If you mean "how does it differ from the kind of eugenics where you murder people", the main difference is that you're not murdering people.
If you mean "how does it differ from the kind of eugenics where people screen for carrier status before they have kids together", the main difference is it's more effective and works for a broader number of conditions.
Maybe you've done this elsewhere (please point me to it if so!), but it would be helpful to justify the idea that embryos aren't people because if they are you're doing precisely that.
I agree with the argument that the couple choosing not to have a 4th child is not equivalent to murder, but I still think you're dodging the actual moral question of eugenics. With embryonic trait selection, the parents are actively making a choice that some kinds of life are not worth living e.g. Downs syndrome. Talking about this as "eliminating disease" is just not morally consistent with how we think about curing disease in living creatures, individuals are deciding those genes shouldn't exist. This is an act of commission by the parent(s).
I reflected on your post about ethical objections based on personhood, and want to offer some reactions. In order:
1. The alcoholic: This is an act of omission. Yes, there is an act of commission to avoid changing an embryo in a harmful way, but really she's choosing to not do something. It is a choice that a life with fetal alcohol syndrome is not as desirable as a life without fetal alcohol syndrome, which has bearing on the topic, but it's not quite the same because she's not already pregnant. A better scenario is if she was pregnant and a heavy drinker, and she has to choose to abort the child, quit drinking, then get pregnant again. Of course, if you think an embryo isn't a form of human life, it may be academic to you, but you may find differing responses to an act of omission vs the act of commission of choosing a life isn't worth living.
2. The intern: This is a delegated act of commission to select for a trait, and the delegate changes its mind based on new data. We’re not arguing about destiny here, we’re arguing about intent of the parents.
3. The parenting workshop: Honestly, this situation feels so manufactured, I don’t understand the relevance. Here, an act of commission (the workshop attendance) has some probabilistic effect on the children in the chaotic, nonlinear system that is life which you can’t control for at all. But polygenic screening is an act of commission with a probabilistic effect on the child born and a completely deterministic effect on the kinds of people not born.
My aunt has some kind of mental disability since birth, though she has led a fulfilling life through her family's efforts and her own. And now I can just choose that life doesn't deserve to exist, at least not as much as another one. That kind of power gives me pause, both for the moral implications and for what it does to the human genetic pool over time. There's probably some useful data about genetic populations that ritually killed children with undesirable traits (Sparta I think?) to understand more about the second-order effects, but I'm willing to grant the first-order effects are almost always greater than the second-order ones.
FWIW these questions are not academic to my family, we meet the criteria you outline, but still both feel deeply uncomfortable with making a decision like that. I am sympathetic that others that have experienced or witness terrible genetic afflictions may feel strongly the other way and simply not want to experience it again themselves, but I feel differently about accepting such tech at the limit and endorsing it as a mainstream technology that everyone should do for any reason they feel like.
I think it's pretty easy to route around your objections here. For example, suppose that it was discovered that a certain prenatal vitamin, let's say Vitamin C, decreased the risk of Down Syndrome (I think this is plausible; even though Down Syndrome is a genetic condition, it has to do with chromosomes failing to sort into sells normally, and some vitamin could improve the sorting process). Do you think it would be immoral for doctors to tell women trying to get pregnant to take Vitamin C?
(sorry, the next part is going to sound kind of anti-intellectual and ad hominem, but I think we're going to go in circles forever without it)
I worry you're going to come up with some set of complicated philosophical rules around probabilistic acts of omission or something that successfully rules in the Vitamin C scenario but rules out the polygenic selection scenario, and I will think those rules aren't interesting/relevant, and we'll be right back where we started. The reason I wouldn't ban mothers from taking Vitamin C isn't because it's *almost* an evil act but barely escapes because of something something probalistic acts of omission. It's because it's not even close to having anything wrong with it in any way.
I feel like it only starts to seem wrong if you're doing some kind of very vague zooming out and squinting, such that every act is judged by things that it vaguely rhymes with as seen through the lens of some kind of extreme and selective paranoia. For example, if I prepare to cut off my child's arm with an axe, and you say "No, don't do that!", then I say "OH, SO YOU'RE SAYING THAT ONE-ARMED PEOPLE ARE SCUM WHO DON'T DESERVE TO LIVE AND SHOULD ALL BE GENOCIDED?", I am doing some kind of weird paranoid drawing-implications-that-I-shouldn't act. We don't require any kind of complex philosophy to justify why I can suggest you not cut off your child's arm, it's a completely normal action that only seems related to evil things if you doing the zoom-out-squint-and-be-paranoid manuever.
I think it is better to not have a disease (including a genetic disease) than to have it (I try to justify that more at https://www.astralcodexten.com/p/nobody-can-make-you-feel-genetically ) I don't think this justifies hurting people with existing genetic diseases (because that would hurt a real, living person). But I do think it justifies creating people without genetic diseases rather than people with those diseases (since this isn't hurting any particular real person with a claim not to be hurt). In order to deny this, you need to create some kind of concept of potential-people's-rights which is going to imply all sorts of insane things later, then come up with new ad hoc shaky philosophical principles to whack-a-mole down each insane implication one by one.
My claim is that this comes from basically the same anti-playing-god place as opposition to vaccines, opposition to antibiotics, etc, that each step gets grandfathered in as it no longer seems like scary new technology and instead is just part of the scenery, and people invent new ad hoc philosophical distinctions each time to explain why last year's "playing God" is obviously okay and it would be horrendous to deny a dying person antibiotics, but this year's "playing God" is still bad. I think instead of doing that we should just abandon the "playing God" concept - at least if we're not religious and don't think there's an actual God who objects to this - and just try to make the world better, as best we can, while moving slowly and watching out for risks.
(an analogy for why I am not so concerned about risks in this case: suppose that we learned most existing disabilities are caused by some kind of radiation coming from the Earth's core. Then there was an earthquake, it disrupted the earth's core, and the level of radiation cut in half. Then only half as many children were born disabled as before. Would you support randomly mutilating / lobotomizing / irradiating babies until we returned to the current disability rate? Why not? It seems like either you must believe that, in expectation, the pre-earthquake level of disability is likely to be better, because there's some compelling reason why having fewer disabilities would be risky or disrupt society, or else you must believe the opposite. If the latter, you should support the random mutilations; if the former, you should support efforts to have fewer disabled children born now. I think almost nobody would support the mutilation in this scenario, and that's because they aren't drawing off some real belief that there are potential risks that outweigh benefits, they're just going off of the belief that all previous ways of playing God are fine but all new ways of playing God feel scary or like they're making aggressive ethical claims)
(here's another thought experiment - suppose that a genetic engineering technique is discovered that lets people "correct" cystic fibrosis after a child is born, but it costs $1 million. A child is dying of cystic fibrosis. Is it ethical to engineer out the cystic fibrosis gene to save the child? Definitely yes, right? Okay, now suppose you are a CF carrier planning to conceive a child. There is a 50-50 chance it has CF. You can either conceive randomly, have a kid with cystic fibrosis, and spend your entire life savings and go deep into debt in order to immediately engineer it out. Or you can do embryo selection for $3000. Either way, you're ending up with a child who doesn't have the gene for cystic fibrosis - you're just wasting an extra $997,000 the first way for no reason)
I am sure your aunt is a very nice person and you are glad she exists. I have a friend with an embryo-selected child. This child seems nice, as far as babies go, and I am also glad she exists. I'm not demanding you kill your aunt any more than you're demanding that I kill my friend's baby. If you accuse me of somehow by-implication thinking of your aunt as "life that doesn't deserve to exist", I will accuse you of somehow by-implication thinking of my friend's baby as "life that doesn't deserve to exist". I think better than that would be for neither of us to accuse the other of anything like that, both agree that existing people can continue to exist (both because they themselves want to exist and because they have many friends and relatives who love them) but keep potential people on the table as a group we can have ethical discussion about without thereby insulting any existing people, who are fine. I think you have to do that to argue about embryo selection at all, but I think once you've done that the argument is strong and we can optimize for things like giving people better lives.
(and it's not academic to me either - my own aunt died of cancer three years ago. I am against this and hope that my descendants don't have to go through this process. I think this is a real harm - in the sense that there will be a real person who dies, and real family members who mourn them - in a way that makes it stronger than the potential harm caused to potential people just because you selected another person before they existed).
First, I appreciate the detailed response. To cut to the chase, of course, I agree that we both agree that existing people can continue to agree. To be more expansive, nobody writes this much in public about these topics, endures the psychic damage of writing in public, then goes back for seconds in the comments because they don't care about people. If nothing else, I have no doubt how much you care, and we are arguing about what is the best way to help people. I also find the idea of complicated philosophical rules to judge each case exhausting and impractical. There are guidelines built on a set of assumptions, and if you're outside that set of assumptions, you have to use your best judgement.
As for keeping potential people on the table we can have ethical discussions about, I'd need to think about it more. The obvious case is homosexuality. If the straights decide no more gay kids, do gays have a right to be angry? Wouldn't your logic dictate they have no right to be angry, and it's on them to make more gay kids? If we get artificial wombs and the gay agenda decides to buy all their supply to flood the world with gays, do the straights have a right to be angry (sorry, this was just an amusing counterfactual to imagine)? Or do we get into a thousand cases of protected traits, lobbying, who gets to decide what to protect, etc.?
It is not my intention to accuse you of killing or judging people walking around, though you are clearly at least judging your own progeny's potential if you are using the service. I think that is the crux of the different perspectives here: you seem to look at the result and if it gives the result you want, the process is judged good, without judging intent, initial conditions, or actions taken. I am not a utilitarian, I lean virtue ethics so the details matter. This is a case of virtues conflicting, which is the main reason I do not like prescriptive morality for these situations because, again, the details matter and there is no substitute for critical thinking.
Consider this also a request for a post about the strength of your conviction in near-term AGI with selecting children for traits that may not show up well past that horizon. I presume it's good insurance if AGI doesn't show up/can't fix them in time?
As I understand it IVF already involves selecting the most-likely-to-implant embryos and discarding others, or using multiple embryos with the understanding that not all of them will implant. This was a relatively common argument in abortion debates.
You're correct this applies in general to ivf. Using multiple embryos seems fine - you're doing the best you can for them to live. Discarding the ones you don't want would not be moral if you view them as having moral standing.
I just left this on The Worst Argument in the World, but I'll put it here too:
"This is such a good post that it's worth perfecting. The Martin Luther King example is a bad one. I've never heard anyone use the argument, "But MLK was a criminal", meaning he had been in jail. On the other hand, frequently people say, "MLK was a plagiarist" or "MLK was a philanderer", and we'll be seeing more of "MLK laughed as he watched rapes". Those are a different fallacy, ad hominem, if the subject is, say, whether MLK was brave. We don't want to confuse that with the noncentral fallacy, which sounds a lot like it but is distinct. " https://www.lesswrong.com/posts/yCWPkLi8wJvewPbEp/the-noncentral-fallacy-the-worst-argument-in-the-world
It is at least as eugenicist for the government to ban this. It is a declaration that it would be more optimal for some people to have Down syndrome or cystic fibrosis or lower IQ or whatever, when they otherwise wouldn't in a world where such therapies were available.
In fact, it is even more pernicious than the alternative, as it requires the state to restrict choice and bodily autonomy. In this sense, restrictions on gene therapy have more in common with the forced sterilizations associated with eugenics than anything else.
eugenics is incredibly obviously good if and only if in the specific form of “open free genehacking clinics on every street corner” where “free” means “it costs zero” and “anyone can enter there and ask for a genetic mod that their geneline will carry henceforth, selected on a catalog of every trait we know how to mod or do iterated embryo selection for”
inb4 “but not everybody would want to use them and then the people who haven’t been genehacked will be unfairly disadvantaged”: oh, so you want Fairness? ok, then sterilize everyone so that nobody will make kids without having thought at least once in their life about making kids or had an opportunity to choose which traits their kids will have
inb4 “but then it will become necessary to have certain genehacks to access certain things” yeah so? right now it is necessary to have certain partially-genetically-determined capabilities to do things, too; at least this scheme permits choosing
inb4 “but then there won’t be disabled people anymore”: Deaf people often want their kids to be Deaf. aspie people often want their kids to be aspie. there are even people who want kids with Down’s, and when i say EVERY TRAIT we know how to mod, I MEAN IT
inb4 “making fertility opt-in would enrage the trads who want people who fuck to be punished with kids they don’t want”: that is a feature, not a bug
inb4 “introducing a trivial inconvenience to fertility would prevent so much births that there wouldn’t be enough people to support the olds”: i’m counting on there being enough people who’d choose to mod for additional intelligence that they’d make fully automated luxury gay space communism happen
The argument against eugenics at the embryonic stage is that it reduces natural diversity due to groupthink about e.g. IQ, which is disadvantageous if it turns out that the narrower set of genetics you are producing has significant downsides or just isn't optimal compared to a wider set. Could be due to you not actually knowing downsides/optimality or due to unforseen changing of the environment. There are some conditions where it seems clear that there is no downside to eliminating them, and as long as the effect isn't significant at a population level (only the rich do it) then who cares, but when you are doing positive things like maxing out a life expectancy or IQ you do risk this sort of thing.
Here the argument of the piece was incredibly weak:
"When I think about these concerns, I consider the following thought experiment: suppose that I had a natural, unselected child, and that child became high school valedictorian and got into Harvard. Would my first reaction be “Oh no! This slightly raises her risk of anorexia!”?"
How about this, you gene select a child, they develop anorexia and drop out of school, would your first reaction be "awesome, that meant on expectation I made the best choice, just need more kids!!"
How is this different from environmental interventions?
That is, maybe teaching your kid to read is bad, since it decreases diversity, since otherwise there would be both literate and illiterate children, and maybe illiteracy is good in some way.
And also, if you taught your kid to read, and then they read a cult pamphlet and joined a cult and dropped out of school, would your first reaction be "awesome, teaching them to read was still a good idea in expectation!"
I think this is one of those cases where our common sense intuitions are good, and we use them for everything in near mode, and then we use a totally different set of crazy intuitions for things like new technology, where maybe all good things are bad and second-order effects dominate first-order ones. I think we should instead use our common-sense intuitions for the new technologies too.
In general why is modifying the genes of embryos different from environmental interventions? Inheritance to future generations and lack of the plasticity/changeability of the mind. A closer analogy would be environmental interventions on infants vs the elderly, higher standards of risk should be held for infants of course.
Yeah we should do good things, some of it is clearly good mainly because it is preventing obvious negative outcomes. The risk/reward for “IQ” modifications is less obvious to me and more likely to change with societal values and so on. You can take all the GMO monoculture ag arguments here as well.
Because we're going to have a fuckton of anorexics before we iron out the bugs for "gimme the higher IQ".
Beautiful theory, but nature does what it wants in practice, and in this case "in practice" means "a lot of experimental babies born so we can see where we're screwing up" but unlike lab rats, it's not (yet) permissible to euthanise the failed results.
The way it’s different from teaching your kid to read or filling the house with beautiful educational toys is that the benefits of these environmental goodies are not heritable. I don’t think selecting embryos for valued traits is ethically wrong, but I do wonder whether it might have unintended negative effects if it becomes common practice.
You know how you don’t like having visible upvotes in the comments? Well, I think selecting for traits is sort of like having visible upvotes.
Teaching kids to read and other educational interventions have had a profound effect that persists across generations and reshapes culture. Children taught to read will eventually teach their own children to read; better educated children tend to educate their own children as well. An educated population has a huge effect on society and could have unintended consequences.
I think you have a valid concern, but it doesn't come from a qualitative difference between embryo selection and other interventions. Rather it's about us having too much power to shape future generations. We would have the same worries if we invented some other way to radically change the personality or physiology (eg, some kind of incredibly effective 12 week training program). People would use this new training program to make their kids like most popular celebrity, reducing diversity in personality. Or maybe they'd use it to make their kids hardcore Republican/Democrat supporters. These kids would grow up and use the same intervention on their kids, perpetuating the change in society.
We want to have some power to affect the lives of our children. But too much power can be scary because it allows our mistakes to have permanent profound effects. New technologies are scary because they give us more power. It doesn't matter whether it's embryo selection or some almost magically effective training program; what matters is how powerful the technique is.
Yes, great point. To be perfectly honest, the thing I'm most concerned about when it comes to selecting for intelligence is that there is a certain kind of high intelligence that goes with being socially awkward, introverted, having trouble reading people and empathizing, inexpressiveness, fascination with patterns, etc. Of course many brilliant people do not have that autism lite thing going on, but what if the genes that get identified as high IQ genes are the ones that tend to produce the Aspergerish form of brilliance?
I am very very fond of quite a number of people of that kind, everybody from Oliver Sacks to many people here. I'm all for having more people who are that brilliant. At the same time, it doesn't seem like a plus for society to increase the number of people who are non-social, low-empathy, etc. It seems like more of what we *don't* need. We already have an ever-increasing portion of the "people" part of our lives filled with internet contact, fictional people in games, etc. and of course AI, which can already ape affection and admiration (though not very convincingly), and will soon learn to ape love and lust. How can it be good to increase the fraction of the population that is not terribly interested in other people?
Probably because of issues seen in animal breeding, where some genetic combinations that can get you to a desired phenotype are associated with genetic disorders, and the correlations wasn’t caught at first. For example, HYPP in Quarter Horses. https://www.tsln.com/news/hypp-and-the-impressive-line-of-american-quarter-horses-the-facts/ Essentially the disorder contributes to the desired phenotype.
“This constant flexing of the muscle cells is what leads to the pronounced muscle patterns on the afflicted horses, hence the popularity of the line in the stock horse breeds.”
“A suite of recent studies has reported positive genetic correlations between autism risk and measures of mental ability. These findings indicate that alleles for autism overlap broadly with alleles for high intelligence, which appears paradoxical given that autism is characterized, overall, by below-average IQ. This paradox can be resolved under the hypothesis that autism etiology commonly involves enhanced, but imbalanced, components of intelligence. This hypothesis is supported by convergent evidence showing that autism and high IQ share a diverse set of convergent correlates, including large brain size, fast brain growth, increased sensory and visual-spatial abilities, enhanced synaptic functions, increased attentional focus, high socioeconomic status, more deliberative decision-making, profession and occupational interests in engineering and physical sciences, and high levels of positive assortative mating. These findings help to provide an evolutionary basis to understanding autism risk as underlain in part by dysregulation of intelligence, a core human-specific adaptation. In turn, integration of studies on intelligence with studies of autism should provide novel insights into the neurological and genetic causes of high mental abilities, with important implications for cognitive enhancement, artificial intelligence, the relationship of autism with schizophrenia, and the treatment of both autism and intellectual disability.”
Conversely, selecting against autism could be selecting against certain kinds of intelligence. Ideally these cancel each other out if you do both kinds of selection, so we neither end up with way too autistic people, nor way too dumb people, and I do expect the technology to have way more upsides than downsides, but we should be careful.
The difference, very stark, is here you are manufacturing babies according to specifications. Treating human life as industrial product would lead to collapse of whatever "dignity of rational animal" that is left.
The specifications come from the combined individual decisions of a lot of separate humans trying to do something good. It is no different.
The animal breeders who care enough to actually evaluate the animals on these traits and pick the best, those people, they almost all genuinely care about the animals and want to breed them to be as good as possible. The people who do not care about their animals, puppy mills etc, also do not care about selecting their animals to have better traits and will just breed and sell as many as possible.
Take rabbits for example. Breeders of pet rabbits who want to improve the breed generally want to make them friendlier and cuter. And due to human perception, having a round face with a short nose is cuter. So breeders choose bunnies with round faces and short noses, and they are indeed very cute, but this is how Holland lops end up with tooth issues, because their mouths are not designed to fit into such a small space. Now, there is another combination of individual decisions breeders make - to compare their animals to those of other breeders in shows. Without this, the selection for shorter noses probably would not have gotten to such an extreme that it has a noticeable impact on health. Compare for instance Dutch rabbits, which do have shorter noses but not to such a ridiculous extreme.
With that background information out of the way, what do we make an analogy to? I would say the genetic testing companies are like the rabbit shows. Both have a very limited view of the situation - genetic testing companies can only see through the filter of a few genes with known effects, while judges at rabbit shows only spend a few minutes with each rabbit and do not get a chance to evaluate complex traits like personality or health. And both use that imperfect lens to decide which genes should be passed to the next generation.
(That being said, I'm still all in favor of selecting against gene variants with known highly negative effects. It's the positive selection that is easier to mess up, and even then, you can't do much harm within a single generation and by the time of the next one we'll have much better research.)
Gian, human dignity has gone out the window, only dumb old believers in the sky-fairy even hang on to that concept any more, and you don't want to be one of the dumb old sky-fairy believers, now do you? That would hold you back from all the fun things we all want to do!
This isn’t really a new technology concern. The issues being raised are as old as deliberate animal breeding. Selection sometimes pulls along things other than what you are intentionally selecting for.
Common sense intuitions from animal breeders would be that avoiding deleterious mutations is pretty safe, and positive selection of desired traits needs to consider the whole organism, not just individual measurements.
For example, some high IQ genes appear to also increase autism risk. In horses, selection for muscling resulted in inadvertent selection for HYPP, a genetic defect. Just looking at one trait can lead you astray.
The diversity discussed is GENETIC diversity, or range of options in the gene pool. Various versions of genes are called alleles. Literacy is not an allele. Alleles can provide immunity to a specific disease or provide other adaptive advantages.
You can see the genetic marks of old plagues in populations from the increased frequency of protective alleles to those diseases.
But this is not a real objection, unless you also object to the idea of people selecting their individual sexual partners. Characteristics like height, physical fitness, intelligence, lack of deformities, et al. are attractive precisely because they predict fitness in offspring. It's a cruder method, but we still practice it. Your logic would have us select sexual partners at random, just in case we were discounting some hidden yet important gene - though a gene pertaining especially to short, unfit, unintelligent people is unlikely to be particularly salubrious.
Choosing a partner allows for observation of the outcome rather than prediction. And after partner-choosing there is the randomness of inheritance that we are discussing.
But the big one is that we cannot all choose the same partner. If you proposed that every child was fathered by either the Super Bowl MVP or a Nobel laureate, I would also object that we are losing genetic diversity. The current system does indeed allow for many relatively short fat stupid people to procreate and it’s not necessarily a bad thing.
Are you saying that people's choice in partners is not based in selection? That they do not select for traits that predict selective advantage for their offspring? I'm afraid I cannot agree with this. You discount an entire field of evolutionary theory.
Inheritance may be somewhat at random, but in any case the difference in question is one of degree, not function. It is not *entirely* at random, because observations concerning genetics become more accurate at the general, not the specific, level. Once we are talking about the group level, genetic prediction can be extremely accurate. And at the individual level, one has only to go back a few generations to make fairly accurate predictions about a given person's traits. This is necessary in animal breeding, for example, and humans do not function differently on a biological level.
What inherent advantage do you believe random 'diversity' to possess? A proliferation of disadvantageous traits may be legal, even morally acceptable, and still have deleterious effects. Because we do not force people into relationships, humans naturally have a tendency to pair themselves with partners whose traits are the best they can manage to obtain. No one is actually choosing short, stupid, fat people as a primary option in our society because they understand genetically on some level that these are traits which are disadvantageous. The consequences of sexual selection have much more to do with the genetic makeup of population groups than random drift. This was well established by Fisher (The Genetical Theory of Natural Selection).
Whatever process affects the genetic makeup of the population is "selection". It's a tautological concept. We are moving on the contiuum between "natural" selection where that process is mostly determined by "nature" meaning physical environmental factors and access to mates/survivability, through selection due to societal-psychological factors and associative mating based on "who do I think has the best traits", into a potential territory where genetics might be shaped by intellectual intent and design.
But yes, eugenics based not on embryo selection or gene-level mapping/modeling/prediction, but rather on intellectually intentional mating practices, espeically when applied at the population level, also seems bad. Of course that is the traditional way to do it and it's probably much more powerful at this point in the technology. It seems that most in this discussion want to run away from connecting this treatment with "breeding humans" so interesting that you embrace it. Producing the human equivalent of the English bulldog is exactly the fear.
All breading habits across all mammals are not random at all, they exist specifically to improving genetic outcomes. Just most people are bottom feeders so they get the best they can.
"making fertility opt-in would enrage the trads who want people who fuck to be punished with kids they don’t want”
Ah, *that's* what I want, and not because I think abortion is murder! Well, nice that you know what I believe better than I do, thanks for opening my eyes!
We don’t want people who fuck to be punished, retard; they punish themselves by destroying their own (and the rest of our) human integrity by enjoying sex as if it were no different from going out for a walk in the park. The ramifications of disociating body, reproduction and soul will be (and is proving to be) disastrous for society. This entire thread is genuinely convinced IVF is good and will have no consequences on furthering our dehuminization ontop of the fact that it’s involves another attempt to ratify the moral loophole of murder that is embrionic killing (just like with abortion).
Maybe get down from your high horse and take a look in the mirror before talking to me from your false virtue. I may be using ugly words to insult you but only as a response to your attempt at defaming an entire group of people by deliberately simplifying a complex topic to misrepresent their intentions. You’re just another wolf in sheep’s clothing.
Blurring the intrinsic value of human life at the embryonic stage is a loophole to morally justify killing that person before their personhood is directly evident.
We're stuck with questions of morality, though, because this opens the door to the dream of social engineering.
We want the future generations to be the best. But what is the best? Well, that they'll have the qualities we value and the values we think correct.
But it's hard to forecast what the society of fifty to a hundred years hence could or will be like. If we had this tech in the 1950s, the qualities and values that were self-evidently correct would be "now there will be no more perversion, we can do away with homosexuality". Today, we have Pride parades.
What will be the Pride parades of 2075? What will be the optimum human for those days? We need to be a little more humble before deciding the future is blue-eyed blondes (that didn't go so well the last time it was tried) and engineering our children to fit.
Okay, so, one of the standard "proofs" from the pro-abortion side that the pro-life side's revealed preferences don't match their rhetoric is that pro-lifers rarely protest IVF (the ultimate topic at hand), even though it creates 5-10x as many aborted viable embryos as a single abortion.
I ask this out of genuine curiosity, because I assume you've at least thought about this, what is your personal opinion there?
Watch me smile like a shark as I pull my Roman Catholicism card out of my sleeve.
Got you covered right there, fam. We don't approve of IVF (and indeed, I caused personal pain to another former commenter on here, which I do regret, by expressing that in a previous debate about the topic).
I'm still curious about your opinions on the magnitude of it. Would you say it is as bad as 5 to 10 "regular" abortions, or is there some sort of mitigating discount rate?
Or is comparing the magnitude of sins at all a fool's errand? I don't recall seeing whether you've ever expressed an opinion on utilitarianism in general.
Edit: Also, is any of this affected by the previous pope's declaration that unborn children can go to heaven instead of limbo?
How long is a piece of string? Is one torture-rape-kill murder equivalent to five or to ten ordinary one-punch murders?
It's all bad.
Now, there is a distinction between venial and mortal sins, which is a whole discussion of theology on its own, but I'll just drop a quick quote from "The Screwtape Letters":
"You will say that these are very small sins; and doubtless, like all young tempters, you are anxious to be able to report spectacular wickedness. But do remember, the only thing that matters is the extent to which you separate the man from the Enemy.
It does not matter how small the sins are provided that their cumulative effect is to edge the man away from the Light and out into the Nothing. Murder is no better than cards if cards can do the trick. Indeed the safest road to Hell is the gradual one — the gentle slope, soft underfoot, without sudden turnings, without milestones, without signposts".
One possibility is that a small amount of a trait is good, but efforts to eliminate the bad, severe cases are bad. A little bipolar gets you artists, severe bipolar ruins lives.
A possibly hard case-- there's a gene which, when homozygous, gives immunity to AIDS. The piece of the immune system that AIDS hooks on to just isn't there. This means the person is a little less healthy in general. Worth it? Worth it going forward?
If I were expecting highly effective embryo selection to happen instantly (and I were young enough to live to see the effects), I'd worry about an "arms race" on traits like height (a race to the top? :-) ) which is, to an extent, zero-sum. As it is, I think too much else (AI, of course) will happen first.
The only acceptable answer to "But then there won't be disabled people anymore" is "Good." I can think of few better indictments of libertarianism than the idea that people should be able to choose to make their kids disabled on purpose.
That sounds uncharitable to me. If we have the technical ability to prevent certain traits that we might be considered disabilities, then *someone* has to decide which such traits (if any) are actually disabilities that should be prevented. I'm not sure that parents are always the right someones, but it's not obvious to me that government is the right someone, either.
That goes double if we can't be sure of side-effects — where we're not just preventing a putative disability but also imposing unknown and potentially unknowable risks. (E.g., maybe it'll turn out that knocking out a certain gene eliminates the risk of nearsightedness but doubles the risk of breast cancer.) *Someone* has to make that call.
That's not how words work. If it were, then "jaunty" and "gentle" would be synonyms (both from French "gentil"), as would "isolate" and "insulate", "diamond" and "adamant", "metal" and "mettle", etc., etc., etc.
Merriam-Webster defines it as "the practice or advocacy of controlled selective breeding of human populations (as by sterilization) to improve the populations' genetic composition" (https://www.merriam-webster.com/dictionary/eugenics), which sounds about right to me.
Note that it specifies several elements not clear from the Greek roots, e.g. "human" (and not, say, selecting for meatier chickens) and "populations" (and not, say, an individual choosing not to have biological kids because they have a condition they don't want to risk passing on).
I'm not sure whether population-scale embryo selection would count as "selective breeding" for purposes of this definition, if every couple is allowed to procreate but not to choose which of their embryos is used. That technology wasn't available when the eugenics movement was in ascendance, so it didn't come up; we'll have to see if the term ends up including it. (Cf. "phone"; if you could explain all the different kinds of 2025 computer to someone living in 1940, I don't think they'd guess that any of them would be called "phones", let alone exactly which ones would and wouldn't.)
Apparently by your definition so is masturbation 2 days before trying to inceminate lol. I'm euginically eliminating the most aged sperm in favor of a fresher sample.
We're about to see some weird generational cycles.
Every generation one trait will get optimized to the point of oversaturation, and the next generation will find its comparative advantage by selecting for a different mix of traits.
If you mean "saturated" in the sense of up to the theoretical maximum, that means nobody ever gets cancer, heart disease, or obesity, and everyone's a genius. Why would you select for a different mix of traits than that next time?
If you mean "saturated" in the sense of up to some current level of technology, the current level of technology is changing much faster than the generational cycle. I expect genetic technologies to be much better than this within ten years.
Make it so everyone gets that 6 point IQ boost, and now the 100 IQ people are 106 IQ and the 120 IQ people are 126 IQ. That sounds great, but now we have to renorm all the tests so now the 106ers are the new 100 IQ.
It might be a better world. Or it might be just more of the same, where the rich and smart get richer and the rest of us pick up the crumbs.
IIUC, the extremely rich are not generally those with extremely high IQ. It's more the "moderately rich". The extremely rich are generally those who start off rich and focus on getting richer. They may have only an average IQ, but they hire people to do things that require such a characteristic.
I think that's pretty obviously an all-boats-lifted scenario. Sure, people's individual experience of whether society treats them as "smart" is just going to do the treadmill thing you're describing, and I certainly wouldn't expect it to particularly help inequality. But, pretty much every job on the planet would get performed just a little bit better here and there.
Yeah, but we're already performing "a little bit better" compared to previous generations and eras, and we're still complaining.
The smarter world will do a little bit better than us, but that will be the new normal, so now people will complain about how it takes a whole two minutes for their customer service problems to be solved.
I recall that Damon Knight once wrote a story where this happened. One character mentions that having beautiful children is like playing the stock market, you have to guess how beauty standards will change in the future.
Imagine if we'd had this technology in the 80s and 90s. It would be a net win for society because of increased intelligence and decreased mental illness. But you'd have to take a moment to pity all the poor white women whose parents selected them to have small butts!
IIUC, that's already happened in China, where people selected for male offspring. (Perhaps that's happened elsewhere, too.) So far, however, the number who could afford the treatment has been a small enough proportion to avoid most of the bad effects.
I don't have super consistent income yet and I could probably pull together $27K for something really important but it wouldn't be, like, remotely easy. My parents mostly pay my medical expenses but my guess is the copays are probably something on the order of $1000 per year, and the vast majority of that is continuous glucose monitors and insulin pumps (which aren't strictly necessary), the insulin itself is like $10 for a 90 day supply I think? No idea what the insurance is company is paying.
Okay, that makes sense. In the post, I said the claim required that "you’re well-off enough that these aren’t impossible sums of money for you", and I was referring to people with Type 2 who tend to be in their 40s and 50s and so probably richer than you. I'll edit "impossible" to "extraordinary".
Strongly agree. I have Type 1 as well, but am in my 40’s and fairly well off. I’d pay >>$100k to get rid of it. Not pay excruciating attention to how much I’m eating would be amazing.
FWIW, I'm type 2 diabetic, but I control it by diet so officially I'm "not currently diabetic" (which is because of a silly definition, not because of remission).
I might pay $5,000 for a complete cure. (If you'd asked me a decade ago, I would probably have been willing to pay more. If you'd asked me in the first year after diagnosis, I'd probably have agreed to the $27,000.)
I think that's fair. If you had gotten the cure for $27k as soon as you were diagnosed you would have saved yourself a huge amount of effort, time, and cost learning and working to manage the condition. Now, you would benefit from the instant cure but you'd never get that time and effort back.
The good news is that I don’t think your 50k number is correct anymore from Herasight. They might reach out and correct you, and we might see just how quickly capitalism brings down costs for technologies like this
I asked Herasight the day before yesterday. They said it was still their most recent number, but that they hope to get it lower in the near future, and that people with unusual situations could ask them for a custom quote that might be less.
Oh, I should probably tell you something: type 1 diabetes is one of the most tractable conditions with gene editing. We can lower the risk by >50% with a few edits. It's actually quite unusual in this way. Conditions like schizophrenia or ADHD involve far more genes with much smaller effects.
Not so with type 1.
And because type 1 has such an earlier age of onset compared with other "highly editable" conditions like Alzheimer's or breast cancer, it's a strong candidate to be one of the first conditions we edit for. We'll be able to demonstrate within ~5 years or so that it actually works.
I think early editing tech is probably going to be roughly on par with the cost of the more expensive embryo selection companies (possibly a bit cheaper), so it's possible cost will still be an issue.
But it's worth noting you can at least lower the cost of egg retrieval by doing it abroad. You can do egg retrieval in the Czech Republic for $3000-$4000, in Spain for $6000, and in Vietnam for $2000. And while I can't really vouch for the quality of specific clinics, the expected cost per child is almost certainly lower than for US clinics.
If you're interested in talking about embryo editing for type one, drop me an email at genesmithlesswrong@gmail.com
Scott, you say: "rich people get first crack at every new technology, but poor people usually follow close behind... Also, it would be crazy for any forward-thinking government not to cover this."
That's reassuring for the classic concern that this tech could, over generations, create genetically-stratified socioeconomic classes *within* a country.
But across countries, still seems like things are gonna get really weird, like "ok now there are *actually* significant genetic differences in IQ across countries". Seems like an important downside.
You would only be one generation "behind" the rest of your country if you had an unselected pregnancy. Probably a wash if we're looking at the impact over multiple generations.
I'm actually a bit more optimistic because there are so many interventions left in developing countries that could raise IQ significantly more cost-effectively than embryo selection. The $/IQ point cost effectiveness for LEAP is orders of magnitude more than that of embryo selection. So I would actually expect to see the gap close in coming years, even if embryo selection is widely adopted.
In the long run, my hope is developing countries can just get a lot richer, and IVF + embryo selection can get cheaper. At some point those two lines will cross. Our job is to increase the slope of both curves.
Unlike wealthy countries, where correlation between fertility and IQ is zero or weak-negative, it's very negative in many poor countries and their brightest are likely to emigrate to rich countries; as environmental gap shrinks, genetic grows...
I am mostly of the opinion that a scenario where country A has IQ 100 and country B has IQ 120 is worse than a scenario where they're both 120 but better than a scenario where they're both 100. I think equality is nice, but not so nice that you should burn down all the houses to give everyone equally good tents.
I do have some concerns that relative advantages can be exploited to get away with evil crap, but we already have some pretty big relative advantages along other axes and the results have not so far convinced me that it's worthwhile to outright destroy value just for more equality, except in some special cases like literal weapons (e.g. treaties that put limits on how many nukes each country can have).
I think there's a somewhat wider range of scenarios where it's good to _redistribute_ some of the value (e.g. progressive tax), compared to destroying it. So hypothetically maybe it would be good if there were a global policy of higher-IQ countries paying some aid to lower-IQ countries.
("But they'd never agree to that!" "Oh, but you think they'd agree to voluntarily lowering their IQs instead?")
All I can say is if I was inclined to do this sort of thing and I already had kids I'd be asking the company to test the existing kids too to calibrate how seriously to take the predictions.
First, your kids probably don't have prostate cancer yet. And child IQ is pretty weakly correlated with adult IQ, and the predictors are for the latter.
Second, it's hard to get good intuitions for statistics, but I think even a somewhat valuable predictor can have a high error rate. If I'm reading the widget on Herasight's site correctly, if you have five embryos and use their IQ predictor with supposed +6 IQ point gains, there's still only a 73% chance that your child born using their predictor will be smarter than an average child born without it. So if you "test" their product and find that actually your unselected kid is smarter than your selected kid, this doesn't prove very much in a scientific sense - it's the chance of two coin flips both landing heads.
Maybe a better tactic would be for them to test you, the adult (since you might have gotten some diseases, and you know your adult IQ). Then you only have the latter problem (sample size of one). So you should get them to test lots of different adults. But then you're back at "this works in large studies", which is how they validated them in the first place!
I've had Nucleus and Herasight genetically test me. There wasn't very much useful on there. For example, I have obsessive compulsive disorder. Nucleus said that I had slightly-higher-than-average genetic risk of OCD, which put me at 2.6% compared to the overall population's 2.3%. Is this good (because it identified higher-than-average risk)? Is it bad (because it gave me only 2.6% chance of a condition which I, in fact, have)? I'm not sure, and I think the real answer is to not pretend that you can conclude much from a single data point in a heavily probabilistic domain.
Herasight did also test my two existing children, but I found I couldn't bear to look at the results. We agreed that my wife would look at the results and agreed to let me know if there was anything actionable. But AFAIK they used the information internally to get a better picture of my and my wife's DNA to predict our future children, but forgot to send it to us. I plan to remind them about it when I get around to it.
This whole thing really smells of “amulet to defend against leopards” - there are no leopards, so clearly it works!
It’s functionally impossible for me to confirm the service achieves what it says it does, and even in the best possible world the effects are so tiny I am unlikely to see them; and when I don’t, well, what did I expect? - the dice don’t often roll all sixes, guess I just got unlucky, LOL cha-ching.
Meanwhile, their margins will be even better if instead of doing any work they just hook an RNG up to the numbers in their report - you can’t even start to check the statistics for a couple of decades, while the half life of a startup is measured in single digit years! - and even if you literally go to their lab and look over their shoulder, since in the long run Moloch says everything that can be enshittified will be enshittified, the fact they had not yet done this when you checked has only a limited bearing on how much I can trust them by the time I get around to using the service. Net result: a generation of Silicon Valley neophiles using their cellphones to call the local shaman to help pick embryos.
Imagine, ten years from now, there are lots of these startups and also just as many of the sort that just make up numbers, and everything in between. How do you tell which one to go with, or where on that spectrum any particular one lies, given only their marketing material?
The situation feels somewhere between buying a lottery ticket and one of those deeds for land on the moon: there is some tiny chance it might pay off, but realistically, you’ve probably spent your money on a worthless bit of paper.
This sounds like the perfect opportunity for scammers. After all, how do you prove a negative? "Your child hasn't got prostate cancer yet". Yeah, and by the time they're in the age bracket to get it, they either do get it or they don't.
If they don't, the scammer company (which did not, in fact, select out embryos) can say "Told you it works". If they do, well, that's forty-plus years in the future. The company may not still exist. Or they can pull statistical fine print out of the air to say "Ah, we never said they'd *never* get prostate cancer, we just said this would reduce the risk. Not our fault some risk remained and they got unlucky!"
You make my stony black heart want to set up a dodgy clinic promising the gullible and desperate that I'll give them superbabies, when it's bog-standard IVF, I pocket the dough and flee to Switzerland, and it's their problem if the kids get health difficulties etc. in later life.
That's a real problem. It's a real technology, that really ought to work if done properly, but it's a lot easier to lie, and it may take multiple decades for your lies to be proven. And when the lie *is* proven, most of the evidence for "who was responsible" will be gone, so the court case may be impossible. (Or, by then, the scammer may be dead of old age.)
...where by "work" we mean if everything goes /perfectly/ we see a single digit percentage effect in statistics across the entire population measured several decades later, rather than anything specific in particular that any person actually paying for the service could point to.
Don't even need deliberate lying, as of yet there's a lot of room for "oops, didn't work out as we thought it would, genuine error, well them's the breaks, that's the way the cookie crumbles, hey at least your baby has the right number of eyes and hands, okay?"
The "oops!" is a different problem. Yeah, it's a real one, but it's a different one. (Well, except that with over-promising they blend into each other around the edges.)
Whether they select or not, very ease to prove, because they have to give customers these genomes data and after child is born you can check if child's genome matches the selected one.
And what are you gonna do if the baby doesn't match up? "Excuse me, I wish to return this deficient item, here's the receipt"?
Sure, this is what American law courts are for, but good luck arguing statistical probability there against company lawyers who are pleading "you shoulda read the fine print, there's no guarantees in biology".
“ Herasight did also test my two existing children, but I found I couldn't bear to look at the results.”
It would be worthwhile exploring why this is so.
(Just using 2.6% chance as a generic chance and anorexia as a generic condition) Is it because
- if you found out a child had a 2.6% chance of anorexia, you regret not having had the chance to lower it to 2.3%
- if you found out an existing child had a 2.6% chance of anorexia, it will have been in some sense unfair on them that your third child has a lower chance
- you will regret knowing that your existing child has a 2.6% chance of anorexia versus 2.3% chance on average and having to deal with the consequences of anorexia in the future
- you will regret knowing one existing child has a 2.6% chance of getting anorexia and the other has only a 2.3% chance and that will influence how you treat your two children
I can definitely imagine ‘not being able to look’ but not having kids myself I’m not sure why it is.
"Herasight did also test my two existing children, but I found I couldn't bear to look at the results."
Presumptuous of me to say this, but I am very glad you didn't. You acted like a father, not someone purchasing from a breeder who wants to check the pups are indeed the pedigree claimed.
Super interesting developments, somehow, after a slow start, the 21st century does pick up steam after all.
But with regard to
> To have a sense of why society works the way it does, instead of feeling like you’re being blown back and forth by institutions you don’t really understand?
Would not institutions become more complex at the same rate that general intelligence increases?
Good question. I don't know. Do countries with very high IQ (like China) have noticeably more complicated institutions than those with very low IQ? I'm not sure (and insofar as they do, I'm not sure if it has more to do with IQ or wealth or stability). See also my recent https://www.astralcodexten.com/p/book-review-arguments-about-aborigines post for some discussion of whether "primitive" cultures have more or less complex institutions.
Well, the Pirahã of the Amazon don't have words for numbers, and they have no abstract terms for colors (they'd describe something red as "like blood"). Yet they seem to have words for every plant and animal species in their environment with the uses and dangers of each (if any). Considering the biodiversity of the Amazonian rain forest, that's quite an accomplishment. Probably they'd have trouble with Raven's Progressive Matrices test because they don't really have words for shapes, and they're not trained to recognize the abstractions (in fact, John C. Raven recommended that his test wouldn't be suitable for populations without formal systems of schooling—can't find the quote now, though).
No. But they're foragers, and like any good naturalist, they can recognize all the plants and animals they encounter in their environment. Upon further digging, their designations seem to be general descriptions—e.g., "the plant whose bark gives visions." This is similar to the Inuit "words" for snow, which are really agglutinative descriptions, e.g., "snow that is slushy and granular".
Identifying all the plants in our environment without the help of a phone app is more than most of us can do. "Oh, that's some sort of tree. I think it's an oak."
I think you have to have a high IQ to be able to learn to read and write in a system with no alphabet and thousands of possible grammatically valid characters.
Don’t assume it, come and look! The answer is: lots of very average and dumb Chinese people are literate, so no, intelligence is not required for literacy in a non-alphabetic system.
For what it's worth, it might be true that China has more complicated institutions than the West. When my family member emigrated from China, they were enormously relieved at the amount of bribery and favor-trading they no longer had to do to acquire basic goods and services
I suppose one could argue that the complexity of human institutions has been correlated with the general rise in IQ in all regions of the world over the past century. But I think a better explanation is that the World's GDP has multiplied 4.5x in the past 50 years. More types of gizmos are being manufactured. More companies to make the gizmos. More paperwork (now electronic) to track the gizmo logistics etc. Plus, companies have gone multinational, which requires more workers and whole new systems to deal with dozens of regulatory regimes, etc.
I joined Cisco in 1998. I was at the bottom of the organizational tree, but there were only four levels of management between me and the CEO (John Chambers). When I left Cisco in 2002, there were ten levels between me and John Chambers. The organization had burgeoned with our product space. New systems were implemented to deal with the new complexity of the organization. This definitely had nothing to do with changes in IQ and the timeline was too short to have been affected by the Flynn Effect.
Generational intelligence has been increasing by ~3 pts per decade in the US. Our institutions are definitely more complex than they were when IQ started being measured a century ago. But probably that increase can't be attributed to increased IQ. I doubt if US citizens are any smarter than they were a century ago.
That ~3 has little sense because it depends on which IQ test is used; Raven's matrices show larger improvement while arithmetic is barely up. It's also that were no large-scale studies of modern people on 100 year old tests; it's extrapolation.
Then why do they keep renormalizing them every 10 to 20 years?
But your statement is incorrect. the Flynn Effect has affected all the major IQ tests. However it seems to have plateaued in the US and Scandinavian countries in 00s and 10s.
It affected all tests, but to a wildly different degrees, the 1st picture on wikipedia page about it makes it obvious.
If population improved 10 points on test A, then improved 10 points on test B, then improved 10 points on test C, it doesn't mean it improved 30 points on test A.
Flynn compared the raw scores (not normalized IQs) of people tested in different eras but taking the *same* version of the test. Someone taking the 1950 version in 1980 would score significantly higher than the 1950 average because the test hadn't been renormed. Flynn did not claim that people were getting smarter. And as you can see from the chart you referenced, the average increase in all 5 performance subsets (across various IQ tests) was ~3 pts per decade, which also closely matches the upward trend in FSIQ. RPM is the only actual IQ test mentioned on that chart (and ironically, it improved the most, even though it's supposedly the least culturally and educationally biased measure of intelligence).
And I very much doubt that people are getting smarter, even though raw "intelligence" (whatever the f**k that is) is improving.
They might do. But (1) if you magically (or technologically) make your children a bit cleverer then the effect on their ability to navigate institutional complexity is much larger than the effect on actual institutional complexity, and (2) if it's true that societies of smarter people have more complex institutions, that's probably because they're getting benefits from those more complex institutions sufficient to justify their complexity.
"Also, it would be crazy for any forward-thinking government not to cover this; it could save hundreds of thousands of dollars in future health care expenses. In countries with public health care, this comes directly out of the government treasury; even in the US, it’s covered by Medicare after age 65. The government should be begging people to select embryos."
A good test case for how this could roll out is genetic screening for stuff like cystic fibrosis, where there's a well understood genetic cause and it can be screened in infants (1).
From a very quick review, Medicare does not cover genetic screening for embryos where there's a family history of cystic fibrosis.
It looks like private insurance does cover screening but...only for patients with fertility problems. I'm not sure I understand this policy:
Is your figure for type 2 diabetes specifically for the US? A quick google suggests that around 29% of people in the US over age 65 have (diagnosed or undiagnosed) type II diabetes, but around 16% in the EU, with fairly large differences between countries. Admittedly all these numbers are higher than I would have guessed.
One can imagine a dystopian future where McDonalds offers free gene editing to improve their future market share...
The other thing that stood out to me about that section is that the logic there does not quite follow. Suppose someone's child would have lived to age 65, but thanks to polygenetic embryo selection was able to live to age 66 instead. This costs the US government more money, not less, because they have to pay one year of medical expenses instead of zero.
I think where this will work, if it works at all, will be in things like "does this embryo have CF or not?" and we can then see if the babies born do or do not develop CF. It'll work where there is a discernible, measurable problem and the result can be discernible and measured: "did the baby develop CF or spina bifida or the rest of it?"
That is a case that can be argued for public health benefits.
The stuff about "taller, smarter, prettier" and the rest? Will always be pie in the sky. Maybe the selected embryo is smarter than if you had gone the natural route, maybe the unselected baby would have been smart anyway. Roll those dice and hope for the jackpot!
It seems to me that using this technology not for vanity but to eliminate such a tragic condition as CF - I recall reading an obituary once, “he was preceded in death by … (his brother, his brother, his brother, his sister)” - would be the necessary justification for selecting for IQ.
Very smart people surely made it possible, but if they can’t get this right … what good were they?
It is important to know that we have made significant advances in treatment for CF over the last few decades. In 2022, the estimated median predicted survival age for a baby born in the UK with CF was 56. In 2023, it was 64. I’d expect to hear the prediction for babies born in 2024 around October. This is extraordinary progress.
It is no longer a disease which guarantees an early death, as it was relatively recently. And the progress is continuing and rapid.
Not yet! They’ve been working on gene-editing therapies for ages: some have looked promising, but I’m not aware of any success yet. It’s been the big thing people have hoped for.
I'm a parent via IVF. When we were weighing the decision to do embryo screening, our providers lightly recommended against it on the following grounds: we are relatively young (mid-30s) and thus comparatively unlikely to have significant age-based aneuploid concerns; and any genetic screening would necessarily involve removal of cells from the embryo which has a non-insignificant chance of disrupting its viability. Can you speak to that latter concern? Particularly for the current userbase of IVF, viability of pregnancy is a pretty paramount topic!
"As with any medical procedure, embryo biopsy carries some risks. There is a small chance (less than 1-2% with a skilled embryologist) that the embryo may be damaged during the biopsy, causing it to stop developing or making it unsuitable for transfer. While current data suggests that embryo biopsy has no adverse impact on the growth or medical outcomes of the resulting children, it's essential to understand that the technique is still relatively new, and there may be potential unknown consequences for live-born babies. If you're considering PGTA for your frozen embryos, our article on getting Orchid's screening on frozen embryos provides valuable information."
If embryo biopsy results in a 1-2% chance of serious enough damage that the embryo is damaged enough "to stop developing or mak[e] it unsuitable for transfer," my guess is that that's just the tip of the iceberg, and that more subtle damage is being done to a larger percentage of embryos as well.
My understanding is that a small percentage -- like 5%? -- of embryos are damaged by genetic sampling. There's a newer non-invasive way to test the genetic material in the substrate around the embryo. I bet this will become more common. (I'm no expert, just have done some reading)
One thing that concerns me is the suggestion in the research that IVF conception itself may increase the risk of certain kinds of cancers, particularly with frozen embryo transfer. I think studies say this is a pretty small risk at this point, but if there's a massive increase in people using IVF, it may become a bigger issue.
So would it be fair to effectively "nerf" all of the health benefits quoted above by 5% given that likelihood of the testing being damaging? That certainly affects the final expected value.
Yes, obviously it wouldn't be transferred. But you've lost that embryo. For many current users of IVF, the number of viable embryos that are retrieved is worryingly small, sometimes even just one or two per egg retrieval. If the embryo is damaged, they've gone from "reasonable likelihood of a healthy child" to "no child".
I guess I'm confused because if the data is about reducing risk of conditions for babies that are actually carried to term and born, then 5% of total embryos lost are unlikely to significantly change those improved outcomes.
Average number of eggs per retrieval is 8-14, they are graded by viability. It's true some women produce fewer eggs per retrieval and in that instance may forgo genetic testing because of that, but those are a minority of current IVF patients who are mostly doing IVF because of fertility problems just like that (insufficient egg supply).
Also there is a new method for testing embryos by testing the genetic material that exists in the substrate around the embryo. It's noninvasive to the embryo, and I think we're going to see more of this technique in the coming years because of this concern.
I'm also interested in the possible impact on the balance of men and women in a population. Most of the people I know doing IVF (which is not that many) are making choices about embryos based on sex along side genetic factors. Is anyone talking about this? Do fertility clinics have data about whether these sex-based choices are tipping in one direction or the other?
I see from reading around that IVF skews male though this seems tied to what kind of IVF technology is used rather than because people's choices are skewing outcomes. Maybe too soon to say on that.
People who tell you they're selecting for sex will tell you they prefer girls. I'm guessing there's substantial stigma around selecting for boys, to the point where you'd get IQ or something instead as the reason.
This may be the result of overly cautious providers, but it may be particular to your embryos. If your embryos developed slowly and were frozen at a later-than-optimal stage of development (day 5 is preferred, but many providers will use and freeze days 6 and 7) or if your embryos appeared low grade under the microscope, genetic testing could cause sufficient trauma to the embryo to lose the embryo, especially if a freeze-thaw cycle is involved. This might also be recommended if you have very few embryos.
Removal of the cells for testing has a pretty small chance of harming the embryo under standard/ideal conditions, which is to say removal of cells for PGT-A/-M after fertilization but before initial freezing.
It's best when testing embryos to know upfront (i.e. before freezing) which testing you want to do, and even then testing is not an equal risk for all embryos.
And on average, two-thirds of patients require six or more IVF cycles for a successful pregnancy. At 15-20K a pop that could make a significant cost difference if the Herasight tests reduce the viability of embryos.
So I'm not actually an expert on this, but I did work at a genetic testing company for a year and I've talked with pretty much everyone in the field.
From everything I've read and heard, a trophectoderm biopsy (the standard kind of biopsy taken in IVF) has a negligible impact on live birth rates. I wish I could give you an actual number here like "the impact on llive birth rates is <X% based on this study", but my knowledge is not that deep.
There ARE studies showing that biopsies of the inner cell mass (the part of the embryo that becomes the actual fetus) do have an impact on success rates, but those aren't commonly practiced by clinics.
No doubt I'm an idiot, but why the focus on Type 2 diabetes, which is far more lifestyle-related, rather than on Type 1, which is far more genetically based? Also, I have Type 1 in my family.
It's only 90% if you do some goofy thing where you restrict cases to people who developed it at <5 years of age. The AUC for a more realistic case cutoff (<15) is still quite good, but not that good.
I'm curious to know if the risks to the baby have been taken into account here. Not the risks of embryo selection, but the risks of IVF in the first place. Obviously if a couple is having IVF for (eg) fertility reasons, an embryo has to be picked, whatever metric is used. But having IVF purely for embryo selection could be counterproductive, if the process itself leads to an increased risk of some adverse effects (eg preterm birth, lower birth weight), as some studies have suggested. That said, most IVF up to now has been done for couples having fertility issues, so it's hard to disentangle causation. Is this something you have looked into, and taken into account?
Many of these traits represent our society's current values. Even the disease ADHD is mainly a disease because of the current cultural milieu. Is it wise to increase selection pressure as much as these technologies are likely to when they trickle down, given the transitory arbitrariness of the criteria? I can see reducing Alzheimer's and cancer. But I'm uneasy about many of the other criteria.
If having ADHD (to the extent it's a disease) is advantageous in some way, then there will end up being a shortage of people who do whatever jobs people with ADHD are good at, meaning parents will stop selecting against it because they want their kids to be able to get good jobs.
I can easily imagine a tragedy-of-the-commons situation where something is beneficial to society but not to individuals : maybe the jobs ADHD people are good at are low status or otherwise undesirable in a way that can't be corrected by economics in the way you describe. Or ADHD suddenly becomes valuable because of environmental changes. But this doesn't seem terribly likely?
This is by far the least convincing argument against eugenics I've ever heard. Oh no, there'll only be one billion kinds of person instead of one hundred billion, only more worthwhile art being created every week than anyone can can consume in a lifetime instead of every day, what a tragedy.
I worry about the environmental changes first, because our world is changing at a very rapid rate, but I think the "something is beneficial to society but not to individuals" is underrated. Imagine AIs paid in "content" -- lolcows become a very valuable commodity for a society, then. Regardless of how well they do for themselves.
re: environmental changes I'm hopeful that reduced genetic diversity won't matter a ton because humans have adapted to pretty extremely different environments in the past with changes in memes not genes. Humans *have* adapted genetically to changing environs (high altitude tibetans adapting to low oxygen, duplication of genes to metabolize grains post agricultural revolution... etc.) But these are really quite minor compared to the way humans have adapted to new environs through new technologies (consider the Inuit! Or space travel!)
Even without embryo selection, every population outside Africa is already very homogeneous due to founder effects and rapid population growth and we've done OK. So to be honest I'm not terribly worried that embryo selection will eliminate some essential diversity. Though it is a frightening idea.
You missed our evolutionary adaptations as "fire users" (which seem to be quite extreme).
Manic Depression or other psychological "brainstates" may be what's necessary to make geniuses. (Not that I'm saying every genius was the Romantic Ideal, many are workhorses -- but even they can go through cycles of "up" and "down" -- or inability to start projects because "there's so many ways to start it!")
By "evolutionary adaptations as fire users" are you referring to co-evolution with technology (e.g. we can't extract enough calories to reproduce without some kind of fire/cooking)?
Right now I'm not persuaded by the idea that genius is connected to mental illness, but if it turns out to be so that definitely would suggest embryo selection against mental illness could be pretty bad! That would be its own interesting moral dilemma. "This embryo has X% chance of debilitating mental illness but Y% chance of being a genius and benefiting lots of other people."
I worry that the equilibrium might not shake out to be that adequate. Parents make bad and silly decisions about their children's lives and future all the time, and status game spirals can often take over parts of human culture quite thoroughly.
One problem is the feedback time is too long: if there are actually jobs people with ADHD are good at (but fewer than the amount of people with ADHD today), the issue that we select too much against ADHD will only become apparent 20–30 years after embryo selection becomes widespread, so we'll have an entire generation of people too heavily selected against ADHD.
The vast majority of congenital conditions are universally considered harmful. No one's arguing that anencephaly, Proteus syndrome, Sickle cell disease, asthma, bladder exstrophy, or nearly all the others improve a society. These conditions cause great pain independent of any "cultural milieu."
The argument for preserving certain congenital conditions only applies to the tiny fraction of them that some people think provide some benefit like ADHD or high-functioning autism. These conditions won't be selected against as much since many parents will think they are fine, so I don't see this as being a problem.
Should we be maybe trying to create a database of whole-genome sequences and alleles before we start artificially changing the population distribution? Ideally one that is large enough to capture a pretty large chunk of rare alleles?
I think I'm pretty well on the side of "this is probably a good thing", but in the off chance that wide spread use of this technology causes unforseen problems, knowing what our populations genetic distribution looked like before we messed it all up seems pretty useful.
(Admittedly, Ensembl and UCSC are genome browsers, but they include variants and their information, such as frequency, predicted impact, and known phenotypes).
Pangenome sounds pretty much like what I'm envisioning, although, if I'm understanding statements on the website correctly, they are talking about the order of hundreds of full genome sequences. That's a great start, but in my opinion, we should probably be aiming for thousands-to-millions to get a fully representative global sample that is large enough to capture non-trivial amounts of rare alleles.
It's going to be a looooong time before this tech has any noticeable impact on genetic diversity. There are about 10 million genetic variants in the human gene pool with a frequency of at least 1%. Even if you did embryo selection for ten generations, and everyone was selected for the same things, you'd barely move that number.
Frankly, even with more powerful techniques like gene editing or iterated meiotic selection you're going to need many generations before this becomes a big concern.
The only area where I'm somewhat worried about this is the human immune system. We've had crazy amounts of selection on the human immune system in the last 10,000 years, to the point where we literally have higher rates of read errors in genetic sequencing when reading regions like the HLA. It's also harder to judge the tradeoffs you're making when changing immune variants because often they're present to protect against plagues or other uncommon diseases. We can't really phenotype those effects well until there's some sort of pandemic.
HLA pops up in GWASes for every damn trait too, so alleles there might be weighted highly in the predictors these companies are using? (Or maybe they are filtered out, since as you noted it's hard to align to that region?) One can imagine a "tragedy of the commons" situation, where every individual is incentivized to select for a certain set of immune haplotypes that e.g. reduce schizophrenia risk, but this counteracts diversifying selection at that locus, making the population as a whole more vulnerable to disease. But that's just a worst-case I have no strong reason to believe currently.
> We can't really phenotype those effects well until there's some sort of pandemic.
That's a good point. It reminds me of an old saw that goes something like:
Student: The mouse knockout model has no phenotype.
Professor: Well, did you take it to the opera?"
(The joke being that you can't know that the gene doesn't produce a phenotype under opera conditions until you explicitly test it.)
Well the plague protection benefits of some HLA variant would increase as the population as a whole became more vulnerable, so if you have good enough predictors and there is at least some transparency into what other parents are selecting for, this is something that in theory we should be able to avoid.
Though it would obviously be more difficult than the usual kind of selection.
Honestly the hardest part is probably going to be figuring out which HLA or other immune variants protect against which pandemics. This is why I think by default we should be prioritizing immune diversity **relative to other people in the population**
>Well the plague protection benefits of some HLA variant would increase as the population as a whole became more vulnerable
Maybe. It also seems very possible that, rather than the main effect of HLA heterogeneity being "at least someone will have a protective allele when a pandemic hits", HLA heterogeneity is *itself* protective. If most people have the same HLA the pathogen just needs to change its protein sequences to avoid antigen presentation by that one HLA. If we all have different HLAs the pathogen needs to avoid being presented by ANY of them, which dramatically constrains it & reduces transmissibility in a population, preventing outbreaks. So your neighbor's HLA diversity could benefit you.
(This is not to mention the specter of other nasty things like transmissible cancers that can result from low genetic diversity in immune genes).
>if you have good enough predictors and there is at least some transparency into what other parents are selecting for, this is something that in theory we should be able to avoid.
Absolutely. I would bet that by the time that embryo selection became wide-spread enough to start seriously affecting allele frequencies the tech will have improved enough that we can just give kids whatever MHC to promote diversity relative to the population just as you suggest & this whole thing will be a moot point.
Who says we aren't artificially changing the population distribution? We have had significant manmade interventions on the population scale side, and they may have reproductive effects.
It seems kind of rude to say Kenyan pastoralists get weather predictions from a Shaman instead of being able to connect to the same exact up-to-the-minute weather satellite data as everyone else
I'll bet I can predict with great accuracy whether 90 days from now it will be hotter or colder than today in a given city in the world. Not perfectly (gets fuzzier in the tropics) but generally, November has different temperatures than August in most places and the direction of weather is practically always in one direcction.
It's exciting to see how this field is progressing! My wife and I will likely be customers of one of these companies in the near future.
Regarding SMN and spinal muscular atrophy, the SMN gene is rather difficult to analyze because there are regions of the genome (SMN pseudogenes) that look very similar. Still, difficult is not impossible and I'd expect Nucleus to do better, given that the focus of their company is genetic testing. (See: https://www.illumina.com/science/genomics-research/articles/HBA-SMN-genotyping.html)
And regarding cost, hopefully this will come down in the near future. The cost for genotyping/sequencing should be, at most, in the low hundreds of dollars per embryo, and I think now that this kind of selection is starting to be within the Overton window, we'll see competition drive down the price. So most of the costs will end up being IVF.
Furthermore I think companies which do the actual genotyping/sequencing (GP and Orchid, right now) will end up with a competitive advantage in the long term.
Here's another concern I have about IVF becoming more widespread in order to access genetic testing... and it's about the physical and psychological impact on women. It's an unbelievably demanding process above and beyond pregnancy, childbirth, and early motherhood which are already incredibly demanding. Anyone who has done it can speak to what a very hard road it is for most women.
Last I checked, something like 70% of women going through IVF experience clinical levels of anxiety and depression. About half of women under 35 get pregnant in the first cycle, so even younger women are likely to need to go through more than one cycle and women over 35 even more.
Though I should add that most (not all) IVF right now happens to address infertility issues and that's no doubt part of the psychological anguish of the process. But the hormones, pain, unpredictability, cost, risk, etc etc even without infertility are still quite a load.
One thing you're not taking into account here is just how much worse IVF outcomes are for women with known fertility issues.
If you're having trouble conceiving, IVF can help but your odds still won't necessarily be that great. This is especially true for women in their late 30s or early 40s. If you're under 35 and you have no fertility issues, your odds of having a kid via even just a single round of IVf are around >85%
What really makes people go crazy with IVF is multiple rounds of egg retrieval, implantation, and failure. Yes there are unpleasant side-effects of the medications you need to take, but the things that really drives people crazy is not being able to have the child they so desperately want.
- I think that all future costs should be time discounted (the why is a whole debate over ethics and economics, but I come down strongly pro). By that argument this somehow comes out as not worth it (per your calculations), despite how great it sounds.
- I'm calling baloney on a 6iq point gain being worth that little. If we were incorporating the uncertainty about it (10% chance it works as promised otherwise you get nothing), then maybe, but an actual 6 point IQ gain can't possibly be worth only a five or six figure sum of money.
Scott cites Zagorsky (2007) for the estimate and there are a few of reasons to think the true value would be higher.
First, Zagorsky (2007) is based on a 2004 survey, but in skimming the study, it doesn't look like they adjusted for inflation between 2004 and 2007, which was about 10%. Scott seems to have adjusted for inflation since 2007, but if I'm right, the adjustment should have been about 10% higher.
Second, simply adjusting for CPI from 2007 (or even 2004) probably underestimates the modern equivalent estimate. That's because over time, people have become richer in real terms. Higher IQ presumably shifts you along the income distribution, but since the country is richer than it was in 2004 (with real GDP per capita increasing by about a third) an equivalent SD shift would have a greater impact.
Fourth, and probably most importantly, reanalyses of the Zagorsky data show higher correlations than Zagorsky found. For example, Dalliard (2016): https://humanvarieties.org/2016/01/31/iq-and-permanent-income-sizing-up-the-iq-paradox/ found that when considering multiple years of data, the return of 1 IQ point to permanent income is $1,130 (95% CI: 1048-1213) in 2011 dollars, or $950 in 2004 dollars, more than double the Zagorsky estimate.
Using log-transformed income data to address non-linearity, he finds that the affect of 1 IQ point is an increase in income of 2.5%, such that the affect of 6 IQ points would be an increase of income of 16%.
Scott's estimate implies that 6 IQ points would boost annual income by $4,020. Per Dalliard, that would only be true if one's salary without the IQ boost would be $25,125. In other words, per Dalliard, it wildly underestimates it.
For the median salary in the US of $62,088, the annual income boost would be $9,915. For readers interested who presumably have higher than median expected earnings, the boost would be larger, yet.
And my third point applies to Dalliard as well as Zagorsky, suggesting that modern data might show an even higher estimate for Dalliard's analysis. Indeed, the Tweet above shows precisely that - for the same data being referenced above (but different years), the effect of IQ on income (for given ages) was significantly larger for the cohort born in 1997 than the cohort born in 1979.
After writing this, I see that the analysis in the Tweet came from here: https://www.sebjenseb.net/p/how-profitable-is-embryo-selection which actually addresses this very topic, estimating the effect of a 1 IQ point increase on total lifetime earnings of between $40,000 and $90,000, ignoring discounting, just as Scott did in footnote 8.
Damn. I was rushing to write something about this since yesterday but Scott beat me to it.
All I’ll say is that I have a feeling this will be one of those incredibly transformative technologies that change the course of human history forever. I’m somewhat of an ASI skeptic, so I imagine a relatively near future where 3 generations of this technology rapidly improving leads to a real sense of “super babies.” There’s enough sci-fi written about this idea that it’s not important for me to go into all the interesting utopias and dystopias this might lead to.
I think ASI will render this stuff moot in a few years, but if humanity had any will to not die, this is how. We'd halt the AI race, make ourselves smarter through genetic engineering, and then try our hands at solving alignment with our newfound wisdom.
I see your point that you're building on the same architecture, and raising them to have positive feelings towards other humans. But depending on how far they are ahead of baseline humanity, they may not be interested in allowing baseline humanity to participate in collective decisions - they would see it as like allowing children to vote and bind adults to their decisions. I agree that there's a difference between that and an ASI, but I'm not convinced there isn't any overlap between the problems.
Hera is goddess of marriage. Ilithyia is the goddess of childbirth, and Hera used her to delay the birth of Heracles so his kinsman Eurystheus could be born first and become king:
"On the night Heracles and Iphicles were to be born, Hera, knowing of her husband Zeus's adultery, persuaded Zeus to swear an oath that the child born that night to a member of the House of Perseus would become High King. Hera did this knowing that while Heracles was to be born a descendant of Perseus, so too was Eurystheus. Once the oath was sworn, Hera hurried to Alcmene's dwelling and slowed the birth of Heracles and Iphicles by forcing Ilithyia, goddess of childbirth, to sit cross-legged with her clothing tied in knots, thereby causing the twins to be trapped in the womb. Meanwhile, Hera caused Eurystheus to be born prematurely, making him High King in place of Heracles. She would have permanently delayed Heracles's birth had she not been fooled by Galanthis, Alcmene's servant, who lied to Ilithyia, saying that Alcmene had already delivered the baby. Upon hearing this, she jumped in surprise, loosing the knots and inadvertently allowing Alcmene to give birth to Heracles and Iphicles."
So, um, yeah: great patron for picking and choosing who is going to be born for which traits, not so much for "child will have happy life afterwards".
Oh definitely, just not (it would appear, if even Scott is confused as to what Hera is the patron deity of) the associations customers think they are invoking.
"Hi! Our patron is in fact the goddess of marriage, not childbirth, and very vengeful about adultery or non-marital relationships, and very much up for revenge on the offspring of same! So let's all be certain that you're married, not to other people, and having babies under traditional norms or else there might be serpents in the cradle - so to speak" 😁
Worked out pretty well for Hephaestus. He got his own forge in a giant volcano, and married to the goddess of love/lust (even if she cucks him most of the time).
I really, really wish that Scott and other effective altruists took the personhood objection seriously.
I'm not talking about the primary concern that Scott mentions in the linked post, "Who Does Polygenic Selection Help?"; I'm talking about the secondary, "pro-life" concern--the concern that, beginning at fertilization, a human embryo has moral value that's similar in magnitude to a newborn baby, and so abandoning or destroying it is morally akin to abandoning or killing a newborn baby.
Although the philosophical positions that underlie that concern aren't ironclad, I think it's the most plausible account of when humans acquire moral value. At the very least, it's a *plausible* account. Consequently, that should make EAs very reluctant to advocate for a practice that might grow to result in the industrialized killing of millions (perhaps billions) of people per year.
The closest that Scott comes to addressing this is saying that if we threw out polygenic selection on these grounds, then we would have to throw out IVF too. So be it. I don't think that the moral correctness of IVF is a reasonable anchor for our intuition whatsoever.
Even Scott's claim that polygenic screening doesn't exacerbate the issue is seriously wrong. In classic IVF, the purpose of creating multiple embryos is to increase the probability of pregnancy. This number has a conservative upper bound to minimize the possibility of triplets, quadruplets, etc. In contrast, the ideal number of embryos for polygenic screening is practically unlimited, because almost none of them are intended to be implanted. Knowledgeable proponents like Gene Smith and Jonathan Anomaly advocate for batches of dozens, if not hundreds, of embryos per pregnancy. This would easily increase the impact of the ethical question by a factor of 10. Furthermore, if polygenic screening becomes the norm (as some advocate), then the impact will increase by an *extra* factor of at least 20, based on current IVF rates.
I understand the appeal of polygenic screening. I really do--and if it wasn't for the issue of embryonic rights, I would be an avid proponent. As it stands, though, we risk creating an entrenched moral catastrophe akin to the next factoring farming, and the people who pride themselves about their quantitatively informed moral sensibilities don't seem willing to properly wrestle with it.
An embryo at the stages we're talking about here is a double-digit number of cells. It doesn't even have a nervous system, and seems less conscious than an insect, let alone a newborn. If you're coming at this from the abortion debate, where we're talking about fetuses with beating hearts and semi-working brains, your intuitions will be totally off. I think if you want to argue for embryo rights, you need some non-materialist conception of what personhood is or why it matters, which is going to be a hard sell.
I don't think that embryonic rights require a non-materialist account of personhood--at least not in a way that's suspicious.
Suppose that while you're walking in the woods, you encounter a hermit who's fast asleep. If you had a way of killing him instantly, would it be morally acceptable to do so? He's not currently experiencing any desires, he wouldn't feel any pain while being killed, and nobody would miss him.
If you consider that to be murder, then what's the morally relevant difference between the hermit and an embryo? Surely it's not the nervous system itself that's morally significant; it's the capacity for consciousness that we associate with the nervous system. A human embryo also has this capacity; it's merely a step farther removed from manifesting this capacity. The embryo has to grow a nervous system, but why is that different from the hermit having to wake up?
How is this different from saying that a couple that has three kids but chooses not to have a fourth is murdering their fourth kid, and exactly as culpable as a couple that has four kids and then kills one?
I think I would solve the hermit situation by saying that the hermit has a preference against dying (ie formed this preference when he was awake), and that preference can't be ignored just because he's asleep right now.
I think a good criteria for moral standing is existence and the potential for consciousness. An embryo has both, a sleeping hermit has both, an unconcieved child does not.
I don't think your hermit explanation holds up - why is an assumed past preference to live any more valuable than the assumed future preference to live that an embryo has?
I agree that the preference explanation doesn't suffice: even if we suppose that the hermit was temporarily depressed, with no desire to survive, it's still bad if they die when they would have otherwise recovered and lived a happy future life.
The better explanation is that the hermit is psychologically continuous with their (possible) future self. An embryo isn't. (See my reply to Matthew's comment for more detail.)
Could you give a more detailed account of why you think an embryo has "potential for consciousness" but (you seem to imply) an unfertilized egg does not?
The unfertilized egg requires an outside intervention to eventually become conscious, but this seems true of an embryo in a test tube as well; neither is going to develop into a healthy baby without a great deal of outside help.
to better phrase what i said above, existence as something with the potential for consciousness gives moral standing. An unfertilized egg does not meet that - it has to completely change through fertilization to have that potential. An embryo in a test tube only needs the right conditions to develop consciousness. other choices can be argued, but i think fertilization is the most reasonable as far as yielding a radical step change from no standing to standing.
It sounds like on your theory, sperm should have moral standing, which is pretty bad news.
Or suppose that one day robots become conscious. There is a lump of metal in the robot factory. Does it have moral standing, since it could one day be turned into a robot?
I just don't think we need to worry about things that don't have moral standing yet, just because they theoretically could in some future which is not going to happen.
I do think you need to have something like an assumption that past preferences extend into the future to value anyone's preferences at all. Suppose I don't want someone to scream in my ear. I'm probably not thinking about that at this exact moment. You need to sort of abstract my into a continuously-existing being in order to respect that preference and not scream in my ear.
(another reason not to grant embryos moral standing is that I think more than half of embryos that implant are spontaneously aborted because the body decides it doesn't like them for some reason. If we thought embryos had moral standing, this would become the worst genocide in history, and the most important cause would be inventing some kind of artificial womb to save them. But in fact I predict most people don't care about this very much.)
We used to have 50% child mortality. Most people couldn’t do anything about it and just accepted it. Not a genocide, because it’s not the intentional killing of a particular ethnic group, but definitely a mass tragedy.
For infertile couples, if we could end that spontaneous abortion, they’d be happy to have their children. For other couple as well, this kind of miscarriage is probably as sad as losing a newborn would’ve been in cultures that prevented naming until 6+ weeks to avoid being attached.
The argument from consequences isn’t a good one because humans have accepted plenty of bad consequences before we had the capability to end them.
I don’t think an individual sperm is alive in the way an embryo is. An embryo needs food, water, and shelter to grow. From the moment of conception its life plan is determined. A sperm needs an egg, and different eggs will result in different life plans (genes). Much much less deterministic. I think you can say that “you” were once an embryo. I don’t think you can say you were once that sperm, because that sperm could really have become anybody else
My line is existence as something with potential for consciousness. I don't think a sperm qualifies as a person, so doesn't qualify.
I'm going to call robot moral standing out of scope (we're having a hard enough time with people).
Your idea about a moral obligation to stop the genocide of embryos through science is very interesting though. I hadn't thought of this from an EA perspective. I think this still works though - GiveWell says right now it costs 3k to save a life in Nigeria. To argue for devoting your altruism to saving embryos, you'd have to be convinced you could find something that costs less than 3k per embryo (accounting for any suffering from the procedure) plus the up front research cost. I don't think that's likely, but if you could I think I'd be for it.
I agree with others that there is a significant difference between an embryo and a sperm. An embryo is an organism that has the complete genetic information it needs for a human body and (unless someone puts a stop to it or deprives it of a healthy environment) is already at work building it's human body. It isn't a building block in the same way that a sperm or a hot meal is. It is a living organism that is consuming energy in order to build up and develop into what it already completely *is*, not in order to become something other. Nothing further is added to it for the rest of its life other than sustenance.
A sperm isn't like that. A sperm is happy to go it's whole life swimming around and then to die. No matter how much sustenance you give it, it won't develop any further. You need to combine it with a whole different organism for it to become something other than it was, and from there it's a human, no longer a sperm or an egg. I agree that you can't grant moral standing to a lump of clay or to sperm, but that doesn't have anything to do with an embryo, which is a human at the very beginning of its development, not a building block of a human.
As for the issue of spontaneous abortions: I admit that it's pretty hard for me to feel emotions instinctively about a human that still only consists of a few cells (at least ones that I can't see--something weird triggers inside me when I see imaging of an embryo), but thinking about the fact you bring up I do think it's tragic. I think there is not much that the average person could do to solve the problem other than not to try to have kids, but I don't think they would be guilty of a death from the woman's body rejecting the embryo, just as they wouldn't be guilty if their child later died of cancer. Both are bodies not operating as they should.
Just because the average person is unable to do anything about a common cause of embryo (human) death, I don't think that justifies them in intentionally killing a different embryo.
That's a simple explanation, but I really do believe it's as simple as that. For someone like you, who has studied and thought about these kinds of things extremely extensively, I would imagine it would be incredibly difficult for you to think of it in simple terms like this. I wouldn't blame you. And if one day you change your mind and feel guilty of the murder of your own children (although if the embryos are frozen then technically they're not murdered, but that's still a really weird moral ground. I was very surprised just now when I googled it, though, to find out that there are embryos from 30+ years ago being unfrozen and allowed to develop into babies) I don't think any reasonable person would condemn you for it aside from your own self. If you don't belive in God, then what's the point anyway. And if you one day come to believe and have faith in Jesus, he will have compassion and forgiveness for you.
Please bear in mind that most of these potential consciousnesses die. Also bear in mind that something like 50% of Americans signed up for interventions which would directly lead to "the potential for consciousness" being terminated. (ahem, okay, only half of them were female, and some of not childbearing age -- you still didn't see people (who weren't Jewish and in Israel) running around issuing fatwahs about the intervention).
True, but lots of unfortunate things happen, I don't think this can be used to justify purposely causing them to happen. I'm not sure what you're referencing with 50% of Americans, but i've seen the choices approximately 50% of Americans have made, and it doesn't really inspire confidence in them being a barometer for morality.
>How is this different from saying that a couple that has three kids but chooses not to have a fourth is murdering their fourth kid, and exactly as culpable as a couple that has four kids and then kills one?
The standard answer would be that fertilised eggs, unlike hypothetical people, have at least some form of moral personhood, by virtue of having a complete complement of DNA.
If you trace a person back through time you'll find they change a lot as they regress into a child, then a baby, then a fetus, but there's one discontinuous point beyond which you can't say that "this person" still exists, and that's the point where that egg combined with _this particular_ sperm rather than any other.
That's the standard (non-religious) version of the Life Begins At Conception argument, anyway.
A sleeping hermit has a mind, even if their mental states aren't being actively processed. It's completely different from merely having a "potential" to form mental states after a bunch of further development.
I think there's an absolutely decisive argument that embryos aren't harmed by death. It's that only beings with minds can be harmed at all.
Many philosophers (from Locke through to Parfit) have been convinced that some kind of psychological continuity is what matters in survival. Here's why: If a mad scientist scanned my brain, disintegrated my body, and then wiped your brain and implanted all of my mental traits (memories, beliefs, values, personality, etc.) in its place, then it seems that I have gotten the better half of the deal. I have survived and you have not. Though your body is the one that lives on, it is our minds that matter, and it seems that your mind has been replaced by mine. In making this judgment, we implicitly judge that it is the content of a mind that matters -- the memories, beliefs, desires, and so forth -- not its "location" in a particular body, or even a particular brain.
Suppose this scenario goes ahead. I wake in your body and return to my old life (as best I can). Does this make any difference to you? Suppose you're told beforehand that after your mind is wiped and replaced with mine, I'll go on to live a happy life. Or maybe you're told that I'll be killed the next day. You might for altruistic reasons prefer the former news, but do you think that you are harmed if the latter outcome occurs instead? If not, this goes to show that preventing future pleasures from being experienced in your body is no harm to you, if you are not the one who will get to experience them.
It follows that mindless embryos cannot "survive" (in the moral sense, i.e. mentally), and hence aren't harmed or deprived of anything by their biological deaths.
I follow what you're saying about the mind being more important than the body, but I dont see how this refutes the idea that denying a pre-conscious, extant being the opportunity to become concious is morally wrong.
Call the embryo 'Emby'. If you agree that you aren't benefited by having me live on in your body, then you should also agree that Emby isn't benefited by having some future person live in their body. Emby can't "become" conscious, any more than you can give my next lecture after I take over your body. Bodies are just husks. It's minds that matter, and Emby isn't one and never will be one.
It seems the crux of our disagreement is the idea that "It's that only beings with minds can be harmed at all"
I think it's pretty common in EA to consider the welfare of future humans who don't even exist yet, much less have minds, in moral calculations. Scott even does in this article ( along the lines of think of the future person who won't have lung cancer because someone selected against it in ivf)
How do you handle this? Do you think future generations have moral standing? If so, how do you square this with them not having minds?
Merely possible people (who never end up existing with minds, e.g. because we go extinct first) can't be harmed. But harms aren't all that matter. It would be better to bring about a future with more happy, valuable lives, and I think we have moral reasons to bring about better rather than worse futures for the world, all else equal. I explain my view of existential harms and benefits in more detail here:
I read your post. I found myself nodding along but wondering why it's reasonable to prefer Joy's existence and not Emby's, especially since Emby is already further down the path to consciousness than Joy.
I believe that you can harm a future person who will exist by making their life worse than it would otherwise have been, but you can't harm a never-existent person by making them not exist at all.
So if I use drugs while I'm pregnant, and this causes me to give birth to a baby with some sort of disability, I am culpable. But if I decide not to have a child, I'm not culpable. And this is true even though if we start with an existing person, killing them is worse than disabling them!
So it's fine to care about future generations in the sense of having good policies that protect the environment for future generations or something, because you are making the lives of people who will exist better off.
So you care about the lives of future generations of people who are theoretical and do not exist yet, but don't care about the existing lives of future people, embryos.
Go back in time to when your kids were embryos. If someone had the power to and decided to destroy them when they were still just a couple dozen cells, and replaced them with different embryos (sorry I don't know what kind of twins you have so modify as needed) would that seem wrong to you?
Now what if someone had the power to, and decided to change which sperm and egg met to create your child/children. There were never embryos of your existing kids, just different ones, but the result is the same as the above example. You wouldn't have the same kids you have, but would it still feel morally wrong?
When I consider this, I think the first scenario is something akin to murder, but the second is not.
This is good insight into EA. And quite predictable from the very observation of their concern with remote future --which does not exist.
They display quite an inversion from the normal expanding zones of moral concern. One goes out from concern with people near one and time close by, to progressively less concern about places and times distant from us.
Anything whatsoever might be justified by the inversion of this normal process.
Arguments are futile in moral matters. Since arguments can proceed to conclusion only if premises are shared. But between materialists and non-materialists key premises are not shared.
Not sure about "futile". It may be futile discussing the "morals of IVF" outside a real world. But IVF is as much real world as are IUDs, miscarriages and all people alive on this planet. 1. Sure, one can argue that condoms are ok, the pill ambiguous but IUDs 'clearly' morally abhorrent, as they keep a fertilized egg from implanting (why my teacher for Catholic religion was vs. IUDS). In real life, people call them all contraceptives and near all use them according to their preferences (STI a potential problem? Not feeling well with the hormones? Long time planning?), not to "moral" considerations. Looking at miscarriages, one may call 'our Lord: the greatest aborterOAT' - but then it may be for the better not to have all embryos become toddlers.
IVF: if one asked Emby: "Hey, you are in an IVF-clinic, together with 3 siblings: we will check you now and implant the "best one". The other go in the bin. Are you willing to take your 25% chance or shall you prefer the bin right away?" - a smart emby would say: I take my chances.
And sorry to our extreme pro-lifers, humans will not implant AND deliver all their IVF-embies. Nor will they refrain from IUDs nor will they have only 2.3 times sex in their life. Except the protestant couple after "every sperm is sacred". ;) It is selection all the way down.
Fwiw, it's not like iuds are the only option. Condoms, surgery, and the pill (depending on who you believe) solve the problem of wanting to have sex but not wanting to destroy embryos.
Just to check, do you disagree with the following premise?
> "If a mad scientist scanned my brain, disintegrated my body, and then wiped your brain and implanted all of my mental traits (memories, beliefs, values, personality, etc.) in its place, then it seems that I have gotten the better half of the deal. I have survived and you have not."
(fwiw I'm not a materialist, but I still accept a psychological theory of personal identity)
More materially (ha), I fail to appreciate the special perspectives within the embryo vs person debate in the first place. Surely these moral objections either way are themselves the product of Fisherian runaway selection - the heightened awareness of a particular trait with selective advantage. What genetics has originated, genetical intervention may cure.
And as for the hard sell of non-materialist conception, I suppose it is hard sell to those to whom anything smacking of non-materialism is hard sell per se.
But I suppose, to the general population, the hardgoing materialism is a harder sell than non-material conception of personhood.
Is it really correct to say that you have reduced someone’s risk of breast cancer by 46%, if what you’ve really done is closer to replacing them with a different person who is 46% less likely to have breast cancer?"
I don't see this is a question. It's not "this is an existing embryo which is a potential baby and you're just picking the traits for them", it's "this is a collection of existing embryos which are potential babies and you are selecting out one of them to be born by a checklist".
By picking "embryo with the lower risk of breast cancer" you are choosing to let that one develop into a baby that will be born. The other embryos will either be re-frozen for a later attempt if needed, or destroyed. You haven't "reduced someone's risk of breast cancer" (as if, for example, they were an adult and a new drug was invented that stops cancers developing), you've killed the competitors who might have been born and left the survivor with the least risk to be implanted.
Now, if you had *one* embryo and you were doing a little snippity-snippity with CRISPR or something to go "delete this bit here, stitch in a new bit there", then yeah: you're reducing that (potential) person*'s risk of breast cancer. But that's not what this procedure is doing. The other embryos are perfectly able, if implanted, to develop into babies that could be born. You're killing those embryos or at the very least preventing their natural development by sticking them back in the freezer, so you are taking potential people out of existence.
EDIT: And no, that's not the same argument as "you have therefore to have as many babies as possible to allow all potential people to come into being". These embryos now exist in reality and in potential to be viable pregnancies. They're not theoretical or notional, they're the result of someone's sperm fertilising someone else's eggs and are now ready to follow the path of conception to birth, only you chose to do all this in a petri dish instead of by the body.
*I think embryos have personhood but yeah: dumb old believer in stupid sky fairy story here.
Thank you for saying this! I don’t think CRISPR would work on any stage but the zygote (well, in the case of something like cystic fibrosis maybe if half the person’s cells are able to produce the missing protein then the effects of CF wouldn’t be as devastating). But anyways I consider IVF immoral so what do I know. It really makes me sad to see all these rationalists’ inconsistencies in moral reasoning and reminds me that as much as I find interesting a lot of content on this blog the fundamental framework is so different we might as well be living in different galaxies (though Father Brown would remind me that the natural law applies everywhere in the universe). We can’t have honey apparently because of bee suffering but creating embryos just to let them die is a-ok!
Ah yeah, I got into trouble with a former member of the commentariat here because of my Catholic objections to IVF (it was a personal family thing with them so naturally they did feel hurt).
I mean, I think the people currently proposing this do have the best intentions about healthy babies first (get rid of the hereditary diseases) and then 'who could object to smart, happy babies?'
But it's not going to stop there, because it never stops there. The particular genetic selection company may be the first one Scott heard of offering to sell hair/eye colour selection, but a couple years back there was at least one chain of fertility/IVF clinics offering this as a deluxe option for clients.
The fertility industry is an industry, and the purpose of business is to make money. So if people want purple eyed kids with silver coloured hair and six fingers on each foot (yes, I said "foot") and are seven feet tall, and the parents have the money to pay for it, that's what they'll get.
Wait, is your conception of the abortion debate that it's a debate between two sides who both have competing materialist conceptions of personhood? This would be a surprise to me, as I don't think anybody but the most hardcore of transhumanist/rationalist/atheists have a purely materialist conception of personhood.
The anti abortion side is a pretty straightforward metaphysical claim. The soul exists, and is created at conception.
Meanwhile the pro-abortion rights side contains dozens of competing definitions and thresholds of personhood, nearly all of them with some nonmaterialist component at their heart (e.g. "personhood only applies to living things" or "personhood only applies to humans" or "personhood only applies to fetuses post-viability" or "personhood only applies after it exits the womb" or "the threshold for personhood is whether the fetus has nervous system that can respond to external pain") The purely materialist conception (one example is purely physical capacity for information processing, e.g. IIT) is hardly held by anybody except a minorities of principled philosophers and rationalists. And to be honest I think even most of these are fundamentally metaphysical claims, with the possible exception of "personhood simply doesn't exist" (which I personally disqualify on the basis that it seems to imply that ethics don't exist either, which in that case, what's the point of arguing about the ethics of killing a fetus)
Do shrimps have personhood? Because Bulldog, for one, is very much active in ending shrimp suffering and getting us to treat shrimps as if they have rights.
As humans, we have a long, long history of treating "this set of us is fine and are people, those set of them are not fine and are not people so we can do what we like to them and that's okay, we're not doing anything wrong because they're not people".
Making personhood a revocable legal status that can be granted and taken away at whim is not a good precedent.
There's already a push for miscarriage leave here in Ireland (we don't yet have a legal entitlement to such, although there is for a stillbirth). So is a seven week old or twelve week old foetus a baby or not? If you want it, it's a baby and you are a grieving expectant mother if it dies. If you don't want it, it's a clump of cells and you are entitled to terminate that pregnancy and expel the contents of the uterus.
"Maybe baby" is not a great way to think about this entire topic, because if we're worrying about AI rights as future sentient beings, we're already laying the legal and social groundwork for "aw, that ain't no person if it's inconvenient for you for them to be a person".
And I don't think it'll stick at the foetus, either. We've done the "adult human but not a person" thing already in several instances, why not in the Brave New World now the unedited, unselected, stupid (because everyone is supposed to be 115 IQ which is the new normal, not a dumb old 105 IQ clod) human-looking animals are not in possession of personhood as an innate property of their humanity, so we don't have to worry about them. In fact, the kindest thing to do is put them down humanely, like we do with feral animal populations. They just can't have a decent quality of life under the circumstances that society demands of its citizens, they're too lacking.
I can't remotely join you on the idea that eight cells has the moral personhood of a newborn baby. I can clarify that IVF doesn't generally any longer involve implanting more than one embryo at a time.
I can also add the reality check that women have to still biologically produce the eggs to be fertilized and that the multiple and high levels of hormones used to produce that many eggs at once is an intense and painful experience and the numbers of eggs that can be produced in any one cycle is quite limited. The average number retrieved in each cycle is 8-14.
"Retrieval" is a surgical procedure under general anesthesia and the process leading up to it is hard and prolonged and demanding. Not all the eggs retrieved will be viable on their own (ie, before being fertilized), others will not grow after being fertilized. It's a long, slow process in other words, and one that does not yield up bunches and bunches viable eggs or embryos. It's extremely hard on women in particular, physically and psychologically.
After you replied, I addressed the question of the embryo's moral value in response to Scott's comment.
As for IVF and multiple implantation: noted, but I hope you realize that this reinforces my point about the seriousness of PS relative to standard IVF.
Regarding retrieval: yes, that's how it currently works, but there's a lot of effort going toward widening that bottleneck. For example, Jonathan Anomaly is banking on in-vitro gametogenesis so that "tens of thousands" of embryos can be created from a single vial of blood.
And will we then have "tens of thousands" of births? I'm thinking "Hell, no!" so we now have the potential for massacres in a test tube.
Ain't science wonderful! Shame our moral intuitions can't remotely keep up with our technological ability.
See how easily we've moved from "prevent preventable hereditary diseases" to "mass produce humans, mass murder humans, but all in the clean scientific way not the messy old King Herod way".
Mostly orthogonal to your point, but I’ve met some couples who use polygenic risk scoring without destroying any embryos. They use the risk scores to inform the ordering of their IVF implantations.
One of these couples reasons along the lines of: “If my child has an elevated chance of breast cancer, then it’s better for them to be born later and hence closer to its cure.“
That's funny (but reasonable!). Do they actually plan to implant all of them? I've always wondered if there are ivf clinics (or patients) that are able to factor in the worldview that embryos have moral standing
For one of the two couples, I overheard them saying that they will implant all embryos at an Orchid Health presentation, because it’s important for IVF to be compatible with their religious beliefs.
The other couple is Malcolm and Simone Collins, they just posted about this on X recently.
beginning at fertilization, a human embryo has moral value that's similar in magnitude to a newborn baby […]
Consequently, that should make EAs very reluctant to advocate [embryonic destruction]
I don’t envision any reluctance: most Effective Altruists are hedonic utilitarians and I think would very straight forwardly reject your argument using their moral framework.
Hedonic utilitarians believe that moral value lies in the sum total of pleasure experienced minus pain experienced across time and space.
Early stage embryos (dozen cell structure) don’t feel pain or pleasure so under this moral framework destroying them is not directly immoral. In fact, hedonic utilitarians will reason that under resource constraints, it’s better to selectively choose the embryos that create people likely to have higher hedonic utility.
I disagree with hedonic utilitarianism because people and other beings can have lives that are net negative from a hedonic perspective, but they still find to be meaningful and worthwhile. It would be wrong to tell these people that it would be better that they were never born (they would disagree, even if a negative utilitarian would tell them they are “delusional”). Furthermore, I am even skeptical about whether pain and pleasure are inherently meaningful. After all, these are “arbitrarily” evolved signals from an ancestral environment we no longer inhabit. When we have the technology to edit such signals, should we in fact coerce all people into wire-heading in a pleasure pod?
Haven’t deeply thought through my moral philosophy, but I’m certainly a consequentialist, I highly value “survival”, and I consider “myself” to also be my descendants (e.g. my descendants are my New Game+, and I am a New Game+ for my ancestors).
I was wondering about coercive wire-heading, that seems to be the most obvious issue with hedonic utilitarianism.
There is also the question of what this evolutionary signal is _to_. Consciousness? If causing harm to others is bad in part because of this pain signal, it seems there is _some_ inherent meaning, although I wouldn't count myself as a pro wire-heading hedonic utilitarian either.
Survival for survival's sake is what cancer is though, I feel like this has to bottom out in "meaning" somehow. But maybe that is a hard-problem-of-conscioussness-complete problem.
I don't think most EAs are hedonic utilitarians - most are utilitarian or utilitarian-leaning consequentialists (I think the 2019 survey was the last one to have this data, where just under 70% were utilitarian consequentialists), but I think further qualifying with 'hedonic' would cut that down a lot, at least in the traditional sense where experience is the *only* thing that matters (and even more so in the pleasure/pain sense you talk about below). I'm not aware of any solid data, but my guess would be that less than half of utilitarian or utilitarian-leaning EA consequentialists are pure hedonic utilitarians.
Thanks for your correction, and for pulling up the EA survey stat. Do you know how typical EA utilitarianians add more nuance beyond hedonic utility in this context? Words that come to mind include "flourishing", "personal preferences", and QALYs but not sure if this is how people would typically frame their utility functions.
I've never heard of the names Gene Smith or Jonathan Anomaly. Are those people's actual names or is it the title of some blogs? If the former, that's some serious nominative determinism.
I think it's worth mentioning that a very large number of embryos fail to develop even in natural pregnancy. I've seen people argue about the figures, so I'm not sure what the exact percentage is, but it's probably pretty high (https://pubmed.ncbi.nlm.nih.gov/28580126/).
This doesn't directly contradict your point, (embryo selection could still theoretically increase the number of discarded embryos) but it does confuse me. Would moral standing at conception suggest that we should *only* perform IVF and ban regular conception to avoid the associated death? (Given the current popularity of non-IVF conception surely it is the greater cause of embryo death?) Would embryonic personhood then be comparable not with selection, but CRISPR modification?
Right, it's repugnant conclusions like this - that in order to avoid creating naturally doomed embryos, the only ethical choice is invasive and alienating procedures like IVF - that are my strongest reason for not being too serious about embryonic value/personhood/rights. That's crazy.
I can see both sides of the argument otherwise. I suppose I come down in a gray area, where I'm uncomfortable enough with IVF for embryonic personhood reasons (and miserable maternal procedure reasons) that I probably wouldn't do it myself unless I were desperate - but there are plenty of people who are understandably desperate enough that I think the value they gain from it outweighs the small value/chance of personhood of the embryos lost.
The movie Gattaca was a utopia for everybody except the protagonist.
If the protagonist's parents had been responsible like apparently everybody else and had a genetically screened baby, then he wouldn't have had any problems.
The movie focuses on his problem that he can't get the job he wants, but the underlying far bigger problem is that he has a genetic defect which may cause him to drop dead in his thirties.
The solution to "it sucks if some people aren't genetically engineered" isn't "nobody should be genetically engineered", it's "everybody should be genetically engineered".
Q: Hi ChatGPT! I hope your early morning is going well. Can you please tell me how many clearly distinct arguments there are in the interminable abortion debate? I realize that these can be combined into a (roughly) exponential number of stances, but, for _just_ the distinct arguments, how many are there? Many Thanks!
A: There isn’t a canonical count—philosophers and legal theorists carve the terrain differently—but you can compress the debate into ~18 argument families. Most specific positions are combinations of these.
Core families (one-liners):
1. Biological humanity → full moral status. Human life begins at conception; killing it is wrong. (SEP overview.)
Stanford Encyclopedia of Philosophy
2. Psychological personhood criteria. Rights track capacities (consciousness, interests, rationality); early fetuses lack them. (Tooley.)
JSTOR
3. Potentiality / “future like ours.” What’s wrong with killing is depriving a valuable future; abortion (usually) does that. (Marquis.)
edX
4. Bodily autonomy. Even if the fetus is a person, no one may conscript another’s body (the “violinist” case). (Thomson.)
5. Responsibility for pregnancy. Voluntary risk-taking (sex) can create obligations that limit abortion claims; dissenters dispute this. (Discussion in SEP.)
Stanford Encyclopedia of Philosophy
6. Self-defense / double effect. Life- or health-saving interventions can be justified even if fetal death is a foreseen side-effect. (Foot; SEP.)
PhilPapers
Stanford Encyclopedia of Philosophy
7. Gradualism & thresholds. Moral status may increase with development (heartbeat, brain activity, consciousness, viability). (IEP/SEP surveys.)
Internet Encyclopedia of Philosophy
Stanford Encyclopedia of Philosophy
8. Viability as a legal/moral line. Before viability state power is weaker; after, stronger (contested and technology-dependent). (Roe/Casey/Dobbs.)
10. Privacy/liberty vs. state interests. Liberty (privacy) grounds a right to decide; opponents stress the state’s interest in “potential life.” (Roe/Casey/Dobbs.)
Justia Law
Library of Congress Tile
Supreme Court
11. Public-health / harm-reduction. Restricting access doesn’t end abortion; it makes it less safe and worsens health outcomes. (WHO, CDC, ACOG.)
World Health Organization
CDC
ACOG
12. Safety comparison. Legal abortion (especially early) is far safer than childbirth; policy should reflect that. (ACOG.)
ACOG
13. Moral uncertainty / risk. Given deep disagreement, precaution (for or against abortion) may supply defeasible reasons. (Moller.)
Squarespace
14. Fetal pain. Some argue pain perception grounds later limits; mainstream reviews place the onset ≥24–25 weeks and dispute earlier claims. (RCOG/ACOG.)
15. Disability-rights (“expressivist”) critique. Selective abortion can convey harmful social messages about disabled lives. (Parens & Asch and successors.)
PubMed
16. Sex-selection / demographic harms. Even if abortion is generally permitted, selection for sex is challenged on justice/inequality grounds. (Ethics literature reviews.)
University of Birmingham
17. Socio-economic consequences. Denial of wanted abortion increases financial distress and harms health; access improves life outcomes. (Turnaway Study, NBER.)
NBER
ANSIRH
18. Paternal interests & parental obligations. Disputes over the moral/legal claims of putative fathers and over duties to gestate/adopt-out rather than abort. (Surveyed in SEP.)
Stanford Encyclopedia of Philosophy
If you like, I can map these into a matrix showing how different policy packages (e.g., “early liberal, late restrictive,” “health-only exceptions,” etc.) combine subsets of these arguments.
Three possible objections I can think of that you don't discuss:
1. You mention how IVF is expensive and very unpleasant, but it might also just have negative effects on the embryo by itself (I know zeta_globin on Twitter is always banging on about this). If we're lucky the selection might yield a net improvement? I don't know, I don't really have enough knowledge here to make this argument or know how strong it is, but I've seen it out there so I thought I should mention it.
2. This is another argument I'm stealing from *someone*, but I don't know who (I remember asking in the comments here once and someone said it was the Unabomber, but I could not in fact find it in the Unabomber manifesto). So far it doesn't sound like this is allowing any sort of selection on personality traits (unless one considers IQ a personality trait). I'm mildly worried about the possibility of selection for agreeableness. That is, I'm worried that parents might select for agreeableness in the hopes of having an easier-to-raise kid, but that this occurring on a large scale might lead to increased conformism that holds back humanity. (Though selection on some other personality traits sounds pretty nice -- who doesn't want lower neuroticism or higher conscientiousness?) (I guess arguably this is covered under the "Selection" category, but it's a little different, I think, in that it's not purely individual.)
3. I can't find it at the moment, and this is even more arguably already covered under the "Selection" category, but I remember an old blog post by Ozy about the possibility of an even more extreme form of the downsides under "Selection" -- i.e., that awful parents might completely disregard their child's welfare in their selection (think of what's happened to certain dog breeds, bred for aesthetics). Like, abusive parents exist, so! Not sure how great that is as an argument because it strikes me as perhaps a bit too "criminals use cars", but thought it was worth mentioning (can anyone find the post? For some reason I can't find it right now!).
Yes, #3 is a good point -- if this becomes generally used/accepted (and hopefully not banned), laws around it will (hopefully) be made to cover this sort of thing.
For #2 that is a counterbalancing factor, but if you get clustering then maybe there's a problem? OTOH, selection on higher IQ -- which is generally desirable -- would probably do a bit to mitigate that...
Natural forms of #3 are, to my knowledge, not illegal currently: AFAIK you are fully within your legal rights to have a child even if you know they will (or likely will) have a serious disability or fatal disease. For example, the couple in this article: https://www.nytimes.com/2022/11/09/health/pompe-disease-treatment.html
I was playing with the Herasight widget for various conditions (plus IQ), and it tells me the expected outcomes for 5, 10, or 20 embryos estimated by their genetic predictors. But I don't know what the implications of these values are. Does this imply they can select the best out of 5, 10, or 20 embryos to obtain the lowest liability (R-squared for risk variance) for that condition? I assume that's what they're claiming, but without a comparison of liability scores between the embryo counts. If not, what are they trying to prove with this widget.
They're not claiming that they can select the best with certainty, they're making a probabilistic claim that they'll do, on average, that much better than average.
If you haven't given them a real set of embryos, then they're simulating embryos with a normal distribution. For example, if two IQ 100 people marry each other, then on average their children will center around IQ 100 with some variance. I don't know exactly what that variance is, but they do and they've calculated it. Maybe on average, five embryos will range from 90 - 110. So if they could always choose perfectly, they could pick the 110 one and get you 10 points above average. But since they can't choose perfectly, they only expect to get you 5 points above average.
That's what I would have expected. But the widget tells me that with 20 embryos from 2 European parents whose IQs are 100, we can expect their child's IQ to be plus or minus half a standard deviation above or below the parents. The kid could have an IQ of between 93 and 108. I don't think they're claiming that, but somebody needs to talk to marketing about improving their value proposition. ;-)
I was playing with their widget. It displays the values of the expected outcomes for 5, 10, or 20 embryos estimated by their genetic predictors. I presume they're claiming they can select the best of 5, 10, or 20 embryos—but it doesn't give me a readout of liability scores. So I can't really make a risk/benefit calculation. Am I missing something obvious here?
Another (tongue in cheek) objection. Your parents will be absolute morons compared to you at some point. You’ll be a mopey 14 year old screaming at your parents for not making obvious stock market picks and running their own companies.
On average, smart parents' children will be less smart than they are, because of regression to the mean. Currently, I think this technology just helps you tread water. I agree that this will be true in 10 or 20 years when the technology is much better.
150 IQ is 3.3 SD above the mean. From an eyeball at a statistical table, 1 in 2000 people are this smart or smarter. And several of them may be reading this blog :)
Typically children of this sort are much more interesting in doing things in their own lives rather than controlling what they do with their parents lives. They're building full-flow 2 stage model rockets with landing capability rather than crying to their parents that they aren't supermen.
Teenagers feel this way about their parents regardless of the real relationship between their levels of intelligence, it'll be nice for them to actually be right about being smarter for a change.
The main thing I'm hoping for from this research is that it gets us closer to adult genome editing. It'll be really awkward if I'm in my 40s and there are 10 year olds smarter than me (Assuming AGI doesn't happen by then and render genetics completely moot).
If we're really concerned about improving our future children's genes, it's worth noting that compared to IVF and embryo selection choosing a partner with strong genetic traits (e.g. intelligence, health, height) would have a far larger impact.
The genetic variation between potential partners is much wider than the small variation among embryos from the same couple. So if you believe improving genetic outcomes is a moral good, does choosing a partner for love or chemistry over heritable traits amount to prioritizing our own happiness over our future child's potential?
I feel like some people already do this! Prioritizing someone hot/smart/healthy/wealthy over other people that might be nicer to you seems like a commonly discussed phenomenon
When i was dating many of my peers got genetic testing to screen for heritable disease that they might be a carrier for. There was a service that saved the results and you could then check potential partners for the same recessive disease. It would tell if you are both carriers but not what specific disease.
I think those traits are typically considered for their first-order effects on the relationship, or *maybe* for their relevance for providing during the reproductive years, not for their genetic effects on the offspring.
I don't think choosing partners for the genetic traits we want our kids to have is at all a commonly discussed phenomenon.
I sure never thought about it. Both of my serious relationships were with people who had siblings with relatively common heritable disorders. I was basically like "huh weird, I guess that's cool/sucks," and not once did it cross my mind that it could affect my potential kids.
Even less data on multi-racial / mixed-race genomes. Multi-racial couples are out of distribution for these machine learning models and plausibly will get the least value from these embyronic screening services.
We don't currently understand much about genetic interactions, either because they rarely happen or because our current tools aren't good enough to detect them. Since embryo selection can only use the genetics we understand, interactions aren't relevant here, and predicting mixed race couples works fine using what we know about the component races.
So for example, on Herasight's calculator, an Asian couple with five embryos can gain 4.1 IQ points, a white couple can gain 5.3 IQ points, and a mixed white-Asian couple can gain 4.8 IQ points - right in between.
And he's free to ban you for being a stupid jerk. My argument for the assertion that you are a stupid jerk is that you didn't provide arguments for your assertion, refused to do so when he asked you to, and said you don't care about his preferences.
Opinions still have an internal logic to them, unless you are claiming to have arrived at yours by rolling dice. If you hope to convince people to adopt yours, you will generally need to engage with that internal logic. For example, you could say *why* you think this will lead to "a very troubling future", or what you think will be troubling about it.
As it happens, I *have* read Brave New World, but to my recollection it objects to nonintellectual hedonism and the creation of a low-intelligence underclass. If you also object to creating healthier, more intelligent people, and think Brave New World is criticising that goal, them I think you have misread the text.
The creation of a high-intelligence upperclass implies the creation of a low-intelligence underclass, even if no active measures are taken towards the latter. Implying it's an intentional, active process may be a POSIWID fallacy because nobody is proposing to bathe embryos in an alcoholic solution or whatever exact mechanism BNW offered, but the overall result is bound to happen anyway.
It turns humans into products. It creates rigid castes divided by intelligence level. It eliminates family structure. It’s a horrifying dystopian world that we are running to as fast as we can.
You're missing the point. People who need to use IVF because they cannot naturally conceive a baby will of course seek to have a healthy child if possible. But that's not what this whole piece is about: It is about people deciding to put a woman through the painful and difficult IVF process just so they can get an athletic kid with blue eyes or whatever.
It's really disingenuous to pretend trying to have a healthy child free of a terrible disease is the same thing as choosing a kid who is tall.
And to use IVF tech when there's no infertility problem is ethically questionable.
There are dozens of people with your opinion participating in the conversation here. Nobody cares that you have this opinion, but this space is used for discussion. The reaction you're getting is not because anyone disagrees with you or feels you shouldn't be allowed a voice, but because you're treating this like a facebook post. You can think and say a lot more here than most other places; you're just expected to contribute signal, not meaningless noise.
but you still didn't answer the question... if you don't support a woman's right to choose what goes into her body, just say so.
(in case it's not clear, this is, indeed, a rhetorical device, pitting one purported value against another in an attempt to get you to make a coherent argument, rather than just being enraged)
An issue that I have yet to see addressed well enough is how sure are we that there is no added developmental risk from the procedure itself that would counterweight the benefits, especially when the benefits are "add an expected 4 IQ points" rather than "not have CF". The risks I'm thinking are mostly:
1) Stress to the embryo from the sampling, either in inducing developmental mispatterning (more likely) or de novo mutations
2) Stress to the embryo and the mother from the IVF (development/mutations for the embryo, hormonal and immune signaling mess for the mom), if comparing against in vivo fertilization
There just seem to be so many risks there that even if each one is very tiny change in odds because doctors optimized it excellently, they should add up to more than 4 IQ points.
Polygenic selection isn’t just a tweak to medicine or parenting. It’s a decision about who gets to exist, and why. But this conversation keeps retreading familiar grooves without confronting the ontological vertigo this new technology introduces.
We’re not choosing a school or a tutor; we’re choosing which possible person becomes our child. That crosses a metaphysical threshold. Potential humans are now compared probabilistically—not for life or death, but for entrance into life itself. And this comparison is based on traits that increasingly resemble KPIs: IQ, BMI, schizophrenia risk, even eye color. That's person-as-product, full stop.
We need plainer language for what’s happening. The core ethical shift isn’t “screening embryos” or “preventing disease.” It’s that the value of a life becomes a function of its forecasted traits. And you know once that logic takes hold, it won’t stay confined to embryos.
The risk isn't Gattaca-style coercion. I'm more worried about soft cultural convergence. No laws required. Just dashboards, rankings, and best practices. And because it feels like choice, not imposition, no one will notice the narrowing until it’s a fait accompli.
Maybe this tech does increase aggregate well-being. But it also reshapes the moral substrate beneath our choices. When you start selecting children the way you select college applicants, your deciding what kind of soul you think is worth betting on.
Is this also true when you take prenatal vitamins to prevent spinal deformity? Isn't that selecting for children who don't have spinal deformity over those who do? Or that the value of a life without spinal deformity is more than one with it? Should we ban prenatal vitamins?
I think there’s a category error here. Taking prenatal vitamins supports the development of a child who already exists. Polygenic selection chooses which possible child gets to exist at all. The former is about nurturing; the latter is about filtering.
Once we’re filtering not for catastrophic disease, but for probabilistic forecasts of intelligence, temperament, or BMI, we’ve crossed from medicine into metaphysics -- defining what kinds of human potential are “worthwhile.” That ain't just banning vitamins.
What happens when the logic of actuarial triage becomes our primary mode of making people?
What you call "catastrophic disease" is also a value judgement. It's a very common value, but saying it's better for a child not to have a spinal deformity is just as much a decision about what kind of lives should be lived as selecting based on predicted IQ.
Not all value judgments are equally loaded. Shifting from healing a child to curating one begins to apply a logic of comparative worth to potential lives. If we’re going to do that, we should at least be honest about what it is.
I'm not sure what you mean by "loaded", but keeping your child from being born with a spinal deformity is absolutely applying a logic of comparative worth to potential lives. At the time you make the decision both lives (with spinal deformity and without) are merely potentials, and your decisions change their relative likelihood of being realised. People act on this because they think the life without the deformity is better than the life with.
Yes, choosing to avoid spinal deformity involves a value judgment. But there's a difference between reducing the risk of profound suffering and optimizing for predicted traits like IQ or BMI. One stems from a widely shared ethic of care; the other starts to encode comparative desirability into existence itself.
That’s what I mean by "loaded": some judgments operate within a shared moral horizon. Others begin to redraw the boundaries of which potential lives are seen as worth bringing into the world.
I don’t think the argument is that the lives of children with spinal deformities are worth less than other children. I think it’s that it would be preferable to live in a world where children with spinal deformities can receive appropriate treatment and accommodation, as opposed to a world where those children simply do not exist.
This tech proposes to “solve” a “problem” by engineering a situation where the “problem” doesn’t exist. That’s fine for some things, but I don’t think it’s crazy to get uneasy when the “things” in question are people.
What’s weird is that this tech simply lets parents choose one randomly generated embryo over another. If we assume there’s no a priori reason to choose one over the other, it doesn’t seem crazy to let the parents’ preferences break the tie. And right now, these preferences are things like “maximize the chance of a pregnancy that’s successful and safe for the mom” and “minimize the expected health care costs over a lifetime”. Not unreasonable!
I do think parents could be a little more honest with themselves about how much these choices are being made for their own sake, rather than their prospective child’s. It’s fine not to want to go through the gauntlet of raising a kid with a rare genetic disorder: I image it’s a huge emotional and financial weight. But probably don’t say the choice is for the child’s sake, because it’s not hard to see how that could shade into saying it’s best for such children not to exist at all.
In my opinion everything gets way more fraught if you start designing embryos from scratch. That’s not what this tech does, so I’m mostly agnostic about it. But I do get worried about slippery slopes, because humans haven’t had the greatest track record in this domain.
>I think it’s that it would be preferable to live in a world where children with spinal deformities can receive appropriate treatment and accommodation, as opposed to a world where those children simply do not exist.
FWIW, I have the opposite preference. Treatment and accommodation have their limits.
Indeed, there is a spectrum of conditions along which I think reasonable people can have differing preferences. A spinal deformity is perhaps not the most interesting example for considering the ethical consequences of this tech, due to the nature of the associated treatments/accommodations. But what about deafness? Or ADHD? Or left-handedness?
(To be clear, I’m not trying to play “gotcha” here, I’m just trying my best to pinpoint the crux of the disagreement on this topic.)
( FWIW, I, personally, happen to be childfree and long past my potential reproductive years, so I don't actually have a dog in this fight... )
I think the best public policy course of action is to leave these decisions up to the prospective parents. As you said
>Indeed, there is a spectrum of conditions along which I think reasonable people can have differing preferences.
Hopefully, the parents have their child's best interests at heart, and will weigh all the possible factors, from catastrophic ones like severe spina bifida to near-neutral ones like left-handedness, in that light. If they _don't_ , the child is screwed anyway, and there is very little that the State can do to remedy the situation. ( Yeah, child protection services exists - and regularly screws up in both directions. )
"And you know once that logic takes hold, it won’t stay confined to embryos."
Why should it? It's factually correct. This is the logic used by the economic system which has produced prosperity unimaginable for the vast majority of human history.
Market logic has produced staggering material gains. It’s also optimized away biodiversity, cultural memory, and non-commercial forms of meaning. Useful doesn’t mean neutral. When that same logic starts governing which kinds of people get to exist, we’re not just scaling efficiency; we’re rewriting the criteria for entrance into life.
If you think the logic that brought you cheaper iPhones should also dictate the selection of human souls, that’s…a position. But let’s not pretend nothing changed.
In any case, the point wasn’t that the logic is flawed. The point was: once it’s normalized at the level of reproduction, it will reshape what we mean by a “worthwhile” human being.
True, what counts as a “worthwhile” human has always been contested and evolving, which is precisely why we should be alert when a new mechanism emerges that may lock in a narrower definition under the banner of optimization.
As for a “positive vision”: I don’t want to halt innovation, and I’m not against reducing suffering or helping parents make informed decisions. But we need to preserve a moral space where unpredictable, unoptimized, even “low-probability” human lives are seen as fully valid -- not just tolerated, but valued. In other words: a future where technology supports life, rather than presorts it.
That will require drawing some philosophical lines before the dashboards fill in automatically.
>which is precisely why we should be alert when a new mechanism emerges that may lock in a narrower definition under the banner of optimization
Well, there are few stronger taboos in the modern world than against "eugenics", so I very much doubt that if this manages to squeak through it would be through lack of alertness.
>That will require drawing some philosophical lines before the dashboards fill in automatically.
Philosophy's track record doesn't inspire much confidence in any lines it would produce, but of course that has never stopped philosophers.
The taboo against “eugenics” exists -- mostly in name. What we’re watching now is its rebranding as choice, optimization, and personal empowerment. Norms don’t tend to collapse through brute force, but through a gradual shift in what feels reasonable.
If your view is that ethics can’t help and vigilance doesn’t matter, then there’s nothing to discuss. That’s a shrug, not a position.
I mean you already can make a decision about the characteristics of your baby that goes much further than polygenic selection: By selecting your partner out of 50% of the world's population. By choosing to have a child with one person, you deny existence to all the babies that could have been born had you chosen someone else. Of course partner choice is probably rarely made just based on gene-based traits one wants to have in one's offspring, but it certainly plays a role.
However I can see where you're coming from. One parallel might be pets, e.g. dogs: Anyone can, and most do, choose a dog breed they like from a wide range available. And I guess dog owners are really happy with that and love their dog and want it to be happy and healthy, nothing wrong with that. However, the problems start when dog owners have such a strong preference in characterisics that they choose breeds that have well known health problems, like french bulldogs or whatever. There the characteristics that the owners like trump the health concerns of the dog.
What if deaf parents lived happily in a deaf community, and select the one embryo out of five that is also deaf, so their child can experience life as they do? Would that be morally right, as most people would percieve deafness as a severe disadvantage in life? Or if the parents really, really want a blue-eyed baby, and choose that one blue-eyed embryo despite a 50% chance of cancer? Also looking at the parents that send their kids to those extreme child beauty pageants, strong guardrails will probably be have to be implemented should this tech get any further, to prevent a lot of suffering in those edge cases.
Your buffet of analogies sidesteps the categorical difference between influencing an existing person’s life and choosing which possible person gets to exist based on ranked traits.
Yes, selection happens at every level of reproduction, partner choice included. But polygenic selection crosses a new threshold by ranking potential humans by algorithmic desirability before they’ve been allowed to exist.
This has now become an existential decision based on probabilistic predictions of cognitive, behavioral, and physical profiles. Let’s at least call it what it is, rather than normalizing it through analogies that don’t fit. We're not choosing a dog breed.
> But polygenic selection crosses a new threshold by ranking potential humans by algorithmic desirability before they’ve been allowed to exist.
Do you mean polygenic selection would lead to the single alogrithm that gets applied to all humans, and only those with the maximum number of points would be allowed to exist? It's the parents who make the choice, and they do so based on their own personality and preferences. They get presented with a table of x embryos vs y probabilities of characteristics from disease to IQ, and make a choice based on their perferences. I'd say it's very much like selecting a dog breed.
The parents make the choice, but the fact that it's individual choice doesn’t negate the nature of the decision being made.
We're no longer just nurturing a child or even choosing to conceive at a particular time. We're comparing multiple potential humans based on statistical forecasts and deciding which one gets to exist -- not based on catastrophic disease, but on probabilistic desirability across a range of traits. That’s fundamentally new.
It may feel like choosing a dog breed, but dogs aren’t people. They’re not moral agents. They’re not asked to find meaning in lives they were selected into based on marginally higher predicted IQ or lower BMI.
So no, I don’t think “it's just like picking a dog breed” holds up once you follow the implications through. It’s a flattening analogy that deflects from the deeper shift underway: from accepting the unpredictability of human life to curating it in advance based on optimized projections.
Did I misunderstand your objection? Vaccination is a decision which influences an existing child.
Personally I think it's morally objectionable to deny a child vaccination because the possible outcome is one which I would call worse, even though the parents would call it better. My understanding is that in your framework, that would be okay (I'm pretty sure I disagree with that, but maybe you have a point I haven't considered).
But I certainly don't share this intuition that excluding one embryo amongst 10 is any worse of an intervention than vaccination.
I guess I can understand the reservations around genetic diversity, and I can understand the reservations for folks who believe that life begins at conception, but your reservation seems to be based on personal outrage, and I'm not sure where it's coming from.
Appreciate the clarifying effort. But no, I’m not defending MMR refusal. I flagged your example as a category error because it conflates influencing an existing child’s health (via action or omission) with choosing which potential child gets to exist in the first place.
My concern isn’t rooted in “life begins at conception” theology, or in squeamishness about medical influence. It’s in the shift from nurturing unpredictable lives to selecting among predicted ones based on ranked desirability.
Vaccination aims to protect a child already in the world. Polygenic selection, at scale, begins to define what kinds of humans are worth bringing into it. That may not feel morally weightier to you, but to me it marks a foundational change in how we relate to human life: not just caring for it, but pre-sorting it.
Except that "ranking potential humans by algorithmic desirability before they’ve been allowed to exist," does not meaningfully differ from "partner choice based on gene-based traits one wants to have in one's offspring." They're different phrases that mean the same thing: deciding beforehand what kind of children you want.
We are, in fact, choosing our 'breed' of offspring every time we decide to have children with one person and not another. We don't always know what that strain will produce, but we definitely make inferences on the basis of pre-existing traits.
The reason we currently accept the unpredictability of human life is because we cannot predict it, not because it should be fundamentally unpredictable in order to remain human. However objectionable you may find the mechanism, a change in kind has not taken place - only one of degree.
A more pertinent question would be whether we have the right to deny people the foreknowledge of genetic information, if it could be used to improve their children's lives. I don't think we do, but I'd be interested to hear your take.
Choosing a partner reflects implicit preferences, including genetic ones, but that’s not the same as explicitly ranking embryos by predicted traits and selecting accordingly. One is inference under uncertainty, the other is filtration by forecast. That’s not just a shift in degree. It’s a shift in kind.
I don’t object to foreknowledge, but to what happens when it’s used to filter human lives before they begin. The right to know doesn’t automatically imply the right to act on that knowledge in ways that redefine who gets to exist.
But people absolutely DO rank their partners on the basis of genetically-transmitted traits, with predictable outcomes for children. Tall, smart, blue eyes, etc. A shift from uncertain conditions to more certain ones is not a shift in kind, because in either case we try to make the same selections in advance.
We act on limited knowledge to redefine who gets to exist every day. The alternative is refusing screening for horrible congenital diseases because we believe that a random affliction is more moral than a conscious prevention, or actively acting against our own trait preferences for sexual partners. That's not a reality I'm eager to indulge.
When you fall in love, or even just choose a partner, you’re not running a regression on their genome. You’re drawn to a person, a whole person, whose traits emerge through culture, chance, hardship, and a lifetime of adaptation. If you hope your future child inherits their curiosity, resilience, or kindness, that’s not “ranking potential humans.” That’s loving someone and hoping what you love continues.
Embryo screening is something else entirely. It takes a cluster of possible humans and scores them on a grid of forecasted traits—IQ, BMI, mental health risk, even “educational attainment” -- and selects accordingly. That’s not preference under uncertainty. That’s predictive exclusion.
Let’s call it what it is: a shift from attraction to optimization, from courtship to computation. It doesn’t just fine-tune a familiar process; it industrializes the logic of better and worse humans.
And yes, it will change us. The tools we use to make people will reshape what we expect from them and what we think counts as a “good” life. If you think that’s just a difference in degree, I don’t think you’ve really looked.
This reads as pure nonsense to me even as someone who has misgivings about this. If you jerk off one additional time before impregnating your wife you're choosing not to let certain "potential humans" be born based on no useful criteria at all. There cannot be value to "potential humans" without ending up with completely ridiculous moral conclusions. Humans who never got past being a clump of cells are not and never were people, and have no moral value.
Libertarian eugenics can create serious issues just like authoritarian eugenics is almost certain to, but the issues arise from increased inequality and unforeseen consequences from removing certain parts of genetic diversity.
The point isn’t that every potential human has moral status. It’s that the criteria we use to decide which potential humans become actual ones reflect and gradually reshape our values.
Jerking off doesn’t involve a value-laden decision between ranked embryos scored on predicted IQ, BMI, or psychiatric risk. Polygenic selection does. The novelty isn’t selection per se -- it’s the filtration of future lives through actuarial models of desirability.
You don’t need to believe embryos have full moral status to see that shifting our selection logic from randomness or love to statistical optimization exerts long-term pressure on social norms, parental expectations, and even our idea of what counts as a full human life. The stakes aren’t just biological. They’re civilizational.
I think I agree that the immediate first-order benefits of IVF embryo selection (and by extension the type of more targeted gene editing discussed in the linked post by Gene Smith) are likely to be extremely positive, considering how much low-hanging fruit there is, even if considering only the disease risk reduction.
Still, I find myself more concerned about the long-term consequences. Gene's post on LW makes an analogy to the development of the modern broiler chicken, which seems to me more alarming than he lets on. Inevitably, if these techniques develop as Scott projects, it will lead to a narrowing of the human gene pool over time (though probably not to a single point, as I imagine there would be some clusters of coordinated traits that are favorable among different values systems).
Slightly closer to home, I look at my purebred yellow Labradors. Fantastic companion animals, and certainly not the least practical breed out there. But I cannot imagine a population of labs surviving in the wild under more adverse conditions. How many generations of hardcore gene selection does it take until we're the same? How robust is that population (to X-risks) compared to the current, naturally-selected order?
Reading some of those objections I'm reminded of that old tagline from Gattaca "There is no gene for the human spirit." A sentiment to which we can now reply "No, but there's probably about 10 thousand SNPs correlated with it."
I've never really understood the objection that with the development of AI there's no point in trying to improve human abilities. Setting aside the fact that I'm very much not convinced that AI is just around the corner, even if it is I still see the value in humans. A diversity of minds has the same benefits as a diverse ecosystem, each organisms filling and exploiting its niche. It would be incredibly short-sighted of us to neglect the development of different mindtypes.
While there's lots to get to, I'm disappointed you didn't even brush up against my main ethical concern about this technology, which is diversity. No, not the race thing -- I mean real diversity. I like living in a world full of vastly different people, with different body types, different tempers and intellects, different tastes in ice cream. A democratization of designer-babies would bring with it a homogenization of the human race towards a consensus 'optimal' form, and that just creeps me the hell out, like a lower-rate version of tiling the universe with paperclips. The natural endpoint of this technology seems to be a world populated entirely with tall, mathy, conventionally handsome, allistic, sociable people. Depending on the way society's blowing they'll either all be heterosexual, or all bisexual. By god, what a waste of a planet.
When I imagine a glorious transhumanist deep utopia, it looks more like Wonderland than Heaven; it's a world like the Culture; it's a world of endless variety, where hardly any two humans have the same skin color and the person next to you on the bus might not even have the same package of *senses*. Granted, modulo AGI which gives adult humans the ability to shapeshift, this sort of technology would be necessary to get something like that. But I don't think the humanity of 2025 can be trusted with it yet. If this goes big before something else changes the course of destiny, I think it's all too likely we'll erase all the interesting but unconventional or stigmatised variations, the left-handedness and the autism and the obligate monosexuality and the unibrows, from the genome long before we grow wise and realise how stultifyingly *boring* it's all become now that everyone is a sporty Von Neumann.
And there's another, related fear. I don't want to go extinct. Maybe *you* want your kids to be smarter than you, but I don't, particularly. There are simple pleasures I wish I could enjoy that bore me to death, because my mind won't sit still; jokes I don't laugh at because I can see the punchline coming a mile away. And sure, I don't know the thrill of winning a high-level chess game against a grandmaster; but that's just it. It's all a treadmill. I don't want our supergenius children to find all our jokes boring, our music simplistic and our novels obvious. I want children to whom we can pass our culture on into eternity; children we can laugh with. Not children to whom everything that defines who we are will look like Dora the Explorer. I think it would do them, and us, a great disservice.
> I like living in a world full of vastly different people, with different body types, different tempers and intellects, different tastes in ice cream. A democratization of designer-babies would bring with it a homogenization of the human race towards a consensus 'optimal' form, and that just creeps me the hell out, like a lower-rate version of tiling the universe with paperclips.
Thank you! I completely agree. There are many traits that probably aren’t “optimal” for maximizing socio-economic status or whatever, but society needs some % of them to thrive long-term. Tortured poets, revolutionaries, etc. Anything that reduces genetic diversity risks “optimizing” these people away. (I’m a diagnosed autist, and I don’t want to “optimize” *me* away!)
How do these concerns not apply equally to giving babies prenatal vitamins?
You're going from a more diverse state (some babies born with deformities from lack of vitamins, some healthy) to a less diverse state (all babies born healthy). And if you have a minor vitamin deficiency, then moving all kids to the no-minor-vitamin-deficiency state is driving "your type" extinct.
I think I just decide that diversity in whether people have deformities and diseases isn't the kind of diversity I care about. I care about the kind of diversity where some people are Einstein and some are Taylor Swift and some are MLK. But all of those are healthy, high-IQ people who didn't die of spina bifida at age 4.
And how, exactly, do you intend to have a glorious transhumanist future where everyone is a different type without any genetic technologies? I'm actually more concerned about this than excited - I worry at some point they'll come up with a predictor for extraversion, and half of parents will select in favor and half against, and then things will get pretty weird.
My main answer is that this technology at its current level of efficacy/cost isn't enough to cause broad social changes. And if you extend it outward until it is, we're already in the glorious transhumanist future (or at most one generation away from it, after the hyperselected people grow up and invent transhuman technologies) and we'll pass off the problem to them.
I still think the equation of raising IQ with curing obvious awful diseases is fallacious. It's not obvious to me that, past a certain threshold enabling e.g. literacy, smarter people's lives are better or more worth living; you're just moving the hedonistic treadmill around, while reducing the variety of minds that exist.
I also want to stress that my two points are separate, even if they come from similar places, and that my concern about "my type going extinct" was specifically cerebral (i.e. "I want there to always be people who think sufficiently like me to appreciate the same kind of art I liked and created"), as distinct from my worry about from the overall loss of variety incurred by erasing pattern baldness or whatever.
Relatedly, you seem to have side-stepped the risk of traits which are not objectively harmful, but are socially stigmatized, being selected against. It's one thing to be maximally spicy and say "we shouldn't select dwarfism and deafness out of existence", but my worries apply even without that sort of thing. e.g. unibrows, (non-debilitating) autism, etc. There are plenty of idiosyncrasies that are not deformities or disabilities, but which parents, for various, shallow reasons, might disproportionately choose against passing on.
> And how, exactly, do you intend to have a glorious transhumanist future where everyone is a different type without any genetic technologies?
I did raise that very point in the first post. You're right, of course. It's just that, like that *other* technology that could bring about a glorious transhumanist future, it could ruin the world as much as it could save it, and I'm not sure current humans are wise enough to use it well yet. If embryo selection remains the *only* transhuman tech for the foreseeable future, I don't think "well, that'll be our transhuman children's problem, they've got this" is a get-out-of-jail-free card. I wouldn't particularly expect the hyperselected transhumans to be wiser just because they're smarter in terms of raw IQ. Potentially they just get stuck in a bootstrap loop where the kind of types that are selected for feel increasingly icky about people who display non-selected traits, and have progressively more homogenized children until the idiosyncratic genes are lost - with the unselected humans who felt the skin-crawling horror I feel now dying of old age before they can grind the process to a halt.
Obviously the "we'll hit AGI first" argument defeats this concern, but then, as you said, it kinda defeats all concerns. I think conversations like this one are best carried out with the understanding that we're talking about the "for some reason AGI takes another hundred years/isn't possible/etc." scenario, not one where extending this particular technology to infinity has to also interact with other technological breakthroughs that would happen in the meantime.
>I think conversations like this one are best carried out with the understanding that we're talking about the "for some reason AGI takes another hundred years/isn't possible/etc." scenario, not one where extending this particular technology to infinity has to also interact with other technological breakthroughs that would happen in the meantime.
Agreed. AGI (or at least ASI, if possible) overwhelms virtually all other concerns.
If embryo selection (and, possibly, genetic engineering of some sort) becomes pervasive, for generations, I share your concern about diversity being selected out.
I agree with Scott's
>I care about the kind of diversity where some people are Einstein and some are Taylor Swift and some are MLK. But all of those are healthy, high-IQ people who didn't die of spina bifida at age 4.
I expect that the long run effect of parental genetic selection wouldn't preserve that degree of diversity. I'd expect that, e.g. we would get _good_ predictors of the big 5 personality traits. And there would be a point in that 5-D space which predicts good life outcomes. And the large majority of parents would choose genes which predict close to that outcome. And the distribution of traits would narrow in the direction of a delta function at that point.
My main reason for not being concerned about this is (a) it will take multiple generations, perhaps even centuries (b) yeah, I really _do_ expect AGI to happen first, maybe 2-10 years, barring an antarctic level AI winter.
EDIT: Re personality traits, currently the effective dimensionality of the 5-D space, while not quite the full 5-D, is a bit over 4 (from ChatGPT, Claude agrees)
Even if pressure towards the optimal choice doesn't push it to a 0-D point (there are a range of socioeconomic niches...), I would be surprised if it doesn't at least collapse to a lower-dimensional attractor.
Your mistake here lies in predicting that choice in heritability leads to sameness. The genetic variance of a given population will not necessarily decrease with a loss of deleterious mutations, and traits that influence genetic variety are found within particular groups, so there's no worry about the loss of unibrows any time soon.
Additionally, you seem to be assuming that every trait will be optimized and selected for, even though we don't find this in practice. Genetic variation is hugely influenced by selective choice, and there is some evidence to suggest that this is genetically-determined and therefore resilient. The fact that you or I might want a tall, mathy, conventionally handsome, bisexual god for a partner doesn't mean that someone else might not have a strong preference for a medium-height, artistic, conventionally handsome (this probably won't change), heterosexual. So long as choice remains free, beneficial variation will persist.
Hi Scott, thanks for the thoughtful post. I was surprised to find myself very convinced, but then reflecting more on my un-ease, I think there is an ethical dilemma that haunts me, and I was curious your thoughts.
I'll assume that in a couple decades we'll have much stronger predictive power than today on a wider range of traits, in order to dispel objections around the efficacy or current menu of options.
Ethical Objection: Arms Race
Diving into "Ethical Objection: Cost," an underlying assumption seems to be that people's nominal choice is indeed a true and freely made choice. If we give people the means to choose, then they can choose freely whether or not to choose IQ of their embryo. But, given how competition is fundamental to human life, if your child's peers are getting ahead with genetic enhancements, then do you really have a choice? This looks a lot to me like an unavoidabe start of an arms race? Those who choose not to participate will be dooming their progeny to failure competing in society. To exacerbate the issue, I would imagine that the positive traits can compound, generation over generation.
I am sold that the category of traits control mere preferences (say, eye color) seem harmless, and traits that fight disease are good. Consider the category of "enhancements" that will make your child more competitive in society (whatever societal context you participate in). Does this lead to an arms race? If so, is that a good/bad/neutral thing?
Karl Marx argues that the surplus labor should be socialized back to leisure. This did not come to be, and China is one of the most competitive societies in the world now.
Deep thinkers from every end of the spectrum of economic theory have predicted that we'll have reduced competition, and more leisure for generations.
I'm starting to think it ain't gonna happen.
Evolutionary biology seems to have woven the need to compete and progress and race into our fibers.
By most accounts, hunter-gatherer societies had a lot of leisure time. We already are plenty productive, that's not what's stopping us from achieving this!
You could imagine some government presiding over a single country and limiting leisure, but as long as some people want a leg-up, or to get ahead, or to push their social status, or to indulge their ego, there will be competition, domestically. Internationally gets even crazier -- Without a credible threat of force from a governing body, no society would feel comfortable "being lazy" while another potentially maniacal society (bloodthirsty dictator, democracy with ethnonationalist ambitions, etc) can out-compete and threaten you.
> If this were a single-use medical treatment, delivered by a doctor after someone got the relevant condition, it would be one of the biggest advances of the decade - imagine a drug that cures 10 - 40% of breast cancers with no side effects! But in fact, it works for breast cancer, and schizophrenia, and heart attacks, and approximately everything else. The only things comparable are antibiotics and GLP-1RAs.
Vaccines too, aren't they?
Also, not to minimize the benefit of embryo selection if it works, but one difference from a drug that cures 10 - 40% of breast cancers (and other diseases...) is that it's like a drug that has to be given to the entire population, including those who don't have an won't go on to have breast cancer, in order to cure 10–40% of those who do get it, so this increases the cost (and, if you wouldn't otherwise use IVF, discomfort) per case cured — kind of like vaccines. Actually this may also be a reason there's more controversy around vaccines than other drugs: they are given to healthy people, whom it's not yet known if the vaccine will benefit, so people tolerate less risk of side-effects than someone who already has a disease and wants to get rid of it.
----
((EDIT: already mentioned by other commenters:) One thing I didn't see mentioned: last time I've read about it, IVF itself had some moderate risks to the embryo; what's the situation with that now, and how does it compare to the benefits of embryo selection? This is relevant to those who consider IVF only in order to do embryo selection.)
Another minor point: Do the companies pay attention to avoiding inadvertent sex selection? If, say, breast cancer is more common than prostate cancer, or if they are worse at selecting against it, that could accidentally lead to the embryo with the lowest risk of the diseases they select against disproportionately being male. Widespread sex selection could cause problems (some people of the more common sex won't find a partner); here in the West I don't think most people have a strong sex preference about their children, but if we don't consider and avoid it, embryo selection could accidentally lead to sex selection even if nobody wants it.
> It depends on the number of embryos they have to work with; the more embryos, the better you can do by selecting the best.
You can do even better by selecting gametes from individuals who have the traits desired. If if you want to have a kid who is good at tennis then get sperm and egg donations from professional tennis players.
In general, reading a phenotypic effect off an actual phenotype is always going to be easier and more accurate than trying to do it from genes.
I'm sure some people will say "but I want my children to pass on my genes"; and fine, those people who want that can do it. But the government should also facilitate prospective parents who want make use of the best genomes out there.
> It’s probably even obvious for governments to pay for all citizens to get these, given how much they’d save on health care costs.
IMO it's a no-brainer for governments to do this, and offer genetic counselling including sperm/egg donation, for those people who want that. Governments would not only save on health care costs, they'd also benefit from a more intelligent and conscientious population with people less likely to commit crimes.
Yes, this technology is most helpful for people who want to have biological children with their life partner, and/or can't get the greatest genius in the world to donate a gamete to them. But I think that's 99% of people.
> and/or can't get the greatest genius in the world to donate a gamete to them
I envisage government-run gamete donor banks where donors are selected for things like: high IQ, high conscientiousness, high achievement, in good general health, not having committed serious crimes, being physically attractive, etc.
Because it's a lot easier to get sperm out of a man than eggs out of a woman, that would probably be what was mostly done.
> But I think that's 99% of people.
I agree with you that it's most people, but I would guess it's more like 70-80% than 99%. My reasons for this are:
(1) If I ever had kids I would want to go this route, as I have been held back all my life by poor mental health, and I wouldn't want the same to happen to my children.
(2) Most parents want their kids to have happy successful lives and if this helps, many would go for it.
(3) Never in my life has any parent bragged to me about how ugly or stupid their child is, but many times have they done the opposite.
(4) As well as optimising for being clever, this process can also optimise for children who're good looking or have pleasant personalities. I feel this last one might be important to many parents considering they will be in close proximity to their kid for the next 18 or so years.
(5) Adoption is a thing in many cultures. In some cultures you could even adopt people after you were dead!
Obviously if 20-30% of kids were created by this route and so had high genetic quality, that would improve things in any country.
This is fascinating and IN PRINCIPLE sounds like great things happening fast, but I have an incredibly strong feeling that the whole text seems to devote much more attention to comparatively lesser concerns ("we don't have this as good for non white people yet", "what if I still get x cancer even if I had a 5 vs 10% chance) and only fairly briefly cover the huge, glaring, elephant in the room: implementing this on anything even remotely societal scale (even within the top 25%) would completely, possibly irrevocably change how one of the key human life-form / species - relevant functions is carried out. That this is "IVF only" option seems to be huge. Unless you have a massive genetic burden (and even then testing and aborting if detected seems better), this seems a real hurdle?
On the other hand, expanding to individual pregnancies would massively up abortion rate in women "trying for optimal" or at least "trying for good enough", which would create its own issues (and a big outcry in groups already opposed to abortion or on the fence regarding it).
> Selecting the best of five or ten embryos is not a very effective way to get the genes you want
I appreciate that Scott mentions this primarily because it is quite common in popular culture to go around claiming that PGS embryo screening will give people “superbabies” and this claim is obviously a marketing gimmick.
The within-family validation that Herasight has done is great and shows that their screening can help parents increase the chance that their kids will lead healthier lives. I think that’s great! But we shouldn’t oversell the technology because that just breeds mistrust and resentment.
If you've already decided to do IVF, I think the rest is footnotes. One has to manually select somehow in any case. (I disregard scare talk using thought-stop words like "eugenics." If someone has a particular objection, I'm happy to consider it.)
For my part, I strongly favor making babies the natural way.
Arguments about the genetic components of heritability (or lack thereof) often seem sort of unresolvable with existing techniques since there is no independent variable to manipulate when you're dealing with humans. e.g. If you hypothesize an allele causes black fur on an animal, you can manipulate it (by breeding or CRISPRing) & see if you're right : a proper attempt at falsification. In humans, there's no manipulation so you have to all kinds of correlative stuff, which kind of can't make causal inference. (Theoretical considerations aside, not being able to directly test stuff seems to make progress way, way slower.)
I have my reservations about current PRS, but I am hopeful the application of these sorts of technologies could produce exactly the sort of data that could resolve some of these arguments & increase predictive power: by actually *testing* trait predictions, not on a held-out set (which could have the same confounders as the training data) but with a real experiment (selecting or CRISPRing an embryo). The resulting data could have long-term benefits. Even in the unrealistic extreme case where everything is population stratification and selection on PRS didn't work at all, that would suck for the people who paid for the service but would be a very valuable result scientifically & medically.
(Of course this assumes that there is enough high quality follow-up work measuring the traits people selected in this way, and enough transparency on the part of the firms offering this service about their predictive algorithms. Also practical considerations mean that this won't exactly be a controlled trial & there will probably be dramatic selection effects based on the sort of people who pay for it. But still better than just running regressions on biobank data.)
Modifying genes was impossible for most time and still very restricted, but many people tried creating environments to create geniuses and barely found anything; most educational interventions produce IQ gains in children that fade in adulthood.
"Confounders" here are not environment variables (like nitrogen in soil for plants); if anti-hereditarians determine that a trait is "confounded" they don't learn how to modify environment to change a trait, but they actually find that people like to self-segregate and the rich will pay money so their child wouldn't attend same school as children of poor. In some sense, when people can migrate where they want, unlike North Korea, is very good!
Some of "confounding" might be womb environment rather than "school quality", did they test that?
But hey, we could just clone a few geniuses instead of trying to analyze what of 6 billion bp makes them, and who is against that the most?
All the companies you've mentioned are going off of public bio-bank data from what I can tell. Is there a reason they haven't tried to prove out their technology by raising some genetically predicted super mice, and provide empirical evidence (through enough overly healthy and smart mice) that this technology works?
I absolutely approve of reducing the risk of disease. It also seems profoundly stupid to spend lots of money on determining hair color when hair dye is a thing. Most people don't even know Christina Hendricks isn't a natural redhead.
>Mark my words, one day there will be entire subfields of therapy dedicated to these issues.
I feel like this is the core of the issue, really. When people say "I am more than my genes," they're probably not saying "I deny that statistical correlations between genes and outcomes exist," they're saying "I do not want my kids to grow up being told they're Genetically Superior/Inferior." Which is a correct thing to say!
> Somebody in Macaronesia in 3525 AD will avoid getting breast cancer because of you (if there is still cancer; if there are still breasts).
Big breasts are a burden (literally), genetically engineer flat-chested women and men who are not attracted to breasts.
I wanted to say that you can go even further, do parallel gender selection away from the accidents of evolutionary attraction toward more functional bodies that we would consider ugly now, but... in reality the fertility crisis will likely result in artificial wombs/vat babies/cloning much sooner than 3525, decoupling procreation from attraction and eventually making sexual attraction a weird rudiment that the enlightened aces (and eventually everyone) would sneer at.
The Selection objection I find strange. If you had been selected for pretty eyes regardless of cancer risk, you still won the billion:one lottery! The remarkable odds against any of us existing at all makes this a silly objection. If they had made a "better" choice regarding cancer, you wouldn't exist!
Yes, and I think people will also complain about their parents selecting them for eyes rather than cancer risk (while still not wishing they didn't exist).
I agree there's a philosophical sense where if their parents had done anything differently, they wouldn't exist, but I think it's a very weak sense. See https://www.astralcodexten.com/p/who-does-polygenic-selection-help . Or compare to the example of someone whose grandparents (before they were born) left their parents money for their college education, but their parents gambled it away and lost everything. Now, there is some sense in which if the parents hadn't gone to the casino, it would have jostled the sperm in their testicles slightly differently, and a different child would have eventually been born instead of them. But I think it would still be reasonable to resent your parents for this, even though in some weak sense it caused you to exist.
Sure, but I think there's a tangible difference in that in the selection case the parents made a conscious choice of me vs another, and presumably I know that. In the gambled away example it *feels* more random, disconnecting the choice from the existence in one's mind. Even though it could be argued that the end result is the same, complaining in the first case seems more unreasonable and less defendable.
Don't IVF babies have modestly higher baseline rates of birth defects and other health problems?
If you already need to do IVF, sure, adding genetic selection largely makes sense. But if you don't need IVF, it would be important to consider not just the health downsides for the mother of doing IVF but also the health tradeoffs for the baby, right?
> Although GP would continue to only test for health outcomes, you could forward the raw data from GP to Nucleus, and Nucleus would predict extra traits, including height, BMI, eye color, hair color, ADHD, IQ, and even handedness.
It does seem important to ensure your kids have an even number of hands.
"Suppose you invest the $3,250 in an index fund that makes 7% over inflation, then give it to your future child when they turn 45 (average age of type 2 diabetes diagnoses). Now it’s worth $75,000. Is this the “true” cost of the intervention?". no, $75k in 46 years is less than the true cost, which is $3250 today. it is less because $3250 can buy $75k in 46 years or it can buy other things which are likely more valuable. we know they're likely to be more valuable because people use their marginal $3250s to buy them and not $75k in 46 years
of course sometimes people do trade $3250 today for $75k in 46 years, usually when saving for retirement. so sometimes the two are the same. but your post seemed to imply that this was not the next-best use of the money
So much rigor here and then "AGI is going to cure all disease and I know that with overwhelming confidence despite the entire history of human folly regarding such predictions"
Eventually we are going to automate scientific research through artificial intelligence. Not clear when. Soon after that happens, I think it’s reasonable to assume that most diseases will be cured, just as we might assume that most diseases will be cured with enough scientific study
I'm less sanguine about what will happen to medicine, but I _do_ expect that, barring an antarctic level AI winter, AI will radically change the whole game.
I think the main worry I have about this technology, (and about selective reproduction programs like ones attempted by Musk or things like The Repository for Germinal Choice) is some variation of "what gets measured gets managed".
That is, we are going to use this technology to optimize things that we can measure. These things that we can measure, like IQ, are imperfect heuristics for things that we actually care about (intelligence or mental capacity or academic potential I guess?). And even those are simpler than the things that we actually-actually care about which are at the level of our society and culture - overall standard of living, kindness, freedom, etc...
So if we gain the power to optimize for things, and we start optimizing for the heuristics that we have available to us, like IQ, we may face an alignment problem. Maybe the things that we can measure right now (like our current conception of IQ), or the things that are in vogue right now (height, particular shades of eye color), are misaligned with the things that we might ultimately care about.
So maybe IQ is negatively correlated with some other important set of traits that are murky and difficult to define and we can't measure very well (say, creativity or being an empathetic human being, or "research taste" or etc...). Just because these things are hard to define and difficult to measure doesn't mean they're any less important.
Furthermore, traits that may be beneficial for individuals may not be beneficial for society. So maybe an individual does better with higher IQ within our current society, but a society where a significant number of individual have a higher IQ undergoes a phase transition which is worse in some way. Maybe it's an overall higher level of cynicism, or a breakdown in the useful fictions that create some form of social cohesion and makes economic cooperation possible. Maybe it's a loss of overall diversity of the population that makes the overall population more fragile.
I guess my overall feeling on this is that we are so primitive in our understanding about these things, and our power to make changes to our genome may advance much faster than our understanding about the societal consequences of those changes. And part of me thinks "but I suppose that's just science in general", and part of me thinks "no, that's just technological determinism!".
This post made me think a lot about the silo series by Hugh Howey. Is humankind doomed because the advances in our power to change the world and ourselves outpaces the wisdom required to wield it without destroying ourselves?
This sounds like the antagonistic pleiotropy suggestion, but see my response above. In practice most good things are correlated with one another.
I am also very worried about our power to change the world outpacing our wisdom, which is why I want to intelligence-enhance children as soon as possible! I agree intelligence isn't exactly the same as wisdom, but it's probably pretty correlated. If it isn't, feel free to submit an ACX Grants proposal to develop a genetic predictor for wisdom!
It seems to me the most important ethical consideration is the one you treat the most glibly, that parents will be making a specific choice about what their kids should be like, that their kids DNA will effectively become a message vector in our society. This isn't just a matter for therapy, this means inviting all the most irrational cultural pathologies into our genetic code, and basically making future generations subject to a certain biological totalitarianism by their parents. It seems to me an exceptional important ethical consideration to standardize any selection process so that it ceases to be a potential message vector.
Stuff like this: "How much would you pay to give your children an extra 1-4 years of healthy life?" and framing things in terms of how much better you're making your child's life is misleading. You're not making anyone live longer, you're bringing into being someone who lives a longer life over someone who lives a shorter life. I might pay a lot to make my actual dog live four years longer, but that reason doesn't automatically translate to a reason of the same strength to buy or breed a chihuaha rather than a great dane, even though in the same sense I am "giving my dog a longer life".
If you're a repugnant-conclusion bullet-biting totalist utilitarian then this difference doesn't matter but for most other plausible moral perspectives (especially those on which your reasons regarding your children are stronger than those regarding strangers) it does. That's not to say you don't have any reason to bring about a longer-lived child, but it's very plausibly not a reason of nearly the same strength as your reason to cure your child of a disease which would end their life early.
It isn't going to end with just wanting to make sure your kid doesn't have diabetes. Genetic disease is the bogus gateway drug. This is a frankensteinian future with a million unintended consequences. No one is going to think about, or care about those unintended consequences until it is too late.
This isn't a spinning jenny moment. This is playing God. Historically, philosophically, morally - that never ends well.
The problem with the "unintended consequences" argument is it's far too broad. When people invented agriculture, writing, metalworking, or movable type, all those had unintended consequences.
If we ban everything that might have unintended consequences, we ban all progress.
The argument about Dude, Genetic determinism has a fine side. Basically, not that this is a reason that we should not do this. But basically, if you really think about the stength of these predictions, they are not very strong. After use for selection, there are very few situations where the information about future traits would be useful compared to not knowing (mostly test order for diagnosis). If these techniques get broader, they are still weak evidence about what someone will become. These companies do not publish the remaining variance from the expected values published (especially for IQ and healthspans), and they should. If you really internalize those, doing that behavior is obviously dumb. The variance remaining (either from what a 3 percent chance means, or the remaining lifespan variance, or the IQ variance not explained), is often high enough that bets against the low probability even need to make sense, and having strong expectations about how children will end up are in general, low to negative EV even if your are expectations are right, and even decisions based on that expectation with bad payoffs in the low probabilty case should be avoided, for obvious reasons. That is, the fact that the obvious case is so obvious is tripping up the complainers.
I don't expect the average ACX reader to be persuaded by my thoughts on the immorality of IVF in general, let alone embryo selection. Those arguments are out there if you want to look for them.
But perhaps you feel a bit of disgust at the thought of people picking out children like they'd pick out puppies from a breeder. If you can have a labradoodle, why not a blue-eyed blond baby with ten more IQ points and a 32% lower risk of cancer?
I also think this post glosses over the cost/inequality aspect of it. There are billions of people on earth who are never going to be able to afford this. Their descendants in the next few generations won't be able to, either (barring some weird post-scarcity techno-utopia AI-world, which I'm not counting on). What does the world look like in 2125 after a few generations of elites selecting embryos (with the technology presumably improving along the way) while 4 billion or so Africans can't or won't? As embryo selection gets better, people will be picking for everything you can imagine: looks, athletic ability, artistic talent, etc. You'd be on the way to having two totally different human species. Doesn't this sound like a dystopian sci-fi novel to y'all?
What's the state of polygenetic prediction for Big 5 personality traits? Supposing you think AI will make IQ and "boring" illnesses like cancer, T2D, and heart disease irrelevant. You might reasonably conclude that you should focus your efforts on having kids who are kind, thoughtful, creative, and emotionally stable (or at least not unhinged assholes). Would love to know how many SDs you could move on various Big 5 traits as a function of number of embryos chosen from.
Has anyone written at any length about the potential for personality selection?
Seems like there are 4 big categories of traits ordered by controversy:
1. Clearly negative health based traits (genetic diseases, propensity for high cholesterol, etc)
2. Clearly positive traits, not clearly health-related (IQ, balance, etc)
3. Preference based aesthetic traits (eye color, height)
4. Personality traits (neuroticism, openness to experience, etc)
Of course, many traits will blend between multiple categories. IQ is correlated with mortality, tons of psychiatric conditions are associated with big 5, etc. By and large, though, I’d guess that there are more who would object to eye color than to IQ and more to IQ than to cancer risk.
Personality traits seem to be objectionable on entirely new criteria than the rest. It’s one thing to charitably select for lower incidence of cancer in your kids and a whole other thing to select for kids that are likely to have a charitable attitude or a certain belief structure. This seems inevitable, though. If we can select against depression we will eventually be able to select for gregariousness, for ambition, for any number of other traits. There’s the question of if this is ethical to do at all and then the secondary question of, if it is, is it ethical to select for a personality for your kids that is, for example, less likely to be happy but more likely to be successful?
"Most genetic data available for research/commercial use comes from the UK, US, and Europe - areas which are mostly white." All three had been working on this problem by making their populations less white, although the US has shifted on that in the past six months.
I think people in the rationalism community should definitely not select for higher IQ fetuses, because they likely already carry risk factors for autism. In another population it'd be fine; in this one it's riskier.
Some guy with a similarly sounding blog wrote a whole post about this! But to sum up, *actual* autism is negatively correlated, but SNPs for IQ are positively correlated.
So if you have two parents that are on the BAP - broader autism phenotype - they already have a much higher risk for having an autistic child. If you do embryo selection on IQ you're also much more likely to select the autistic child because there's overlap between IQ SNPs and autism SNPs; some of these are the same SNPs.
I'm glad you have addressed this concern, but you have not really internalized it properly.
> Ask anybody dealing with regular insulin injections (let alone limb amputations) whether it would be worth $27,000 to wave a magic wand and not have Type 2 diabetes! It’s not a hard question!
The proper question is, whether it would be worth $27,000 to wave a magic wand and be replaced by your sibling who doesn't have Type 2 diabetes. I agree this is not a hard question.
Also, while I follow and agree with all your points intellectually, I just can't share your seeming enthusiasm. "romantic and aesthetic"? Yeah, nothing says romantic like selecting an embryo based on the polygenic score calculations.
I read recently about multi-PGS where they take PGS for seemingly unrelated traits and make composite predictor (which might be non-linear) with better score. But I found very little about it.
Eugenics carries a deeply negative legacy, for reasons any student of history will understand. That said, it’s worth distinguishing modern genetic screening from its past abuses.
Preventing heritable diseases is widely considered a positive goal. Reducing the burden of serious medical conditions can improve individual lives and contribute to public health. Using technology to improve the overall condition of the human species is a valid long-term goal.
Moreover, I submit that the defining feature of our species is intelligence. In today’s society, lower cognitive functioning can be a barrier to employment and independence, while higher cognitive ability is linked to greater productivity. If reproductive technology allows parents to select for intelligence, does it really matter if they, at the same time, choose their children's eye color?
You go on to mention race. While obvious characteristics like skin color are irrelevant, science has repeatedly observed group-level differences in cognitive outcomes across populations. These differences correlate strongly with the prosperity of the respective societies. At the same time, fertility tends to decline as societies become more prosperous - something noticed even in antiquity. This is, effectively, a brake on human progress. Whether this can, in the long-term, be addressed by genetic screening?
If a company is selling a product that costs tens of thousands, but there's not really any way to verify the product worked in practice without doing a study across large numbers of the product's customers over 80 years... then I'd have a very high prior that it doesn't work as advertised. The potential for fraud or mistakes is too high.
If it doesn't work for whatever reason, by the time this can be proven to a legal standard (particularly the parts relating to the longevity of the selected children, and their disease frequency in later life), the customers and the founders will be dead, and the company may no longer exist for your grandchildren to sue.
Or, to put it another way, if I wanted an unethical way to get rich, I could sell expensive products whose efficacy couldn't be disproven until after my death.
I think this is a concern, but we already have a culturally normal practice for something similar: the construction business. We regularly contract companies to build houses and bridges and so on intended to last for 100+ years. I think many of the same ways we assess the reliability of construction can be applied to embryo selection, albeit with a lot more uncertainty at this stage.
Because the selection involves a variety of traits that become evident in different years after the purchase, we can assess how reliable the company is to some extent long before we get to the 80 years out threshold: firstly you can genotype the baby to see if the claimed genes are actually what's in the kid. Secondly, there are traits like eye color that show up within a few years, traits like height that show up within a few decades etc.
In that time, we also get more data on the genetics of cancer and other long-term risks in the population in general, and we can compare the results of the embryo selected vs the baseline.
That really doesn't seem to apply to the case where a store sells you an item with a promise that it will or won't do a thing 40 years later. Or that it's 10% less likely to do a thing 40 years later. How much class action do you need for a bunch of customers to get together 40 years later and say "hey, this thing you sold me wasn't actually any more likely to do the thing than the version we could have got for free!"
If I go into a store to buy a stand mixer, and they promise to sell me a stand mixer, and then wheel out a concrete mixer - well yeah technically it's a mixer but come off it, mate.
At the moment, I think the technology is at the level of "we can tell you if the embryo has Down's syndrome or CF" and they're promising (but unlikely to deliver on) "if millions of people select on these criteria we supply, then in twenty years time out of the hundreds of thousands of babies born, there may well be an overall 1 point increase in population IQ" but leaving the clients to parse that as "my baby might be 6 points smarter!"
You and your wife are reasonably smart people, she's going to take care of herself during pregnancy, and you as parents are not going to fold, spindle or mutilate any kids you have. So there's a good chance any future babies will turn out okay. Will that be due to the genetic selection? God alone knows! But I'm fairly sure the company does not.
They're selling hope in a bucket and badmouthing their competitors in the effort to grab market share, market advantage, and those sweet, sweet, aspirant dollars before they go as bust as 70s cryonics companies.
I'm guessing (I haven't read any PGS company's offers) that they are promising to do only what they can do - to offer customers the choice of embryos to implant based on the polygenic risk scores they have scored the embryos for. The actual outcome will be indistinguishable from chance on an individual basis, as you say. It's only in retrospect, across hundreds of thousands of people, that any impact will be noticed.
I was trying to think of an analogy, and I think it's like an impressionist painting. Up close, every splash of colour is just a splash of colour. You have to take a good step back to appreciate the image. Up close, each individual is just that, and, except for some Mendelian single-variant conditions, it's impossible to pin down a talent, ability, condition or illness to any one cause. It's only from a distance of hundreds of thousands of people that you can see a pattern.
EDIT: And for that reason, I rather think that what these companies are really selling is a chance for parents to believe they have chosen the best they can for their child.
1. This allow parents to select sex of a child as well, which could move us away from natural balance in which for every 100 girls 105 boys are born.
2. Selecting for obvious traits like risk for certain diseases, IQ, etc, might also unintentionally select for some and against some other traits that weren't even considered. This too might skew the natural frequency of some traits... making some more prevalent than is optimal, and some less prevalent than is optimal.
3. This opens the can of worms which will allow kids to blame their parents for pretty much any negative outcome in their lives - whatever they did: if they used this technology and kid turns out to be unhappy or to have some disease, they can blame parents for wrong choice; if they didn't use this technology and conceived naturally, kids can blame parents for NOT using this.
4. In general kids lose some of their dignity points by not just coming into their parent's life, but becoming in some ways their explicitly engineered product. Imagine a situation in which a kid is selected for high IQ, and then is pushed by ambitious parents to become some wunderkind or to excel academically, while the kid itself turns out to be completely uninterested in such a thing and would prefer to be a soccer player or a plumber. Imagine the kind of arguments that could ensue. "Did I in vain select you to be so smart? Now you want to waste all that potential?!
- F**k you and your embryo selection! I'm my own person!"
5. If eventually personality traits become a part of the equation, so we can select for them as well, there is a risk that humanity could become much more homogeneous, we could become a society in which popular traits are very frequent, even if "popular" doesn't necessarily mean optimal, especially not on a level of whole population. (For example, for a single person, it might be optimal to have some "good" personality, but for the entire population it might be more optimal if not all people have such "good" personality, and we also have a fair share of quirky people, neurotics, introverts, etc... For example high neuroticism might be required for taking seriously X-risks)
6. Achieving genetic perfection could put more pressure on people to perform, because now they have no excuse. If you fail in life, it's only you to blame.
7. In general it interferes with natural selection and introduces elements of artificial selection. A thing that we do to dogs, for example. Many of the breeds created by this process actually have worse health and fitness than mongrels.
8. There is a open question about how many embryos to select from. It could be 5, 10, 50, 100... or more? Each choice could be criticized.
1 is already an issue. Parents using IVF in the US can select by gender, and apparently it's socially acceptable in educated circles to do so, as long as you are selecting for a girl. In the EU, it's illegal to select by gender (though I'm not sure of how this can be tightly enforced).
Gender-selective abortion bans came about in the EU because, surprise surprise, turns out parents (particularly those of certain ethnicities whose culture place much more value on sons) wanted boys and were aborting girls, even if that was already illegal since they could find a compliant doctor if necessary, although it is a complex question - some studies find a skewed ratio in live births, others don't, and there is a natural ration of 105 male to 100 female live births anyhow.
Not sure if you, or the companies, have answered this elsewhere, but I can think of two more issues. You've discussed the relative risk of developing the condition, but not the absolute risk, nor the severity of the condition. Some of the conditions tested are exceeding rare, and so even a 90% reduction might not mean much. And to pick breast cancer survival for example, while apparently the risk is rather high for the average woman (about 14%), the outcomes if not diagnosed late are actually pretty good. And you don't know the severity of those who will get the disorder. Maybe the ~7% who do get breast cancer will have a kind that is more aggressive or treatment resistant, which means the the relative risk is misleading and this process is actually not worth it.
The other issue is that in removing these genetics from the gene pool, you're removing beneficial traits that are harder to identify. Lets say that people with schizophrenia genetics are more... adept at being altruistic or something like that. You could develop a society that's perfectly healthy but but there's a statistical predisposition to selfishness that changes the whole culture.
About action when the future is unknown: 500 years ago, when Martin Luther was asked what he would do if the world were to end tomorrow, he answered, “I would plant an apple tree today." That story got rephrased into "Even if I knew that tomorrow the world would go to pieces, I would still plant my apple tree."
3 fun facts: 1. there is an eco-bestseller in Germany titled: "Let us plant an apple-tree! It is about time." 2. The original quote might be Islamic: “If the Final Day comes upon one of you while he's in the middle of planting a sapling, let him finish planting it.” -Prophet Muhammad ﷺ
"suppose that I had a natural, unselected child, and that child became high school valedictorian and got into Harvard. Would my first reaction be “Oh no! This slightly raises her risk of anorexia!”? If not, why should this be our reaction to artificially increasing IQ?"
With natural, unselected children, you don't probably think about their traits (might be using "trait" wrong here) in terms of tradeoffs, because you never opted for anything in the first place. Like I don't actively think about stuff like "Gawd, I wish I was born with worse eyesight so I had more voluminous hair" or such correlations there might be between different traits, because they're not commonsensical and, more importantly, you don't opt for them.
There's no sense of responsibility or regret over traits we didn’t select, while selecting involves intentionality and responsibility.
I think, though, that the question is fair to present just as "If I could be smarter but have anorexia, would I still want to be smarter?" assuming the elevated risk of anorexia matches the risk of being smarter.
I would probably select for non-anorexia, but my guess is that this community would tend to prefer the IQ? Does this point out a flaw in the selection process? I freely admit I would probably be bad at making these kinds of decisions for my baby.
Assuming the selection only yields potential positive effects and no potential negative effects, it's of course a no-brainer that this would be good for everyone with any selection. Assuming there are tradeoffs, it raises legitimate ethical concerns about whose values guide selection and whether those values are always in the best interest of the child.
Also, let's assume the tradeoffs aren't that clear. Having lesser risk to heart disease has a slight potential risk of heightened aggressiveness? I assume this community would baulk at aggressiveness as an undesirable trait. But assume other communities you can probably name, where aggressiveness is desirable, where people think that it's a dog-eat-dog world and being big and intimidating and a bit of a rabid dog gets you places and means you can protect your family better and so on.
I guess I am essentially making the slippery slope argument and highlighting the vast differences between cultural preferences for desirable and undesirable traits.
I think if you asked parents of anorexics if they would prefer their child were a little dumber but healthy, they'd pick that.
"Smarter, but a risk of anorexia" sounds great when it's a theoretical discussion. Who wouldn't want their child to have the advantage of being that bit smarter than their peers? And my kid will be too smart to fall into the trap of self-starvation! When it's your teenage child starving herself down to skin and bones, you may wish you hadn't ticked that box in the clinic, because intelligence has little sway over compulsion:
"Anorexia can lead to severe health problems associated with malnutrition (not eating enough healthy food). But these will usually start to improve once your eating habits return to normal.
Possible complications include:
- problems with muscles and bones - feeling tired and weak, osteoporosis or physical development
- fertility problems
- loss of sex drive
- problems with the heart and blood vessels
- problems with the brain and nerves - such as fits, difficulties with concentration and memory
- kidney or bowel problems
- having a weakened immune system or anaemia
Anorexia can also put your life at risk. It's one of the leading causes of deaths related to mental health problems. Deaths from anorexia may be due to physical complications or suicide."
I'm extremely sceptical, not to mention cynical, about promises of earth-shattering benefit to all humankind. So this makes me murmur "Oh, really?" very strongly.
First and foremost, this whole industry is a business. Yeah, yeah, healthy babies is the selling point. That's not what the share price is about, or even the business (else a company wouldn't be promising 'pick hair colour and eye colour').
So one company takes a swingeing criticism brickbat to its competitors? Well, colour me surprised that a business is telling potential customers "ignore those fake lying snake oil purveyors, spend that 30 grand on *us*, the only true genuine real deal snake oil!"
Second, as pointed out, we'll only know how all this works once once many genetically selected babies are born and are observed as they grow up. Did you get that 6 point IQ increase after all, or would your kid have been smart anyways? Oh, Junior *is* smarter, but they're also troubled?
What we're forgetting is that we're playing dice with real beings, the babies who will be born and grow up after this process. I'm happy if they aren't going to get cancer or whatever, but I remain dubious.
(As an aside Type II diabetes doesn't necessarily need insulin, other medications and lifestyle changes are tried first and it's only if they're not working that you're put on insulin).
Thirdly, I think the potential for more family conflict this opens up is horrendous. It's bad enough when parents compare you to other family members/classmates/the neighbour's kid, but now you'll be competing in an unwinnable contest against an unbeatable rival: the ghost of the ideal you that your parents planned out when they selected for and against traits.
You're not *quite* as smart as you should have been (that 6 point IQ gain was in a range, in the small print, and you only got the 3 point gain). Not quite as tall. Not quite as attractive (your hair isn't the exact shade of blond/blonde, your eyes not that bright enough blue). Not quite as athletic, or academic, or musically/artistically talented as the brochure promised.
You can pick up that your parents are quietly disappointed that you're not as good as you could have been, even if they hide it (and as for the parents who don't bother hiding it? yeah that's gonna be a fun childhood).
And fourthly, I damn well could see people selecting for anorexia. Not the full "starve yourself to death" kind, no, but a stronger inclination towards it? A kid who is constantly dieting, exercising, calorie-counting, and reliably only ever eats exactly six peas, two carrot batons, and a square of tofu for every meal - well, they're going to hit that BMI target as promised by the company, aren't they? No dangers of obesity here, no disgusting bucket of lard boy or girl to shame you in public for raising a fat kid and being unworthy parents! Couple that with the tallness genes, and if you get a tall, skinny, weight-obsessed kid, there's a good chance of a modelling career which will more than repay you for that extra expense in gene-selecting! And if a little IQ bump comes along too, that's just gravy.
Good luck to those that choose this road of embryonic selection, I know you all mean well (in this generation and context, at least). All ye holy men and women, pray for us.
I share your concerns and I'll add another -- the dynamic you describe could be reversed, and parents may end up blaming themselves if their kids don't realize the advertised benefits. Channeling that disappointment inward, on the decision maker, is less bad than channeling it outward, but it's still bad and can have spillover effects on the children.
Generally a good comment. Yes, I agree with the hazards you cite. One quibble:
>What we're forgetting is that we're playing dice with real beings, the babies who will be born and grow up after this process.
Embryo selection is _reducing_ the role of chance in reproducing. The dice are already there, in the natural process. Selection makes it _less_ of a dice roll, not more.
But we are still rolling the dice, only this time we've loaded the dice without being sure which side is going to come up. We're hoping for sixes, we might in fact be setting ourselves up for snake eyes.
Exactly agreed. On each individual roll, embryo selection is reducing the fraction of the variation due to chance.
>without being sure which side is going to come up. We're hoping for sixes, we might in fact be setting ourselves up for snake eyes.
Yes, you are right that we are introducing a systemic risk that the selection may have unknown deleterious effects. I think _most_ of that is covered by Scott's section on correlations. E.g. generally speaking, higher IQ correlates with a bunch of _better_ outcomes in general health.
Admittedly, we _never_ know the long term effects of any innovation when we introduce it. If it had turned out exposure to the 120 Hz flicker in fluorescent lights (unprecedented in our evolution) mades people go stark staring bonkers after exactly 30 years of exposure, we would have turned every office worker into a maniac a generation after those tubes started flickering. If we wish to innovate _at all_, we have to accept that kind of risk. Mostly, we monitor short term side effects, and, mostly, that works well enough.
A moral premise implicit in trait-based embryo selection is that a reduction of difficulty leads to better lives: that traits associated with struggle (lower predicted IQ, higher BMI risk, neurodivergence, etc.) are inherently undesirable, and that fewer of them means progress.
But not all difficulties are defects. Some challenges are formative. Some constraints shape the trajectory of a person’s development in ways that aren’t easily captured by metrics of “health” or “success.” Many people living with lifelong limitations -- not just physical, but cognitive, social, or emotional -- report that their identity, moral insight, and empathy were forged through precisely the conditions others might now screen out.
I don’t like suffering, and I don’t romanticize it. But I reject the idea that a desirable life is necessarily a frictionless one. Yet, that’s the vision this technology seems to be nudging us toward, without much pause.
One thing I find confusing about Scott's stance on eugenics is that he takes Chesterton fences and the idea of cultural evolution seriously in other contexts. His stance on central planning, at least for the pre-1970s era, seems to be that the theoretical case was strong but conservatives were right to resist for illegible reasons instilled by what made societies successful in the past.
Normal selective breeding has been known about since the Neolithic and could change a society's population genetics much faster than embryo selection, and having better genes would have been really advantageous for any society. But in 10,000 years no society has had a serious eugenics program. Isn't the fact we aren't already the descendants of Bronze age eugenicists who took over the world evidence against eugenics? If you believe in cultural evolution, it must really hate eugenics for some unknown reason.
It doesn't seem consistent with his caution for other kinds of social progress.
Eugenics has been suggested since our earliest philosophical writings (Plato and Socrates both recommend it in the ideal society). Beyond philosophical writings, the knowledge or ability to influence population-wide genes simply didn't exist. You can't control who has children, especially with more primitive social structures, and people were mostly having as many as they could to start with.
The closest example in ancient societies is probably the Spartans, which effectively practiced eugenics, but they also had a lot of other cultural baggage which made them more stable, and more effective in war, but ultimately lead them to have population issues and eventual collapse.
I don't think it's right at all to claim that normal selective breeding could change a populations genetics faster than embryo selection. When I've seen people model it, it requires massive changes in the demographic of who's having children and how many, only for tangible benefits to be realized after 4-5 generations at immense cost. I don't think there's a society on earth that has enough excess power and foresight to make it happen.
>Eugenics has been suggested since our earliest philosophical writings
Yeah, I think once you know about selectively breeding animals it quite an obvious idea, which makes it more surprising that it's not already the norm..
>I don't think there's a society on earth that has enough excess power and foresight to make it happen.
Possibly, but there have just been so many societies since 8k BC. Autocratic ones, God-king cults, ones with lots of different sexual norms, slave societies etc. None of them could get a eugenics program going?
>only for tangible benefits to be realized after 4-5 generations at immense cost.
I think this applies to all arguments from cultural evolution. There are also strong immediate pressures against eg. monogamy, private property, state subjugation etc. which cultural evolution proponents generally accept. And eugenics seems like it could potentially give a bigger competitive edge than those things.
I'm not personally that convinced by Chesterton fence/cultural evolution arguments, but if you do accept them it seems you should at least consider eugenics through that lens.
This is a chimerical issue, if only because of human selfishness. We appear to have no difficulty dealing with the same question in other respects. I know of several people whose parents squandered their inheritances, for example, and went on expensive trips or spent it all in other creative ways. They garner no sympathy from anyone if this is ever mentioned, even though it materially affects their lives and possibly the lives of their children. Why should we imagine that the new genetic selection will be treated any differently?
(I agree that helping people to be born healthy is important, and I don't understand the technology enough to see its potential faults, this is just a side note that I thought might be useful):
This seems to give parents an even greater amount of control over the bodies of their children than they currently have. This can do great good in the hands of parents who are altruistic toward their children, but having seen my share of dysfunctional families, and parents who are actively harming their children, I'm moved to wonder what this technology might do in the hands of the worst abusers. A controlling parent selecting for a disease that would make emancipation difficult — selecting for *lower* IQ, perhaps? Or a molester selecting for a child to look to their liking, the better to eat away their childhood?
Perhaps my own issues are showing, noticing this amount of control being given up and flagging that there's potential for even greater abuse in the hands of the minority of villains, but I wanted to bring up that this would also be one side of the question to keep in mind.
The eugenics performed by the blind idiot god already maims us sometimes, and so does the negligence and abuse in our childhood, but being able to just add some problem right into the DNA is an interesting new situation.
***
Now that I think about it, I wonder if opponents of homeschooling might be driven by something like this, knowing how horrible parents sometimes may be. I. e. someone like Scott and me might say, "the school is a horrible experience for many children, it's insane that our culture gives them no choice and doesn't listen to them when they say they're miserable and forces them to suffer for years on end", and at the same time, the opponent would be saying "being at home is a horrible experience for many children, it's insane to force them to be there with no alternative" etc. But I guess this is less applicable in a country with any kind of social services?
Go tell the dying person that the reason they can't be genetically enhanced is because you have "ethical concerns" about eugenics.
There is no inherent difference between enhancement and healthcare and that is fine.
People who have a problem with non-violent eugenics are motivated by envy and not some fear of the second coming of Hitler. It's really gross and disgusting to say to someone that they have to live with a disability because you have an ethical problem with enhancement.
Literally millions of people have died because of the ethical concerns of left-wing academics.
Everyone you know who has ever gotten cancer probably only got it because leftists have opposed genetic enhancement on the basis of arguing against eugenics.
I believe that a person should support a world where everyone is better off even if they themselves would not be better off in that world. If a person would oppose other people having greater health because of concerns over inequality or eugenics that person has a low and disgusting soul and is simply not worthy of any respect at all.
The only acceptable equality is the equality of spreading privilege around. To advocate for the deconstruction of other people's privilege is to advocate for a worse life for other people. That's disgusting logic, and the people who believe it are gross and disgusting people.
I can't respect them at all. I can't respect anyone who opposes making the world better on the basis of so-called fairness, equality, or equity.
People who oppose non-violent eugenics are the worst human beings on Earth. They are all literally mass murderers with the blood of millions on their hands. Everyone they know who ever got cancer or diabetes only died because of them.
The reason the dying person can't be genetically enhanced is because if they're an adult, it's too fucking late (to be coarse about it). The genes are already expressed and you have the deficient body, you can't just use an eraser and wipe out the little mistakes then. "Oh no, if only my parents had done embryo selection before I as born!" First, the tech didn't exist then and second, had they done so, the resulting baby would not have been the you that exists right now to complain about it.
Very different from picking an embryo on the grounds that "we think this one will have a lower risk". And of course the lower risk one if it survives implantation etc. will not be your dying person.
You're trying emotional manipulation here which is not going to work.
The time horizons involved in practically testing these claims are so long as to make verification impossible. Without verification, the field is going to be extremely vulnerable to motivated reasoning/wishful thinking/publication bias/etc. from the various companies offering this service.
The description of type 2 diabetes isn't true to my experience, but I have no idea whether I'm very unusual.
My blood sugar is well controlled with metformin, two pills per day. I test my blood glucose daily. I can eat fairly freely, though with some limits on carbs. It's been over five years, My eyes and feet haven't taken obvious damage.
I'd rather not have it, but the biggest downside is being dependent on the insurance/prescription system.
Unrelated, but anorexia does seem to be strongly correlated with OCD, bipolar, etc. Possibly the real solution is selection against mental problems.
On the other hand, selecting against anxiety might be out there in fantasy land at this point.
And select against anxiety: It turns out there are kids who don't eat-- sometimes so much anxiety about choking or other problems that they can't/won't swallow anything. https://archive.is/iZRcs
I think there's another risk you're not considering -- systemic risks and pathologies that come from lack of diversity.
There's a common trope in disaster movies where a subset of the population rides out the apocalypse in a bunker so they can rebuild civilization. The lucky few are inevitably selected for being really smart or skilled in something that seems important for rebuilding civilization, at least to the people in power (who often just so happen to have many of those highly valued traits).
That's always seemed like a recipe for disaster to me. By decreasing diversity (genetic, behavioral, cultural, etc.) you increase the chance that some vulnerability that would normally only impact a subset of the population becomes a systemic, existential risk. For example, imagine a society that is entirely dependent on a single crop that experiences famine when that crop is blighted. Or a society where everyone gets the news from one or two organizations and are thus extremely susceptible to propaganda. Diversity is important for what Taleb calls anti-fragility, while homogeneity tends to increase fragility.
I also think you run the risk of creating dysfunctional societies. Imagine a world made from only the hardest working and most ambitious among us. They could put that drive to use creating all sorts of wonders! Maybe...but I suspect they'd first spend a lot of time fighting and maybe even killing one another because everyone would want to be at or near the top of the hierarchy.
It takes all kinds to make a healthy society. We need ruthlessly ambitious people who will stop at (almost) nothing to fulfill their dreams, because that's how seemingly impossible dreams become reality. But we also need empathetic people to prevent the ruthlessly ambitious from running amok. We need introspective, philosophical people to come up with institutions that can check our worst impulses. And we need a lot of people who are just chill and average and who aren't worried about "fully comprehend[ing] what we know of nature, and be[ing] able to fully appreciate the mystery of the rest", because those are the people who do all the boring things that make the world work. My grandparents were pretty simple people -- they weren't all that ambitious and they weren't particularly deep thinkers. But holy shit! -- they could put their noses to the grindstone and get monotonous but necessary stuff done in ways that I just don't seem capable of replicating, even though I can spend way too much time writing a long comment on a blog post about polygenic embryo selection. There have been more than a few occasions on which I've envied them.
Now, I will readily admit that I can't point to a particular correlation between, say, IQ, or ambition, and some other high-risk or anti-social characteristic. But I'm a Hayekian and I worry a lot about about the pitfalls of trying to design or otherwise control complex systems with lots of emergent, difficult to discern but absolutely critical characteristics. There's also an evolutionary argument that the characteristics that people are are using to select their embryos would already be the norm if they are such an unalloyed good. And finally, I worry that we won't appreciate some of the characteristics that people have until we've bred them out of existence in the pursuit of a better polygenic success score.
At the end of the day, I just get really nervous about all the subtle, unforeseen side effects of trying to optimize on a small number of traits.
Footnote, and disclaimer: My view is very likely biased by ethical objections to IVF itself, so consider that when evaluating my arguments.
"Suppose you invest the $3,250 in an index fund that makes 7% over inflation, then give it to your future child when they turn 45 . . ."
Good for you, Scott! You thought about discounting. As an economist, I was working it out in my head and then saw you had the idea. But you didn't do it right. Here's my back-of-the envelope thoughts:
Suppose we can add 2 extra years of life by spending $10,000 today, and we can get 4% if we invest the money instead. By the Rule of 72, the investment will double every 72/4 = 16 years. Average longevity is about 80 years, so we get 5 doublings, so the $10,000 would be $320,000 at age 80. Would someone rather have $320,000 at age 80, or 2 extra years of life? (Or, if you don't care about heirs, $160,000 at age 64 or 2 extra years of life at age 80?)
The average American would prefer the money, I think. But the person thinking about spending the $10,000 is surely top 10%, and they might well prefer the extra years. Except, for them average longevity is higher, maybe 90, so it would be maybe $500,000 for two years. Keep in mind also that for anybody, their last two years of life are apt to be feeble, even though most rich people are doing pretty well at age 80 because they're living 10 more years.
I don't know, man, the vibes on this are fucking awful. And while I acknowledge how weak an argument that is, the post ends on this note:
> At some point your answer has to be romantic and aesthetic
My romantic, aesthetic self is screaming "bad idea" while my rationalist self makes fussy arguments about how this is fine, actually.
The trouble so much in the current incarnation of these technologies, which really do seem at worst likely to be harmless and at best to meaningfully improve some people's lives. It just seems somewhat obvious that there is an arm's race dynamic to this that won't permit us to get off this ride even if we don't like where it leads.
It depends on how much sense people have. Discussions below about animal breeders breeding to standards which aren't designed for the benefit of the animal give a hint about the risks.
For example, height seems like a purely competitive trait. It's not like there's something intrinsically good about being tall, but there are advantages to being taller than other people. There comes a point where being taller leads to health problems.
It's possible that there'd be selection for girls with very slender builds. Osteoporosis later in life? Who cares?
I wouldn't be surprised if some parents select for children who are extremely "typey" for their ethnicity. Very light, very dark, facial features on the extreme end of their ethnicity.
The net effect might still be good, but considering how obsessive people are about competition, I don't think trait selection can be reliably centrally controlled.
>For example, height seems like a purely competitive trait. It's not like there's something intrinsically good about being tall, but there are advantages to being taller than other people. There comes a point where being taller leads to health problems.
Yup! I inadvertently echoed you before seeing your comment here.
"I care about IQ because I bang up against the limits of my own a thousand times a day, and I hate it." I'll take Scott's word for it, but can't it also be the case that what Scott truly excels at - let us call it syndetic intelligence - isn't covered by IQ tests?
IMHO, the right thing to do when getting into a discussion that brings about the comment "how exactly does this differ from eugenics?" is not to either run for the hills or dig in, but to reread Franz Boas's classic article
The passage most relevant here is the one starts in page 476. Sure, you can say some of these things about human educational systems, but, say whatever you want about them, they are tolerant enough to allow for all of the following: (a) Scotts, Asimovs, etc., (b) great researchers, (c) great artists, etc. (Of course only a fraction of humanity gets the chance to truly develop its talents, and that's intensely shameful, but it's a different issue.)
This. Per Wikipedia, only 30% of natural conceptions result in pregnancies, the rest mostly fail to implant. [1]
Because, as you say, unborn children are still children, I consider couples who try to get pregnant through PIV sex to be monsters who are willing to statistically murder an expected two children to get one life birth.
Are you arguing that having a miscarriage (something you can't control) is equivalent to creating excess embryos with the PURPOSE of doing so being to only select the best one, and to destroy the others?
It's the pro-abortion set's attempt at false equivalence: an accident is exactly the same thing as deliberate murder, because death is the result in both cases, so murder is not in fact murder.
So for instance, speaking purely hypothetically, if you laid in wait for quiet_NaN and beat them to death with a baseball bat, that is *precisely* the same thing as if they slipped on the wet street and fell in front of a bus that ran over them. You can't be blamed for killing them any more than the bus can! (hence, I suppose, all the news reports about "cars running into crowds" and not mentioning the driver who rammed his car into that crowd, unless the driver was a white conservative guy).
Intention? Deliberate action? One was caused in order to bring about death, the other was an accident NaN might have survived even if severely injured? La-la-la, can't hear you, sticking my fingers in my ears, abortion is the exact same thing as miscarriage or stillbirth!
Indeed. I have seen a lot of false equivalences made in this comment section. Frustrating to see Scott arguing that choosing to take prenatal vitamins to prevent spinal bifida (nourishing an existing human life) is equivalent to selecting one embryo with a lower chance of spinal bifida out of a pool of embryos and destroying the others (deliberately killing embryos)
If this technology didn't require killing unwanted embryos and somehow was only ensuring only very healthy embryos are created, that would be a different story and I wouldn't have this particular objection to it.
A miscarriage is something which happens after implantation. We are talking about the worth of an embryo before implantation, which you assert is equal to that of a child, because "the unborn child is still a child".
I will grant @Deiseach that my comparison was a bit inexact. On your "embryos are kids" basis, the couple using IVF (with or without embryo selection) is basically creating a bunch of kids, then picking (either at random or through careful selection) one to keep and murders or freezes (which comes to the same thing for most of them) the others.
The couple trying to conceive naturally is doing the equivalent of creating a baby and then trying to smuggle it through the cargo hold of an airplane, where it will suffocate 70% of the time and be fine 30% of the time, until they succeed in getting one to survive. Sure, they do not want any of their smuggled babies to die, but they are also willing to accept that risk to get a kid, in a way which is not entirely dissimilar to the feeding shelter cats to coyotes meme. [1]
I will grant you that in this hypothetical, the IVF equivalent is likely slightly worse, deliberate murder, while the smuggling might be depraved-heart murder or manslaughter through gross negligence depending on whom you ask.
Furthermore, I will grant you that there is already historically established precedent that it is fine to bring kids in the world even though they have a high chance to die in infancy, because that was the case throughout most of human existence and is considered acceptable in all cultures. (But then again, we should be careful to conclude "something is ok" from "something is widespread", for example, some forms of rape were rather widespread historically and yet I am still opposed to it.)
However, even if I were to accept that these gambles are necessary sometimes, I would still expect the parents to do their utmost to optimize their kids chances to survive. A mother who refuses a supplement for a child which would raise its survival chances from 50% to 55% because she believes that it is up to god to decide the survival commits 5% of a depraved heart murder. With the implantation probability, I do not think anyone tries very hard to optimize it.
So, in short, my problem with religious right generally is that while I recognize that the idea of protecting the unborn is ur-Christian, they do not generally act in a way which is consistent with what they claim they value.
Take abortion. If you consider abortion to be equivalent to murder, then you would obviously want every girl to get on birth control before menarche. Sure, this might enable women to commit all sorts of lesser sins, like fornication or having-sex-on-birth-control, but surely a single act of filicide would stain her soul more than a lifetime of sinful sexual encounters?
Instead, the religious right is also bearish generally on birth control, which should be the lesser evil by orders of magnitude. From this, I conclude that the main goal is to coerce women not to have sex outside committed relationships on pain of being a single mother. If half of the unwanted pregnancies in Texas are terminated out-of-state and the other half are carried to term and serf as a cautionary tale for the other woman, that is clearly preferable to trying to avoid most of these pregnancies through state-sponsored contraceptives, even if that option would be much preferable to a utilitarian Christian.
With the status of the unimplanted embryo, I observe the same, but on steroids if anything. While I am sure that the onset of menstruation will make a couple trying to conceive very sad, I am also sure that they will be orders of magnitude more devastated if their two-year-old drowned in a pool.
From IVF, we know that implantation rate goes down with age, from 37% at 34 to 12% at 41. This is likely similar when trying to conceive naturally. So even if you consider it acceptable to create "kids" where two are expected to die for every one which gets born, if you care about about unimplanted embryos, you should probably stop trying at age 35. At age 41, you might make as many non-implanting embryos per birth through PIV sex as a 34yo undergoing IVF.
Again, if the religious right genuinely cared about the fate of unimplanted embryos, they would tell woman to be celibate or at least use birth control between age 35 and menopause. I mean, unlike the drowned 2yo, these are human beings who die before being baptized!
It feels a bit like the US caring a whole lot about the human rights situation in Iran (which is their enemy) and not a whole lot about the human rights situation in Saudi Arabia (which is their ally), so that one might conclude that human rights are mostly of instrumental value to them as a way to bash their enemies.
Leaving aside the political and religious right discussion, I'll say that embryos should have status as the earliest stage of persons and the decision to use the technology Scott is describing should include consideration of their status. Destroying an embryo is not morally neutral.
The topic at hand is whether it is good to use technology to select from multiple embryos that are created for the purpose of being weighed against each other, and most destined to destruction.
I am just arguing that giving unimplanted embryos the same rights as born children, as the OP was heavily implying, would have a lot of unintended consequences as far as normal human behavior is concerned.
Personally, if I was an embryo created through IVF, I would very much prefer to be created, even if there is a 90% chance that I am discarded and another 50% chance that I will die through natural causes. Suppose that the IVF technician already has picked up a sperm and an ovum an just as they move one to the other, they get a phone call "the implantation was successful, we will not need another embryo", and they proceed to terminate both cells separately. For the "don't destroy embryos" perspective, that would be better, but from the perspective of the prospective embryo, getting created, put in the freezer and then (very likely) destroyed would still seem like a better deal. It is not like things made out of a few cells can suffer, any plant will likely outperform them cognitively by a mile.
My very argument is that creating and destroying an embryo is morally neutral in itself (apart from being a waste of time and resources which could be better used elsewhere).
One way it would make a difference is if the embryo was ensouled at the moment of fertilization, and souls were either in limited supply or would suffer. But again, if we suppose that the human reproduction system is divinely inspired and not designed to cause a lot of unnecessary suffering (which seems a bit unlikely to me given what I know from history, but theodicy is another problem), then we have to conclude that during a woman's life, a few dozen embryos which never implant are created and starve was fine with god, and he certainly did not give us commandments which could have easily reduced their numbers.
"I care about IQ because I bang up against the limits of my own a thousand times a day, and I hate it."
From what I understand of very intelligent individuals who have made major achievements, I think this might mischaracterize what intelligence means. I think very intelligent individuals "bang up against their limits" all the time. In general they don't find it simple to do challenging and complex things, they spend tons of time and effort developing their skills in a domain and often feel like they aren't able to surpass a perceived limit. It often seems like obsession with something is a key characteristic to achievements that cause someone to be perceived as a genius.
I'm not saying "your limits aren't real." There are also people who feel like they are banging up against their limits and never surpass them. I'm arguing that, qualitatively, "intelligent person who feels like they are hitting their limit and never surpasses that limit" and "slightly more intelligent person who feels like they are hitting their limit but persevere and eventually achieve at higher and higher levels" probably feel about the same in the moment. And polygenic selection isn't going to change that basic phenomenon.
I might even argue that "banging up against limits" might be more of an intelligent-person phenomenon rather than an unintelligent-person phenomenon. It's not hard to live a life where you never challenge your mind to do new things. Continuing to challenge yourself and seek your limits is something I associate with intelligence.
Yeah, IDK Scott. Maybe you get 10 embryos, but they might not all stick, and many families need every embryo they get for successive tries. IVF communities are turning away even from conventional PGT-A testing, which just looks at whether there is a typical number of chromosomes. The accuracy behind PGT-A is being questioned, and with good reason, as there are some people having normal healthy successes with embryos that were deemed abnormal or mosaic. Because each embryo chance is so valued, people can't afford to throw any of them away, and the more you can retain the better. For these reasons, people are trending away from testing at all; most people just want 1 healthy baby, and my experiences along with participation is many IVF communities suggests that's all you can really hope for.
The personhood argument is well-worn territory, but apparently "proper thematic understanding of Gattaca" is not.
It's been some years since I watched it, so I'm sketchy on a few details, but from what I remember:
- It's not like "everyone else" genetically engineered their kid. The implication is that there's a large underclass of people whose parents couldn't or wouldn't genetically engineer them.
- Then there's Jude Law, who becomes disabled for non-genetic reasons, and realizes that he's suddenly in a world that only values you if you're a perfect Adonis.
- There's the doctor at the end of the movie who <SPOILER ALERT> lies to get Ethan Hawke into the rocket because he too has a non-genetically engineered kid who's been denied opportunity because of his genes.
The dystopia is that in the world of Gattaca, your genes are what define who you are, from career opportunities to what you pay for insurance. The system goes far enough down this rabbit hole that genetic potential replaces revealed achievement in a kind of false meritocracy.
At least one of the themes is, struggling against our imperfections is part of what makes us human. On, as Scott would say, an aesthetic level. And there's something insulting to our humanity about the idea that we could select away our imperfections.
This seems like "I have read a dystopic novel where doctors are trying to cure broken legs and depression with antibiotics and cause a lot of suffering, therefore I am against antibiotics."
If you hire an employee for a non-reproductive role, their actual IQ will obviously predict their performance much better than a genetic prediction of their IQ.
I do not see why this would run into a false meritocracy when a real meritocracy is so much more profitable for everyone involved.
> At least one of the themes is, struggling against our imperfections is part of what makes us human. On, as Scott would say, an aesthetic level. And there's something insulting to our humanity about the idea that we could select away our imperfections.
Tell that to the woman who decides to gets her breasts replaced preemptively because she has a family history of breast cancer, and to her sister who does not and dies of cancer at age 40.
Tell it to the IQ 75 person who will likely never achieve a lot of things most people here would take for granted.
Tell it to me, who has spent most of his adult life dealing with depression.
Are our struggles aesthetically pleasing to you?
While struggling is human, it is also very human to improve ourselves where we can. If a human is looking for a reproductive mate, they will not pick one at random but carefully consider which person likely would pass beneficial genes to their offspring. And it is bloody human to cheat where we can, too. If a human gets attacked by a coyote, the human will not try to fight with tooth and claw, he will use a club or a gun when he can. Do you really expect us not to cheat at genetics now that we can?
FWIW, using embryo screening, humans will likely still remain diverse and imperfect for the foreseeable future.
> Here’s the same table for other conditions, courtesy of Genomic Prediction (except I flipped the colors from the original, to match the one above):
I do not think that this is the "same table". The psych table is clearly symmetrical. For that one, I think I at least have an intuitive understanding, roughly "the Pearson correlation coefficient between the alleles which contribute to PTSD and the alleles for ADHD is 0.7".
The other one is much weirder. First off, there is likely a typo in the caption (4000 base pairs, not kBPs).
I can kinda make sense of figure 6 in the preprint. There is a bit of asymmetry, but this comes from different normalization: if you have 100 predictors for asthma and 10k predictors for basal cell carcinoma, and the overlap (within 4kBP) is 10 predictors, then you might have a correlatedless of 10 (implied: %) in the one direction and 0.1% in the other, fine.
But I am still puzzled by figure 7, where they put in the sign of the correlation somehow. First off, I would say that if your method says that genes for height are 83% correlated with genes for height, your assumption that anything within 4kBP is the same is probably not totally solid.
Systolic and Diastolic blood pressure shared 35% or 38% of their predictors in figure 6. (Only in the text, they say 18% and 19% instead.) In figure 7, their not-quite-correlations are 50% and 51%. I interpret this as "the most important predictors for systolic are also the most important predictors for diastolic, and point in the same direction".
The most significant negative correlation is between gallstones and hypertension in one direction, -80%. In the other direction, the effect is much smaller (likely because the figure 6 relatedness in one direction is 11%, but in the other direction it is 0.18%). But it is also positive: 0.067%.
Or take pulse rate and systolic blood pressure, which I picked because they share the same correlatedness in figure 6 (12%) and are also next to each other in the (for some unfathomable reason alphabetically ordered) list of measurements and conditions. At least, for them figure 7 has a positive sign in both cases, which seems sane. But why is the magnitude so different? 7.7% vs 18%.
Ah, I think that the answer is in equation 5, where beta_j (the correlatedness with the SNP which is currently not normalized) is only used directionally, in sgn(beta_i*beta_j). So if you have a correlation where beta_j is positive but very small, that will yield a large correlation in that direction (because the amplitude of beta_j does not matter), but a small correlation in the other direction.
For that matter, how do you even define the direction of correlations for single nucleotide permutations? Say that at a certain point in their genome (site X), half the population has an A and the other half has a G, (with C and T being directly lethal, for simplicity). The A variant has an expected systolic blood pressure which is 2 mmHg higher. 200 base-pairs downstream (site Y), you have another SNP, where you have either A or T, with T having an expected height which is 1cm higher. Due to our "less than 4kBP apart is the same thing" approach, we need to lump them together. But should we say "X=A and Y=A are basically the same thing" or "X=G and Y=A are basically the same thing"? This will affect the direction of our correlation! If on both sites, one variant is dominant, we might call the other variant "the weird one" and lump the weird ones together.
Finally, I do not like the overall smell of the data. For example, consider the field "education years". From priors, you would not expect that to be strongly correlated with most clinical diagnoses, and likely rather negatively than positively. Well, in the one direction, education length variants are positively correlated with everything except for "CAD 20 k" and T1 Diabetes, the largest effects are for two types of cancer (>30%). In the other direction, it is positively correlated with everything except that mysterious "CAD 20 k" again, but mostly on a level of a few percent. On priors, I would have expected that the correlation between education years and genetic diseases is mostly negative, because I would expect most variants which increase the disease risk not to have a positive side effect (if you count a longer education as a positive side effect).
In conclusion, I support superbabies so that future generations will have more brainpower to understand such papers.
Anorexia seems like a strange example to use for the More IQ Costs argument when there's a dozen alternative examples. Isn't anorexia much more prevalent in girls or is there an understanding this is only a bias in treatment?
If you ask almost every parent who is about to have a child, “What is your greatest wish?” You will most likely hear “I just want my child to be healthy.” If you would ask a public health professional, “What is the most important factor that determines whether a child is born healthy?” You will probably hear: “The age of the parents.” Consequently, on a population-wide basis, the most effective way to have a higher percentage of the children born healthy is to decrease parental ages. This is opposite of the current trend for increases in parental ages, especially among people who are likely to achieve graduate-level education. Restructuring the education system to accommodate parents better, including undergraduate parents is probably the best solution to this problem. It is extremely unfair that so many women need to wait so many years to have children and then feel compelled to go through IVF. I say women, because they are the ones who have to struggle through it physically. I also recognize that men are also extremely worried about having healthy children and may not feel confident enough to take on the responsibility of fathering children until they have enough income to provide for their families. So, it this way, they are under pressure too. Ultimately, a better public policy solution will bring better medical outcomes, as well.
Importantly, I think most of the age effect comes from the age of the genetic material, not the age of the (surrogate) mother having the baby.
So a kind of high tech solution would be to just put gametes of everyone into the freezer after puberty -- which admittedly will be a much more pleasurable process for men than women. Then 30 years later, instead of going through the process only to get gametes which have accumulated decades of mutational load, you just open your freezer and take the material from when you were 15.
On priors, I would expect that sperm would benefit more from early extraction and cryo-conservation as opposed to ova, because the oocyte is basically just passively waiting while the spermatogenic stem cells probably pass through a lot of generations during the life of a man. (Hm, a computer scientist could easily create a scheme where the generation count since conception of sperm cells would only scale logarithmically with the total amount produced, but knowing evolution no computer scientist was consulted when implementing the system which decides which primary spermatocytes undergo meiosis and which are kept for mitosis as required.)
The developments you describe are very positive for prospective parents in the future. I hope that most people will have such choices to help them if they find themselves in situations when they have to have children later in life. According to your example, the biological mother will be 45 years old when her first child is born. But this is not a viable solution on a population-wide basis. The 45 year old mother would have difficulty baring her own child, even if she wants to. She would not be able to breastfeed her own child, even if she knows that this is what’s best for the health of her child. Psychologically, she would have suffered through suppressing her natural instinct to have a child, which is a heavy emotional burden for years. Plus, the constant fear that the eggs that were stored for her have not been damaged, somehow. There are many issues regarding surrogacy as well. So, I still feel that a better public policy solution is preferable for the general population.
Waiting until ~after 30 as a matter of course really is an awful trend; my wife and I woke up to it juuust in time to manage a normal healthy family.
Reforming the education system as you describe would be a good start, but then there's the more subtle problem of career advancement in the workplace/job market. That's much harder to fix without some pretty heavy-handed intervention that would seriously clash with the notion of free enterprise. And, ultimately you're either spending a given daylight hour on accomplishing stuff at work, or on childcare: even if you force the structure to not penalize child-havers in terms of material rewards, there's still the issue of wanting the feelings of accomplishment that working hard in an interesting job can bring.
I really think this is the single most difficult and important issue in terms of how modern society is shaped. If it somehow gets solved, life in developed countries will become much less dysfunctional for everyone, and more just for women. I so far can't even in my imagination think of a good solution. At the very least, though, we're at least clearly on our way out of the woods on "more👏45-year-old👏unmarried👏girlbosses👏having👏IVF👏babies". Now here's hoping the tradwife meme doesn't take us too too far in the other direction...
I applaud all these developments. I have a cousin who has an advanced degree and was able to bare her youngest child with the help of IVF and without the help of a surrogate at the age of 52. I am very happy for her! I had another cousin, who got married after age 30, went through a few rounds of IVF unsuccessfully. She had serious complications after her IVF treatment that required major surgery. Unable to cope with that psychologically, she committed suicide at age 45. I still talk to her bereaved mother every day to give her some emotional support after the death of her only child. In sum, I definitely support IVF. I think it should be widely available. At the same time, it is not a viable solution for many women. We still need a better a better public policy to support young families. Yes, it is very complicated, but we must think about workable solutions.
Sasha Gusev has waded in on X with some questions about the lack of transparency in Herasight's methodology—"while clearly conceived and marketed around the idea of IQ prediction, has also elected to hide all the relevant validation until some date in the future."
He also lambastes Orchid, Nucleus, and Genomic Prediction for their sloppy methodologies: "People who care about this technology should be furious at Nucleus and their collaborators (as well as Orchid and Genomic Prediction for their own errors). Finding such flaws should not require reverse-engineering by a competitor. These products clearly need independent audits."
And there's a cast of sketchy promoters of race-based genetics who are in the Herasight orbit.
As for the risk-benefit calculus of using this product, Gusev wrote: "I personally think the benefits don't outweigh the risks and am genuinely bothered by the dodgy roll-out and affiliations. But I don't have any special insight on the risk/benefit calculation and I think people should be in charge of their own reproductive choices."
With regard to claims of accuracy, it seems to me that this should be something which should be easy enough to benchmark.
Find elderly sibling volunteers who are not in any genetic database. Sequence them (there genomes, not the humans), and hand the data to the prediction companies. Ask them to predict the relative genetic outcomes ("sibling A will have five more IQ points", "sibling B will have a +2% absolute risk to develop breast cancer", etc).
Of course, the benchmarking has to be done by a neutral third party. Preferably, have the predictors run their models on an offline laptop on-site, so that nobody will be tempted to id the volunteers and google them.
In the long run, I think we should have open source predictors, so that independent researchers can examine how they come to their prophecies and how much they are just testing for confounders.
Number-crunching through genetic databases is likely not something I would want to try on my laptop, but it also does not seem like the kind of thing which is on par with training LLMs. Basically academics should be happy to do this for publications ("we found a SNP which explains 0.01 IQ points between siblings"), and then you "just" need someone to sanity-check and curate all these findings into a model.
>Everyone else probably wants a generically healthy embryo with low risk of all conditions. Exactly how this works depends on the customer’s own values - would they prefer an embryo with lower cancer risk to one who will have fewer heart attacks? - and the exact benefits will depend on how parents make that decision.
"Good problem to have" though it may be, this level of information/control sounds like an absolute nightmare. I need to make a choice among several points on a many-dimensional pareto front? And if I choose wrong my children suffer? Wow, sign me up!!!!
Actually, I *would* sign up, but I would be absolutely miserable over it.
I see huge risks in wide adoption of such selection. If evolution hasn’t weed out some genetic variants over millions of years it means that they were useful at some point in time in some context. Even if we ignore possible interactions between genes (which can at least potentially produce unique traits), there are still environmental factors making some genes more or less important. We can expect that some genetic variants widely presented and seen negative by scoring was positive some time ago so collective evolution process “decided” to keep it.
You can say “evolution fails at fast changing world like ours, we are no longer exposed to lots of past environment factors”, but are you sure we are good enough at determining variants no longer relevant? Will our quite simple polygenic models and weighting of diseases hold in 20 years? Are we sure there wouldn’t be local famine where genes giving us diabetes become critical for survival? We once thought appendix is something unnecessary and dangerous and blamed evolution for it, but now we know it’s important. Are we at point where we can outplay evolution at world population scale?
Of course we can say that this scores are best estimations for now, but I haven’t heard about any adaptation of models for expected advances in medicine for example.
Speaking about IQ specifically. We know it depends on so many genes and we know that they were under high selection pressure. Have we checked that good variants from 20 years ago are still good today?
Polygenic scores or other models might be overfitted for current state of the world and it’s not just calories intake. Imagine gene that allows child pay 30% more attention to face on screen. Probably negative in 90s (more TV), probably positive during Covid. Or imagine gene reducing long term effects of Covid on brain but lowering IQ one point. It probably kept in gene pull by evolution because we get pandemic infections once in a while but of course it’s negative for carriers between them. Since IQ is depending on so many factors and so sensitive to them I expect any model not taking it into account to lose accuracy fairly quickly (of course there will be more robust genetic marker). It is not a problem for personal decision, you just expect less accuracy, but it will have unintended consequences if half of population adopt such models.
Also wide adoption of technology will significantly reduce variance of population and we can end up with higher average but lower let’s say 90th percentile of IQ. At least at this point of time I see 90th percentile more important for technology and quality of life progress
This frustrates me immensely, because you are focus on material issues when cultural issues will be a far bigger threat.
The big issue is that children routinely struggle against being dominated by their parents's wishes and values. Currently the idea of serendipity helps keep this in check: while children are begotten from parents's genetic material, they are a blending affected by time, chance, and many other factors.
If you select for IQ, you are going beyond just giving them material advantage; you are changing the relationship between parent and child into parent and product. It's bad enough that parents expected their kids to be little geniuses and get mad they underperform or struggle, what happens when they bring out the invoice from the company saying they paid for you to be built better? Psychic damage will be immense, because products are defective.
Plus you always push values when you do this. The focus on IQ is mostly due to upper-middle class competitiveness in seeking power and status. It's not always about the kids because the kids may be happier with less status and less pressure. But you don't want to enshrine values through genetic selection of traits because a lot of them are full of shit. That kid may be happier not trying to be a master of the universe, and maybe we need less jockeying for power and status overall.
I think "Frankenstein" is the cautionary tale here; Victor sets out to create his idea of the perfect being, achieves it, and is then horrified by the result and abandons the child he made, with bad results.
"Begotten, not made" may be good enough for the Creator of the Universe, but not for middle-to-upper middle class white Westerners competing in the ceaseless strife to get Junior into the best preschool onwards.
my favorite example in media is an anime called Geneshaft, which is 50% silly anime tropes like cute girls and giant robots and 50% weirdly hard science fiction about this.
Like the spunky protagonist gets down about her ability to pilot the robot and calls her "mother," who assures her she hand picked the necessary traits to succeed in her role, talking as if she is a robot.
Or how with genetic engineering society does prefer girls, with boys mostly there only as needed and with a monitor to keep them in check. Girls are generally more stable and reliable in that future, and in a weird way our own.
anime and manga are odd. Oh there is now a rationalist manga lol, Qualia the Purple. You want to explain what a p-zombie is that one is a terrifying example of it.
>I think "Frankenstein" is the cautionary tale here; Victor sets out to create his idea of the perfect being, achieves it, and is then horrified by the result and abandons the child he made, with bad results.
Yup, oops! As a cautionary tale (and cautionary tales, both as fiction and as poor predictors of reality, themselves need to be treated with caution...) it is indeed a good match.
>"Begotten, not made" may be good enough for the Creator of the Universe, but not for middle-to-upper middle class white Westerners competing in the ceaseless strife to get Junior into the best preschool onwards.
Hubris? Who, us? :-)
( Admittedly, as a childfree man well past his potentially reproductive years, I don't actually have a dog in this fight... ).
Not sure I get the snark, as that's my whole point: we don't relate to our kids as a creator does to his creations, we beget them. God can make but He is assumed to always be good, and also far superior to us in nature. But we are neither.
The "only begotten" is theological in part to show why Jesus can be the sacrificial lamb, and also to prevent more divine sons of God popping up to claim the mantle. We are sons of God by adoption, and also the bride of Christ in analogy. we pointedly don't relate to God only as things to fulfill a purpose, in Christian theology at least. The more strains of it think that, the more inhuman they become (predestined vessels of wrath)
the angels were created, and christians don't know what to make of them.
Well, I was poking fun at the theological distinction, and agreeing that embryo selection (and genetic engineering, if it becomes an option) does have some resemblance to the inverse of the theology (modulo the Nicene Creed's "begotten" isn't _really_ the same as human begetting...)
Now, I, personally, don't see this change as all that much of a big deal. Parents try to mold their sons and daughters in many ways today, so I see this as an incremental change.
( Admittedly, as a childfree man well past his potentially reproductive years, I don't actually have a dog in this fight... )
>The "only begotten" is theological in part to show why Jesus can be the sacrificial lamb
I've always found the force of the distinction to be odd (admittedly looking at the whole thing as an atheist/agnostic). An omnipotent deity could do whatever it wants - making a sharp distinction between the possible _mechanisms_ by which it generates a more-or-less related deity, or anything else, seems like a strange thing to care about to me. Maybe for a deity with limited powers we'd care about side effects of different mechanisms, but that isn't the model in Christianity, as far as I know.
> What if they selected your brother for everything great, then had you naturally?
Presuming your brother is older (because "then had you"), there are two possibilities for change: they could have avoided selection of traits for the first child, or they could have selected traits for the second child. Considering the first case, we're just envious, and somewhat despicably so. Considering the second case, that would only improve things for this child if they believe it was better had they never been born. There's no "you but with better genetics," because that "other you" is actually someone else, which is more transparently obvious in this case than the usual "put yourself in someone else's place" thought experiments, since it's about embryos that already literally exist.
Creating dozens, or hundreds of embryos with the purpose of destroying all but the "best", "best" being determined not by viability but by subjectively desirable traits, is unethical. Two reasons off the top of my head.
1. embryos have moral value because under natural circumstances (regular ol sex resulting in conception) they will likely develop into people. Yes miscarriages occur, this doesn't change the argument. No, sperm and egg separately do not have the same moral value.
2. It devalues human life in general by making the explicit determination that someone with blue eyes is more worthy of existence than someone with brown. Or that someone with a higher IQ is more worthy of life than someone with a lower IQ. The foreseeable downstream cultural effects of this are frightening.
Yeah, there appears to be an incentive to create as many embryos as possible.
The upper limit is just the point at which the woman becomes too physically uncomfortable for additional retrievals and the pool of trait combinations tapers off. The more traits get selected for, the more embryos would need to be created to get a desired combo.
Embryos created via IVF do not "likely develop into people". You have to go through the active step of carefully inserting the embryo into the uterine lining of the woman for this to happen.
Yes, that's true, but my point was that the embryo has a purpose and a potential. The purpose and potential is to develop into a baby and eventually into an adult human. The fact that an embryo is an early stage adult human warrants consideration before creating them and destroying them.
Vaccines are one of the most amazing human inventions and have saved countless lives, beyond any statistical doubt and with a very large effect size. Yet there are countless anti-vaxxers who say "my cousin got the flu vaccine and then felt sick, vaccines don't work, I'm letting my kid grow up naturally."
The effect sizes for these effects are far smaller. If this becomes popular, millions of families will have a designer baby that grows up to get lower grades than a non-designer baby and they will say "that designer baby thing doesn't work."
Humans have a very strong naturalistic fallacy, that anything natural is good and unnatural is bad. Obviously these aren't ironclad -- not everyone believe the naturalistic fallacy, most people give their kids vaccines. But still, I predict those forces will put a lower ceiling than optimists might like on rapid adoption of this technology (assuming it works as well as claimed to begin with.)
The naturalistic fallacy is not a fallacy. It is a fundamental human value.
If your view is that unnaturally screening embryos for higher IQ is a necessary evil to combat the challenges of this world, then your opinion is worthy of a discussion. But if your view is that screening embryos for IQ has no moral problems whatsoever, even though it goes against nature in an unprecedented way, then your morals are alien to mine.
>The naturalistic fallacy is not a fallacy. It is a fundamental human value.
Yet you type that on a machine with transistors of elemental silicon, a material found nowhere in nature.
And I'll bet that your lifestyle is as far from our environment of evolutionary adaptation as mine or Dylan Kane's is.
The naturalistic fallacy is indeed a fallacy. "Natural is good" is profoundly mistaken.
EDIT: A point you know, but perhaps have forgotten:
_LITERACY_ is unnatural
Reading and writing are _not_ part of our environment of evolutionary adaptation. We have been human for around 300,000 years. Writing was invented only about 6,000 years ago. Children acquire spoken language spontaneously in a family. Reading and writing have to be actively taught to them. If you want to reject the unnatural, you must reject the written word.
There are decent arguments for being skeptical about embryo selection. Scott covered a number of them. The naturalistic fallacy is not one of them.
> “Although in general higher scores mean you are more likely to have a condition, many healthy people will have high scores; others might develop the condition even with a low score”, says the Society for Psychiatric Genetics, as if they have just blown the lid off some dastardly conspiracy.
I know that many people here are used to being surrounded by intellectuals trained to reflexively do Bayesian statistics, but you have to remember that *this is not at all obvious to the average person/most people*. In fact, there's a significant portion of the population who I've come to realize might actually be incapable of understanding that.
For someone who interacts with the public and is aware of the existence of these sorts of people, you can see how someone might come to the conclusion that PGES companies are trying to mislead people about efficacy with the claims they're making (even if it's almost certainly those companies just speaking from a place of the curse of knowledge).
" I care about IQ because I bang up against the limits of my own a thousand times a day, and I hate it. I fantasize about ways to make my children smarter than I am for the same reason a dog confined in a tiny crate might fantasize about getting her puppies adopted out to a nice house with a big grassy yard.)"
Well said. There's another reason, though. High-IQ people can be evil, but they're not stupid. They don't shoot their best friend by accident when they get angry. They understand that someone who said something offensive didn't mean to offend them. They know how to have good manners that please other people at low cost to themsevles. In short, they have positive externalities. And that't even without talking about innovation or artistic creation.
an average person will cheat on his wife, a smart person will cheat on his wife, write a book about how cheating is morally justified, and convince others to cheat too.
Intelligence is just a force multiplier, it isn't any better in a moral sense. There's a bad idea that evil is stupid, but no, smart people commit it too, and can even systemize it.
"High-IQ people can be evil, but they're not stupid."
No, they don't shoot someone on impulse when angry; they go off and plot a tidy murder. As to "understanding someone didn't mean to offend them", please tell the wider world that. There's plenty of smart people (or people who like to think they're smart) getting very angry over offence even if it was not intentional.
This article is deeply concerning. It fails to address the most obvious argument against selecting embryos for IQ or appearance: that it breaks from nature in a way no other technology has. Selecting embryos for these traits disregards the natural process by which humans are created—something no previous intervention has done so thoroughly. Once we begin down this path, we risk losing what it even means to be human. What do you truly value? Aren’t your values rooted in the human condition itself? Isn't "the natural way humans are produced" an essential value? Some problems are so ugly that they are worth betraying nature to solve. Increasing IQ is not one of them.
You say that selecting embryos for IQ or appearance breaks from nature in a way no other technology has. I disagree. Birth control pills are a technology that separate sex from reproduction. 14% of women in the USA aged 15-49 take birth control pills. Only 2% of children are conceived using IVF. This new polygenic scoring technology provides parents with additional information. IVF parents are going to choose which embryos to implant, should they make the choice randomly? Are you against PGT-A screening for down syndrome? Hiding data from parents is not the answer. Humans are diverse with diverse values and will make diverse decisions when it comes to embryo selection. This technology is inevitable.
It's unprecedented not because it's the first technology to "separate sex from reproduction," but because it's the first technology that allows parents to make value judgements about which types of people get to live, beyond screening for debilitating diseases. Intelligence may be a valuable quality in the abstract, but is a smarter child more deserving of life than a less intelligent one? Are intelligent people worth more than less intelligent people? I assume you would say no to these questions. In short, it's immoral to choose who gets to live based on subjective values. I'm not sure if uptake is inevitable, but even if it is, that obviously doesn't mean I need to support it.
Decisions based on "subjective values" happen all the time in mate selection. Talk to people about how they chose their partners and you’ll hear stories like “I noticed her because she wore a red dress to the dance,” or “he looked cool smoking outside.” Subjective value judgements in mate selection are far more impactful than embryo selection. Consider sperm donor selection: a woman flipping through donor profiles is making decisions of who is more "deserving of life" based on appearance, education, health, and personality all before conception even occurs. In comparison, choosing among a small number of very similar embryos is relatively inconsequential. And you asked if an intelligent child is "worth more" than a less intelligent one. Does society value intelligence? Yes, it often does. Intelligence correlates with prosocial behavior and the ability to make meaningful contributions, and society tends to reward it. It is innately human and natural to want healthy, happy & smart children. Breast cancer risk, diabetes risk, IQ, risk of Alzheimer's... I wouldn't exactly call these "subjective values." But, I respect that you don't support it. Society will ultimately decide whether or not to accept this new embryo selection technology.
Good point. You’ve brought me closer to your perspective. That said, I’ve always been uneasy with the idea of mate selection as a moral framework. It feels wrong to judge someone’s worthiness as a partner based on objective traits like IQ, rather than personal, indexical ones like how deeply you connect, or the good they’ve done for you.
Still, mate selection is a natural process; embryo selection is not. And I don’t think the appeal to nature is a fallacy here. Valuing what’s natural is a deep human intuition. Part of me feels that using artificial technology to amplify natural ugliness crosses a moral line.
Don't assume that. We've plenty of people who have commented on here that yes, in fact they *do* think smarter people are better and more deserving, and that Huxley's "Brave New World" with embryos treated and decanted and conditioned to fit into certain niches is something we should be striving for, not reading as a cautionary tale: but just think of the soma and the feelies! the dumb people do the dirty dangerous work without complaint and are no threat to us smart types at the top of the pyramid!
“Consider e.g. Genomic Prediction, which costs $3,250 for five embryos and claims to lower absolute risk of Type 2 diabetes by 12%. That implies that not getting Type 2 diabetes is worth $27,000. Ask anybody dealing with regular insulin injections (let alone limb amputations) whether it would be worth $27,000 to wave a magic wand and not have Type 2 diabetes! It’s not a hard question!”
Linear extrapolation of the utility function to zero Type 2 diabetes risk is wrong. The PRS scores have a floor (well) above zero risk. While I applaud an attempt to quantify what these screens may be worth, a linear utility function can’t be the right way to think about this. It will always be biased towards making this technology seem far more effective per dollar than it is.
First of all the floor for type 2 diabetes is pretty damn close to zero if you can modify enough genetic variants. In theory you could get up a ~98% risk reduction if you could change enough genes (though you're not going to be able to get anywhere close to that with embryo selection)
But second of all... I don't understand your objection here.
If I can reduce the odds of getting diabetes by 12%, then surely that is worth 12% of the value of not getting type 2 diabetes, right? Maybe you want to argue we should do some time discounting (frankly I would agree with that)
Scott's 12% reduction is absolute risk. It's obviously true that if the average risk among a family's embryos is lower, the intervention on that disease will be worth less.
You can reduce your risk of getting Type II diabetes by being damn careful about what you eat in what amounts, getting exercise, and getting regular health checkups.
The irony here would be that the genetically selected babies now have much, much less risk of Type II diabetes so the food industry can continue to turn out highly processed, filled with sugar, junk food for fun and profit since the consumers are now able to eat as much slop as they want with much less fear of consequences.
Your “98% risk reduction” switches the topic from embryo selection using PRS to hypothetical large-scale gene editing. Even under that premise, it can’t be 98% of total T2D risk—non-genetic drivers remain, heritability is <1, and PRS captures only part of the genetic share. At best, that number might refer to a slice of the polygenic component—not total lifetime risk.
Back to the actual technology: the blog’s linear extrapolation—“$3,250 buys a 12-point reduction, so a cure costs ~$27k”—assumes constant returns and that the remaining risk is purchasable. But it doesn’t work that way.
Maybe this analogy helps: a seatbelt might cut fatality risk by 50% in a crash. But you can’t wear eight seatbelts to get to near-zero risk. There’s irreducible residual risk from the crash itself.
The linear extrapolation overstates the benefit and risks distorting how we think about cost-effectiveness and where this technology is actually useful.
And maybe you think this calculation isn't literal. It's illustrative. I'd buy that except the author brings up the prospect of not having amputations and insulin injections for only $27,000. That's rhetorically pretty loaded. Either the number is a rough utility estimate and we stay in abstraction, or it’s a claim about tangible medical outcomes—at which point it needs much more caution and nuance. You can’t have it both ways, and that's the point of my initial critique of the framing.
>And then there’s AI. The average age at diagnosis for Type II diabetes is 45 years. Will there still be people growing gradually older and getting Type II diabetes and taking insulin injections in 2070? If not, what are we even doing here?
Very much agreed.
I wish the people in this endeavor well, but, barring a truly Antarctic class AI winter, I can't see them making much difference (from a broad policy view).
Here's my question: out of the four kids in the header image, which would you keep?
Personally I've spent some time staring at it and probably the third? It's got the lowest chance of diabetes, Alzheimer's, and hypertension. But on the other hand it's also got the lowest IQ. How should I be thinking about trading off IQ for a slightly reduced probability of hypertension? Should I just take the smarter kid and be careful what I feed them?
If this is a typical result of embryo selection then this small menu of barely-different options doesn't seem worth the effort or expense.
There is something about this entire discussion that makes me nauseous. I feel like I’ve woken up from a coma 15 years in the future. It’s just so surreal to see this topic garner such serious, meticulous discussion because we’ve actually reached the point where it can’t be avoided. My unwillingness to even want to face the issue also just makes me feel old.
Unhappily, unless we face it now, we'll have nothing to say when it is commonplace. I'm quietly amused, in a grim way, by the silence on stem cell therapies.
You remember, if you're old enough, the dazzling new advance that was going to do away with all hereditary diseases? Being blocked by those stupid old sky-fairy believers and Luddites? We were only five/fifteen years away from no more [insert ill of your choice]!
And then they got the permissions they wanted, after hard fighting and resistance, and the embryonic version was sidelined without fanfare (because that was all about research, in fact,. and not miracle cures) and the emphasis was now on adult stem cell research. Which did *not* require destruction of embryos, which was what us stupid old sky-fairy believers were happy with in the first place. And then attention moved on to "hey maybe we don't need embryos at all, we can convert other cells into stem cells".
Yes, stem cell therapy is working. But it's not the dazzling miracle cure for every ill that the promoters sold it as, and I wonder if this embryonic selection is going to go the same way: give it twenty years or so, and it'll be the preserve of for-profit companies holding out the lure of miracle cures:
>suppose that I had a natural, unselected child, and that child became high school valedictorian and got into Harvard. Would my first reaction be “Oh no! This slightly raises her risk of anorexia!”? If not, why should this be our reaction to artificially increasing IQ?
...because her genetic risk of anorexia was already locked in at birth, whether or not she went to Harvard, and there's nothing you can do about it now, and therefore no reason to suddenly react as if something new has happened? So you may as well celebrate the good thing?
Whereas, if you are deciding whether or not to hit the 'Harvard + Anorexia' button *before* they are born, then you absolutely have to consider both the upside and the downside.
I think a more analogous story would have captured the point that you are making a decision. Something like: 'If you hear that a child got into Harvard, do you put slightly more effort into checking their eating habits and asking them how they feel about their body, to screen for early warning signs of anorexia?'
But the answer to that is... yes, sure, you should do that if the connection is predictive enough for it to be a rational precaution.
(Note, I take the point of the paragraph, which I think is just 'many people would happily take this tradeoff.' I just think that asking what your emotional reaction to someone getting into Harvard would be is a poor way to frame the intuition, because celebrating a good thing in a world where bad things exist is very different from deciding which tradeoffs to make)
If everyone starts using this technology, it will have unforseen consequences for sure. For example, there is a correlation between schizophrenia in the family and creativity. Maybe if in future, we filter out many of the genes associated with schizophrenia, we will also filter out many of the geniuses.
It’s not just certain ethnicities, believe me. It’s now normal in the US upper middle class to implant only female embryos, and of course (white, native-born) women of the same class talk very freely about how they would abort a male embryo/foetus if they got pregnant the natural way. (Epistemic warning: anecdotal observation.)
I think this article overlooks a few important points.
First, there is a concerning genetic correlation between IQ and Autism, which I raised in my article (https://theinfinitesimal.substack.com/p/science-fictions-are-outpacing-science) but which this post avoids mentioning in favor of anorexia. Given the general concern and anxiety people currently have about ASD risk, especially in the high-edu / high-income clientele that Herasight is catering to, this really should be front and center. If I had to guess, Scott ignores ASD in favor of anorexia because proponents of PGT argue that IQ is only genetically correlated with Aspberger's and not clinical ASD and therefore poses less of a risk. To the best of our current knowledge this is not true, see for example Figure 3 in Grove et al. (https://pmc.ncbi.nlm.nih.gov/articles/PMC6454898/) showing a statistically significant association between the IQ PGS and childhood autism (hCHA). If you're selling parents a product that might increase the risk of childhood autism, I think that needs to be on the warning label.
Second, all of the results for antagonistic pleiotropy are based on population-level estimates which will be confounded by assortative mating (in particular the fact that people with high education / high income / high IQ / high health mate assortatively will make "good" traits look correlated; and the fact that people with mental health problems / low income / low health mate assortatively will make "bad" traits look correlated). What is relevant for embryos is the within-family genetic correlation and this is simply too noisy from the current data (e.g. Tan et al. family GWAS https://www.medrxiv.org/content/10.1101/2024.10.01.24314703v1). There also plenty of traits that do not have good GWAS data or good biobank measurements to evaluate this question at all. So this continues to be a leap of faith.
On the topic of population genetic correlations, the post considers the market benefit of IQ versus the cost of embryo selection. But this analysis uses phenotypic correlations between IQ and income, which is even more confounded, and will include any paths where SES -> IQ and SES -> Income. Interestingly, in Tan et al. the population level genetic correlation with individual income is actually much higher for the Educational Attainment GWAS than for the IQ GWAS. If the goal is purely return on investment and we're relying on population r_g, then selection should be on EA rather than IQ. EA also has generally higher genetic correlation with various health outcomes, too. I'm surprised this wasn't mentioned.
Finally, there are a few references to embryo selection curing or eliminating certain conditions but it is important to understand that embryo selection works very differently for polygenic traits than monogenic traits. When you screen out a rare Mendelian disease, you are removing carriers of the mutation and therefore removing that risk from the offspring and arguably "curing" the disease. But when you screen out for common PGS, you are subtly shifting the underlying liability from just above some clinical disease "threshold" to just below the threshold -- this is where the majority of the "risk" reduction comes from. The easiest way to understand this is with the example of obesity, which is typically defined as BMI>40. Embryo selection can appear to reduce obesity risk by an impressive 50%, but what it actually does is reduce BMI from a mean of 41 to 40 (in obese families). YMMV on whether a BMI reduction of 1pt is worth the cost/benefit but it is clearly not a *cure* for Obesity except for in a very pedantic way that happens to make the statistics look good. This is not a nitpick, because many of the traits that are being screened on are actually just arbitrary dichotomizations of an underlying continuum (Alzheimer's: underlying cognitive / decline tests; Schizophrenia: defined by a sum of symptoms; T2D : Glucose/hba1c levels; Hypertension: blood pressure; Gout : uric acid; VTE : clotting factors; Glaucoma : eye pressure / nerve damage; Osteoporosis: bone mineral density scores). Even for cancer, the most likely mechanism is an extension of the age of onset or rate of growth rather than a "cure" to the disease. In my opinion, it is misleading to present large relative risk reductions on threshold traits when the reality is a small quantitative reduction on the underlying continuous trait.
I was curious about pleiotropy! I wondered if the correlations shown were population or within-family; concerning to hear they are population. So I take your comment to mean we still don’t have good estimates of pleiotropy from within-family data?
For people already doing IVF, this seems like a no-brainer. But you also suggest people might electively do IVF in order to do this, and it's much less clear to me that is a net win. There are a lot of selection effects that occur in natural reproduction that aren't replicated by IVF (for example ICSI replacing sperm competition), and many potential insults to the embryo that are IVF specific (e.g. freeze-thaw). If you have a small number of embryos to choose from you could easily come out net behind.
This is really compelling—and raises some quiet discomfort for me. It feels like we’re drifting toward this subtle optimization trap, where even our kids’ traits start to feel like variables we’re supposed to tweak.
I get the appeal—if you can reduce suffering or give your child a small edge, why wouldn’t you? But I also wonder what we lose when everything becomes selectable. Some of the deepest parts of being human come from the unchosen: the weird mix of strengths and struggles that shape who we become.
So how exactly does this differ from eugenics?
If you mean "how does it differ from the kind of eugenics where you murder people", the main difference is that you're not murdering people.
If you mean "how does it differ from the kind of eugenics where people screen for carrier status before they have kids together", the main difference is it's more effective and works for a broader number of conditions.
If you just mean "nice technology you've got there, shame if I were to associate it with a vague and inconsistently used term that has negative valence", see my old post on the Worst Argument In The World - https://www.lesswrong.com/posts/yCWPkLi8wJvewPbEp/the-noncentral-fallacy-the-worst-argument-in-the-world
Maybe you've done this elsewhere (please point me to it if so!), but it would be helpful to justify the idea that embryos aren't people because if they are you're doing precisely that.
See discussion downthread at https://www.astralcodexten.com/p/suddenly-trait-based-embryo-selection/comment/140960824
I agree with the argument that the couple choosing not to have a 4th child is not equivalent to murder, but I still think you're dodging the actual moral question of eugenics. With embryonic trait selection, the parents are actively making a choice that some kinds of life are not worth living e.g. Downs syndrome. Talking about this as "eliminating disease" is just not morally consistent with how we think about curing disease in living creatures, individuals are deciding those genes shouldn't exist. This is an act of commission by the parent(s).
I reflected on your post about ethical objections based on personhood, and want to offer some reactions. In order:
1. The alcoholic: This is an act of omission. Yes, there is an act of commission to avoid changing an embryo in a harmful way, but really she's choosing to not do something. It is a choice that a life with fetal alcohol syndrome is not as desirable as a life without fetal alcohol syndrome, which has bearing on the topic, but it's not quite the same because she's not already pregnant. A better scenario is if she was pregnant and a heavy drinker, and she has to choose to abort the child, quit drinking, then get pregnant again. Of course, if you think an embryo isn't a form of human life, it may be academic to you, but you may find differing responses to an act of omission vs the act of commission of choosing a life isn't worth living.
2. The intern: This is a delegated act of commission to select for a trait, and the delegate changes its mind based on new data. We’re not arguing about destiny here, we’re arguing about intent of the parents.
3. The parenting workshop: Honestly, this situation feels so manufactured, I don’t understand the relevance. Here, an act of commission (the workshop attendance) has some probabilistic effect on the children in the chaotic, nonlinear system that is life which you can’t control for at all. But polygenic screening is an act of commission with a probabilistic effect on the child born and a completely deterministic effect on the kinds of people not born.
My aunt has some kind of mental disability since birth, though she has led a fulfilling life through her family's efforts and her own. And now I can just choose that life doesn't deserve to exist, at least not as much as another one. That kind of power gives me pause, both for the moral implications and for what it does to the human genetic pool over time. There's probably some useful data about genetic populations that ritually killed children with undesirable traits (Sparta I think?) to understand more about the second-order effects, but I'm willing to grant the first-order effects are almost always greater than the second-order ones.
FWIW these questions are not academic to my family, we meet the criteria you outline, but still both feel deeply uncomfortable with making a decision like that. I am sympathetic that others that have experienced or witness terrible genetic afflictions may feel strongly the other way and simply not want to experience it again themselves, but I feel differently about accepting such tech at the limit and endorsing it as a mainstream technology that everyone should do for any reason they feel like.
I think it's pretty easy to route around your objections here. For example, suppose that it was discovered that a certain prenatal vitamin, let's say Vitamin C, decreased the risk of Down Syndrome (I think this is plausible; even though Down Syndrome is a genetic condition, it has to do with chromosomes failing to sort into sells normally, and some vitamin could improve the sorting process). Do you think it would be immoral for doctors to tell women trying to get pregnant to take Vitamin C?
(sorry, the next part is going to sound kind of anti-intellectual and ad hominem, but I think we're going to go in circles forever without it)
I worry you're going to come up with some set of complicated philosophical rules around probabilistic acts of omission or something that successfully rules in the Vitamin C scenario but rules out the polygenic selection scenario, and I will think those rules aren't interesting/relevant, and we'll be right back where we started. The reason I wouldn't ban mothers from taking Vitamin C isn't because it's *almost* an evil act but barely escapes because of something something probalistic acts of omission. It's because it's not even close to having anything wrong with it in any way.
I feel like it only starts to seem wrong if you're doing some kind of very vague zooming out and squinting, such that every act is judged by things that it vaguely rhymes with as seen through the lens of some kind of extreme and selective paranoia. For example, if I prepare to cut off my child's arm with an axe, and you say "No, don't do that!", then I say "OH, SO YOU'RE SAYING THAT ONE-ARMED PEOPLE ARE SCUM WHO DON'T DESERVE TO LIVE AND SHOULD ALL BE GENOCIDED?", I am doing some kind of weird paranoid drawing-implications-that-I-shouldn't act. We don't require any kind of complex philosophy to justify why I can suggest you not cut off your child's arm, it's a completely normal action that only seems related to evil things if you doing the zoom-out-squint-and-be-paranoid manuever.
I think it is better to not have a disease (including a genetic disease) than to have it (I try to justify that more at https://www.astralcodexten.com/p/nobody-can-make-you-feel-genetically ) I don't think this justifies hurting people with existing genetic diseases (because that would hurt a real, living person). But I do think it justifies creating people without genetic diseases rather than people with those diseases (since this isn't hurting any particular real person with a claim not to be hurt). In order to deny this, you need to create some kind of concept of potential-people's-rights which is going to imply all sorts of insane things later, then come up with new ad hoc shaky philosophical principles to whack-a-mole down each insane implication one by one.
My claim is that this comes from basically the same anti-playing-god place as opposition to vaccines, opposition to antibiotics, etc, that each step gets grandfathered in as it no longer seems like scary new technology and instead is just part of the scenery, and people invent new ad hoc philosophical distinctions each time to explain why last year's "playing God" is obviously okay and it would be horrendous to deny a dying person antibiotics, but this year's "playing God" is still bad. I think instead of doing that we should just abandon the "playing God" concept - at least if we're not religious and don't think there's an actual God who objects to this - and just try to make the world better, as best we can, while moving slowly and watching out for risks.
(an analogy for why I am not so concerned about risks in this case: suppose that we learned most existing disabilities are caused by some kind of radiation coming from the Earth's core. Then there was an earthquake, it disrupted the earth's core, and the level of radiation cut in half. Then only half as many children were born disabled as before. Would you support randomly mutilating / lobotomizing / irradiating babies until we returned to the current disability rate? Why not? It seems like either you must believe that, in expectation, the pre-earthquake level of disability is likely to be better, because there's some compelling reason why having fewer disabilities would be risky or disrupt society, or else you must believe the opposite. If the latter, you should support the random mutilations; if the former, you should support efforts to have fewer disabled children born now. I think almost nobody would support the mutilation in this scenario, and that's because they aren't drawing off some real belief that there are potential risks that outweigh benefits, they're just going off of the belief that all previous ways of playing God are fine but all new ways of playing God feel scary or like they're making aggressive ethical claims)
(here's another thought experiment - suppose that a genetic engineering technique is discovered that lets people "correct" cystic fibrosis after a child is born, but it costs $1 million. A child is dying of cystic fibrosis. Is it ethical to engineer out the cystic fibrosis gene to save the child? Definitely yes, right? Okay, now suppose you are a CF carrier planning to conceive a child. There is a 50-50 chance it has CF. You can either conceive randomly, have a kid with cystic fibrosis, and spend your entire life savings and go deep into debt in order to immediately engineer it out. Or you can do embryo selection for $3000. Either way, you're ending up with a child who doesn't have the gene for cystic fibrosis - you're just wasting an extra $997,000 the first way for no reason)
I am sure your aunt is a very nice person and you are glad she exists. I have a friend with an embryo-selected child. This child seems nice, as far as babies go, and I am also glad she exists. I'm not demanding you kill your aunt any more than you're demanding that I kill my friend's baby. If you accuse me of somehow by-implication thinking of your aunt as "life that doesn't deserve to exist", I will accuse you of somehow by-implication thinking of my friend's baby as "life that doesn't deserve to exist". I think better than that would be for neither of us to accuse the other of anything like that, both agree that existing people can continue to exist (both because they themselves want to exist and because they have many friends and relatives who love them) but keep potential people on the table as a group we can have ethical discussion about without thereby insulting any existing people, who are fine. I think you have to do that to argue about embryo selection at all, but I think once you've done that the argument is strong and we can optimize for things like giving people better lives.
(and it's not academic to me either - my own aunt died of cancer three years ago. I am against this and hope that my descendants don't have to go through this process. I think this is a real harm - in the sense that there will be a real person who dies, and real family members who mourn them - in a way that makes it stronger than the potential harm caused to potential people just because you selected another person before they existed).
First, I appreciate the detailed response. To cut to the chase, of course, I agree that we both agree that existing people can continue to agree. To be more expansive, nobody writes this much in public about these topics, endures the psychic damage of writing in public, then goes back for seconds in the comments because they don't care about people. If nothing else, I have no doubt how much you care, and we are arguing about what is the best way to help people. I also find the idea of complicated philosophical rules to judge each case exhausting and impractical. There are guidelines built on a set of assumptions, and if you're outside that set of assumptions, you have to use your best judgement.
As for keeping potential people on the table we can have ethical discussions about, I'd need to think about it more. The obvious case is homosexuality. If the straights decide no more gay kids, do gays have a right to be angry? Wouldn't your logic dictate they have no right to be angry, and it's on them to make more gay kids? If we get artificial wombs and the gay agenda decides to buy all their supply to flood the world with gays, do the straights have a right to be angry (sorry, this was just an amusing counterfactual to imagine)? Or do we get into a thousand cases of protected traits, lobbying, who gets to decide what to protect, etc.?
It is not my intention to accuse you of killing or judging people walking around, though you are clearly at least judging your own progeny's potential if you are using the service. I think that is the crux of the different perspectives here: you seem to look at the result and if it gives the result you want, the process is judged good, without judging intent, initial conditions, or actions taken. I am not a utilitarian, I lean virtue ethics so the details matter. This is a case of virtues conflicting, which is the main reason I do not like prescriptive morality for these situations because, again, the details matter and there is no substitute for critical thinking.
Consider this also a request for a post about the strength of your conviction in near-term AGI with selecting children for traits that may not show up well past that horizon. I presume it's good insurance if AGI doesn't show up/can't fix them in time?
Hi Scott-In aggregate are you convinced with the IQ screening data of these companies? And if so, would you select Herasight over the competitors?
Of course not advice. Jeff
As I understand it IVF already involves selecting the most-likely-to-implant embryos and discarding others, or using multiple embryos with the understanding that not all of them will implant. This was a relatively common argument in abortion debates.
You're correct this applies in general to ivf. Using multiple embryos seems fine - you're doing the best you can for them to live. Discarding the ones you don't want would not be moral if you view them as having moral standing.
*If* you're using IVF, and already making a bunch of extra embryos, then you have to select which to implant, so it's not any new problem.
But if you were just going to do it the old fashioned way, and now are making a bunch of embryos, you've created a new problem.
I just left this on The Worst Argument in the World, but I'll put it here too:
"This is such a good post that it's worth perfecting. The Martin Luther King example is a bad one. I've never heard anyone use the argument, "But MLK was a criminal", meaning he had been in jail. On the other hand, frequently people say, "MLK was a plagiarist" or "MLK was a philanderer", and we'll be seeing more of "MLK laughed as he watched rapes". Those are a different fallacy, ad hominem, if the subject is, say, whether MLK was brave. We don't want to confuse that with the noncentral fallacy, which sounds a lot like it but is distinct. " https://www.lesswrong.com/posts/yCWPkLi8wJvewPbEp/the-noncentral-fallacy-the-worst-argument-in-the-world
It is at least as eugenicist for the government to ban this. It is a declaration that it would be more optimal for some people to have Down syndrome or cystic fibrosis or lower IQ or whatever, when they otherwise wouldn't in a world where such therapies were available.
In fact, it is even more pernicious than the alternative, as it requires the state to restrict choice and bodily autonomy. In this sense, restrictions on gene therapy have more in common with the forced sterilizations associated with eugenics than anything else.
eugenics is incredibly obviously good if and only if in the specific form of “open free genehacking clinics on every street corner” where “free” means “it costs zero” and “anyone can enter there and ask for a genetic mod that their geneline will carry henceforth, selected on a catalog of every trait we know how to mod or do iterated embryo selection for”
inb4 “but not everybody would want to use them and then the people who haven’t been genehacked will be unfairly disadvantaged”: oh, so you want Fairness? ok, then sterilize everyone so that nobody will make kids without having thought at least once in their life about making kids or had an opportunity to choose which traits their kids will have
inb4 “but then it will become necessary to have certain genehacks to access certain things” yeah so? right now it is necessary to have certain partially-genetically-determined capabilities to do things, too; at least this scheme permits choosing
inb4 “but then there won’t be disabled people anymore”: Deaf people often want their kids to be Deaf. aspie people often want their kids to be aspie. there are even people who want kids with Down’s, and when i say EVERY TRAIT we know how to mod, I MEAN IT
inb4 “making fertility opt-in would enrage the trads who want people who fuck to be punished with kids they don’t want”: that is a feature, not a bug
inb4 “introducing a trivial inconvenience to fertility would prevent so much births that there wouldn’t be enough people to support the olds”: i’m counting on there being enough people who’d choose to mod for additional intelligence that they’d make fully automated luxury gay space communism happen
any other objection?
The argument against eugenics at the embryonic stage is that it reduces natural diversity due to groupthink about e.g. IQ, which is disadvantageous if it turns out that the narrower set of genetics you are producing has significant downsides or just isn't optimal compared to a wider set. Could be due to you not actually knowing downsides/optimality or due to unforseen changing of the environment. There are some conditions where it seems clear that there is no downside to eliminating them, and as long as the effect isn't significant at a population level (only the rich do it) then who cares, but when you are doing positive things like maxing out a life expectancy or IQ you do risk this sort of thing.
Here the argument of the piece was incredibly weak:
"When I think about these concerns, I consider the following thought experiment: suppose that I had a natural, unselected child, and that child became high school valedictorian and got into Harvard. Would my first reaction be “Oh no! This slightly raises her risk of anorexia!”?"
How about this, you gene select a child, they develop anorexia and drop out of school, would your first reaction be "awesome, that meant on expectation I made the best choice, just need more kids!!"
How is this different from environmental interventions?
That is, maybe teaching your kid to read is bad, since it decreases diversity, since otherwise there would be both literate and illiterate children, and maybe illiteracy is good in some way.
And also, if you taught your kid to read, and then they read a cult pamphlet and joined a cult and dropped out of school, would your first reaction be "awesome, teaching them to read was still a good idea in expectation!"
I think this is one of those cases where our common sense intuitions are good, and we use them for everything in near mode, and then we use a totally different set of crazy intuitions for things like new technology, where maybe all good things are bad and second-order effects dominate first-order ones. I think we should instead use our common-sense intuitions for the new technologies too.
In general why is modifying the genes of embryos different from environmental interventions? Inheritance to future generations and lack of the plasticity/changeability of the mind. A closer analogy would be environmental interventions on infants vs the elderly, higher standards of risk should be held for infants of course.
Yeah we should do good things, some of it is clearly good mainly because it is preventing obvious negative outcomes. The risk/reward for “IQ” modifications is less obvious to me and more likely to change with societal values and so on. You can take all the GMO monoculture ag arguments here as well.
Because we're going to have a fuckton of anorexics before we iron out the bugs for "gimme the higher IQ".
Beautiful theory, but nature does what it wants in practice, and in this case "in practice" means "a lot of experimental babies born so we can see where we're screwing up" but unlike lab rats, it's not (yet) permissible to euthanise the failed results.
The way it’s different from teaching your kid to read or filling the house with beautiful educational toys is that the benefits of these environmental goodies are not heritable. I don’t think selecting embryos for valued traits is ethically wrong, but I do wonder whether it might have unintended negative effects if it becomes common practice.
You know how you don’t like having visible upvotes in the comments? Well, I think selecting for traits is sort of like having visible upvotes.
Teaching kids to read and other educational interventions have had a profound effect that persists across generations and reshapes culture. Children taught to read will eventually teach their own children to read; better educated children tend to educate their own children as well. An educated population has a huge effect on society and could have unintended consequences.
I think you have a valid concern, but it doesn't come from a qualitative difference between embryo selection and other interventions. Rather it's about us having too much power to shape future generations. We would have the same worries if we invented some other way to radically change the personality or physiology (eg, some kind of incredibly effective 12 week training program). People would use this new training program to make their kids like most popular celebrity, reducing diversity in personality. Or maybe they'd use it to make their kids hardcore Republican/Democrat supporters. These kids would grow up and use the same intervention on their kids, perpetuating the change in society.
We want to have some power to affect the lives of our children. But too much power can be scary because it allows our mistakes to have permanent profound effects. New technologies are scary because they give us more power. It doesn't matter whether it's embryo selection or some almost magically effective training program; what matters is how powerful the technique is.
Yes, great point. To be perfectly honest, the thing I'm most concerned about when it comes to selecting for intelligence is that there is a certain kind of high intelligence that goes with being socially awkward, introverted, having trouble reading people and empathizing, inexpressiveness, fascination with patterns, etc. Of course many brilliant people do not have that autism lite thing going on, but what if the genes that get identified as high IQ genes are the ones that tend to produce the Aspergerish form of brilliance?
I am very very fond of quite a number of people of that kind, everybody from Oliver Sacks to many people here. I'm all for having more people who are that brilliant. At the same time, it doesn't seem like a plus for society to increase the number of people who are non-social, low-empathy, etc. It seems like more of what we *don't* need. We already have an ever-increasing portion of the "people" part of our lives filled with internet contact, fictional people in games, etc. and of course AI, which can already ape affection and admiration (though not very convincingly), and will soon learn to ape love and lust. How can it be good to increase the fraction of the population that is not terribly interested in other people?
Probably because of issues seen in animal breeding, where some genetic combinations that can get you to a desired phenotype are associated with genetic disorders, and the correlations wasn’t caught at first. For example, HYPP in Quarter Horses. https://www.tsln.com/news/hypp-and-the-impressive-line-of-american-quarter-horses-the-facts/ Essentially the disorder contributes to the desired phenotype.
“This constant flexing of the muscle cells is what leads to the pronounced muscle patterns on the afflicted horses, hence the popularity of the line in the stock horse breeds.”
I have heard concerns in human selection that some (not all) genes associated with high IQ are also associated with autism. https://pmc.ncbi.nlm.nih.gov/articles/PMC4927579/
Abstract is as follows,
“A suite of recent studies has reported positive genetic correlations between autism risk and measures of mental ability. These findings indicate that alleles for autism overlap broadly with alleles for high intelligence, which appears paradoxical given that autism is characterized, overall, by below-average IQ. This paradox can be resolved under the hypothesis that autism etiology commonly involves enhanced, but imbalanced, components of intelligence. This hypothesis is supported by convergent evidence showing that autism and high IQ share a diverse set of convergent correlates, including large brain size, fast brain growth, increased sensory and visual-spatial abilities, enhanced synaptic functions, increased attentional focus, high socioeconomic status, more deliberative decision-making, profession and occupational interests in engineering and physical sciences, and high levels of positive assortative mating. These findings help to provide an evolutionary basis to understanding autism risk as underlain in part by dysregulation of intelligence, a core human-specific adaptation. In turn, integration of studies on intelligence with studies of autism should provide novel insights into the neurological and genetic causes of high mental abilities, with important implications for cognitive enhancement, artificial intelligence, the relationship of autism with schizophrenia, and the treatment of both autism and intellectual disability.”
Broadly speaking, your geniuses have really really weird bodies, chemically speaking. Not terribly habitable for your normal infectious parasites.
Conversely, selecting against autism could be selecting against certain kinds of intelligence. Ideally these cancel each other out if you do both kinds of selection, so we neither end up with way too autistic people, nor way too dumb people, and I do expect the technology to have way more upsides than downsides, but we should be careful.
Except that’s not, in general, how it’s being marketed.
Right now we don’t have enough data to distinguish between IQ genes that increase autism risk and those that don’t.
And someone who is selecting for a high IQ baby probably is also someone who does not want to accept increased risk of autism.
The difference, very stark, is here you are manufacturing babies according to specifications. Treating human life as industrial product would lead to collapse of whatever "dignity of rational animal" that is left.
The specifications come from the combined individual decisions of a lot of separate humans trying to do something good. It is no different.
The animal breeders who care enough to actually evaluate the animals on these traits and pick the best, those people, they almost all genuinely care about the animals and want to breed them to be as good as possible. The people who do not care about their animals, puppy mills etc, also do not care about selecting their animals to have better traits and will just breed and sell as many as possible.
Take rabbits for example. Breeders of pet rabbits who want to improve the breed generally want to make them friendlier and cuter. And due to human perception, having a round face with a short nose is cuter. So breeders choose bunnies with round faces and short noses, and they are indeed very cute, but this is how Holland lops end up with tooth issues, because their mouths are not designed to fit into such a small space. Now, there is another combination of individual decisions breeders make - to compare their animals to those of other breeders in shows. Without this, the selection for shorter noses probably would not have gotten to such an extreme that it has a noticeable impact on health. Compare for instance Dutch rabbits, which do have shorter noses but not to such a ridiculous extreme.
With that background information out of the way, what do we make an analogy to? I would say the genetic testing companies are like the rabbit shows. Both have a very limited view of the situation - genetic testing companies can only see through the filter of a few genes with known effects, while judges at rabbit shows only spend a few minutes with each rabbit and do not get a chance to evaluate complex traits like personality or health. And both use that imperfect lens to decide which genes should be passed to the next generation.
(That being said, I'm still all in favor of selecting against gene variants with known highly negative effects. It's the positive selection that is easier to mess up, and even then, you can't do much harm within a single generation and by the time of the next one we'll have much better research.)
Gian, human dignity has gone out the window, only dumb old believers in the sky-fairy even hang on to that concept any more, and you don't want to be one of the dumb old sky-fairy believers, now do you? That would hold you back from all the fun things we all want to do!
So does selecting your mate. Should we just force a random function on all breading habits? Use your brain.
Robin Hanson might unironically say you have a point
https://www.overcomingbias.com/p/beware-moral-confidence
https://www.overcomingbias.com/p/libertarianism-as-deep-multiculturalism
This isn’t really a new technology concern. The issues being raised are as old as deliberate animal breeding. Selection sometimes pulls along things other than what you are intentionally selecting for.
Common sense intuitions from animal breeders would be that avoiding deleterious mutations is pretty safe, and positive selection of desired traits needs to consider the whole organism, not just individual measurements.
For example, some high IQ genes appear to also increase autism risk. In horses, selection for muscling resulted in inadvertent selection for HYPP, a genetic defect. Just looking at one trait can lead you astray.
The diversity discussed is GENETIC diversity, or range of options in the gene pool. Various versions of genes are called alleles. Literacy is not an allele. Alleles can provide immunity to a specific disease or provide other adaptive advantages.
You can see the genetic marks of old plagues in populations from the increased frequency of protective alleles to those diseases.
But this is not a real objection, unless you also object to the idea of people selecting their individual sexual partners. Characteristics like height, physical fitness, intelligence, lack of deformities, et al. are attractive precisely because they predict fitness in offspring. It's a cruder method, but we still practice it. Your logic would have us select sexual partners at random, just in case we were discounting some hidden yet important gene - though a gene pertaining especially to short, unfit, unintelligent people is unlikely to be particularly salubrious.
Choosing a partner allows for observation of the outcome rather than prediction. And after partner-choosing there is the randomness of inheritance that we are discussing.
But the big one is that we cannot all choose the same partner. If you proposed that every child was fathered by either the Super Bowl MVP or a Nobel laureate, I would also object that we are losing genetic diversity. The current system does indeed allow for many relatively short fat stupid people to procreate and it’s not necessarily a bad thing.
Are you saying that people's choice in partners is not based in selection? That they do not select for traits that predict selective advantage for their offspring? I'm afraid I cannot agree with this. You discount an entire field of evolutionary theory.
Inheritance may be somewhat at random, but in any case the difference in question is one of degree, not function. It is not *entirely* at random, because observations concerning genetics become more accurate at the general, not the specific, level. Once we are talking about the group level, genetic prediction can be extremely accurate. And at the individual level, one has only to go back a few generations to make fairly accurate predictions about a given person's traits. This is necessary in animal breeding, for example, and humans do not function differently on a biological level.
What inherent advantage do you believe random 'diversity' to possess? A proliferation of disadvantageous traits may be legal, even morally acceptable, and still have deleterious effects. Because we do not force people into relationships, humans naturally have a tendency to pair themselves with partners whose traits are the best they can manage to obtain. No one is actually choosing short, stupid, fat people as a primary option in our society because they understand genetically on some level that these are traits which are disadvantageous. The consequences of sexual selection have much more to do with the genetic makeup of population groups than random drift. This was well established by Fisher (The Genetical Theory of Natural Selection).
Whatever process affects the genetic makeup of the population is "selection". It's a tautological concept. We are moving on the contiuum between "natural" selection where that process is mostly determined by "nature" meaning physical environmental factors and access to mates/survivability, through selection due to societal-psychological factors and associative mating based on "who do I think has the best traits", into a potential territory where genetics might be shaped by intellectual intent and design.
But yes, eugenics based not on embryo selection or gene-level mapping/modeling/prediction, but rather on intellectually intentional mating practices, espeically when applied at the population level, also seems bad. Of course that is the traditional way to do it and it's probably much more powerful at this point in the technology. It seems that most in this discussion want to run away from connecting this treatment with "breeding humans" so interesting that you embrace it. Producing the human equivalent of the English bulldog is exactly the fear.
Genetic diversity in a gene pool is strongly correlated with the health of that gene pool and its odds of long-term survival.
Pretty good discussion at https://www.usgs.gov/news/why-genetic-diversity-important
All breading habits across all mammals are not random at all, they exist specifically to improving genetic outcomes. Just most people are bottom feeders so they get the best they can.
we aren't selecting partners eugenically, though. It's individual love. We'd mock anyone who did if they'd admit it, because its dehumanizing.
"I adore Jeremy. He has such beautiful eyes!"
"I couldn't possibly date Steve; he's cross-eyed and short."
we call this "being shallow" in Narnia.
dehumanizing is reducing people to traits like a pair of pretty eyes.
"making fertility opt-in would enrage the trads who want people who fuck to be punished with kids they don’t want”
Ah, *that's* what I want, and not because I think abortion is murder! Well, nice that you know what I believe better than I do, thanks for opening my eyes!
We don’t want people who fuck to be punished, retard; they punish themselves by destroying their own (and the rest of our) human integrity by enjoying sex as if it were no different from going out for a walk in the park. The ramifications of disociating body, reproduction and soul will be (and is proving to be) disastrous for society. This entire thread is genuinely convinced IVF is good and will have no consequences on furthering our dehuminization ontop of the fact that it’s involves another attempt to ratify the moral loophole of murder that is embrionic killing (just like with abortion).
If you're an example of someone who hasn't been "dehumanized", I'd hate to see what less human people talk like.
Maybe get down from your high horse and take a look in the mirror before talking to me from your false virtue. I may be using ugly words to insult you but only as a response to your attempt at defaming an entire group of people by deliberately simplifying a complex topic to misrepresent their intentions. You’re just another wolf in sheep’s clothing.
Whatever you say, extremely weird person who thinks casual sex destroys human integrity and clumps of cells are people.
Who said anything about a moral loophole? Some people have no moral issue with the elimination of an embryo. That would make them morally consistent.
But of course, you mean that *your* morality is the correct one. That's an argument that I have no interest in pursuing.
Blurring the intrinsic value of human life at the embryonic stage is a loophole to morally justify killing that person before their personhood is directly evident.
See, there you go again: using phrases like "the intrinsic value of human life." (What value? How intrinsic? Whose human?)
Not everyone agrees with your moral perspective. That doesn't make their arguments "moral loopholes" - it makes them at variance with yours.
We're stuck with questions of morality, though, because this opens the door to the dream of social engineering.
We want the future generations to be the best. But what is the best? Well, that they'll have the qualities we value and the values we think correct.
But it's hard to forecast what the society of fifty to a hundred years hence could or will be like. If we had this tech in the 1950s, the qualities and values that were self-evidently correct would be "now there will be no more perversion, we can do away with homosexuality". Today, we have Pride parades.
What will be the Pride parades of 2075? What will be the optimum human for those days? We need to be a little more humble before deciding the future is blue-eyed blondes (that didn't go so well the last time it was tried) and engineering our children to fit.
a) That was pointlessly cruel
b) If you look at Deseach's comments, you will find that you have just insulted an ally or near-ally.
You’re right, i jumped the gun. I’m the retarded one.
Many Thanks! Best wishes.
Okay, so, one of the standard "proofs" from the pro-abortion side that the pro-life side's revealed preferences don't match their rhetoric is that pro-lifers rarely protest IVF (the ultimate topic at hand), even though it creates 5-10x as many aborted viable embryos as a single abortion.
I ask this out of genuine curiosity, because I assume you've at least thought about this, what is your personal opinion there?
Watch me smile like a shark as I pull my Roman Catholicism card out of my sleeve.
Got you covered right there, fam. We don't approve of IVF (and indeed, I caused personal pain to another former commenter on here, which I do regret, by expressing that in a previous debate about the topic).
https://www.usccb.org/issues-and-action/human-life-and-dignity/reproductive-technology/begotten-not-made-a-catholic-view-of-reproductive-technology
Well, I can appreciate the moral consistency.
I'm still curious about your opinions on the magnitude of it. Would you say it is as bad as 5 to 10 "regular" abortions, or is there some sort of mitigating discount rate?
Or is comparing the magnitude of sins at all a fool's errand? I don't recall seeing whether you've ever expressed an opinion on utilitarianism in general.
Edit: Also, is any of this affected by the previous pope's declaration that unborn children can go to heaven instead of limbo?
>Or is comparing the magnitude of sins at all a fool's errand?
Wild guess (I find deities implausible), but I think most followers of many religions put that judgement in their deity's hands.
How long is a piece of string? Is one torture-rape-kill murder equivalent to five or to ten ordinary one-punch murders?
It's all bad.
Now, there is a distinction between venial and mortal sins, which is a whole discussion of theology on its own, but I'll just drop a quick quote from "The Screwtape Letters":
"You will say that these are very small sins; and doubtless, like all young tempters, you are anxious to be able to report spectacular wickedness. But do remember, the only thing that matters is the extent to which you separate the man from the Enemy.
It does not matter how small the sins are provided that their cumulative effect is to edge the man away from the Light and out into the Nothing. Murder is no better than cards if cards can do the trick. Indeed the safest road to Hell is the gradual one — the gentle slope, soft underfoot, without sudden turnings, without milestones, without signposts".
Deiseach you are truly irrepressible! I'm glad our faith has a representative like you here in the commentariat.
God help us all if I'm taken as a representative of Catholicism, except maybe in the "see, we're not an organised religion!" sense.
Well, we are all sick and in need of the heavenly physician, I suppose.
One possibility is that a small amount of a trait is good, but efforts to eliminate the bad, severe cases are bad. A little bipolar gets you artists, severe bipolar ruins lives.
A possibly hard case-- there's a gene which, when homozygous, gives immunity to AIDS. The piece of the immune system that AIDS hooks on to just isn't there. This means the person is a little less healthy in general. Worth it? Worth it going forward?
Good example!
If I were expecting highly effective embryo selection to happen instantly (and I were young enough to live to see the effects), I'd worry about an "arms race" on traits like height (a race to the top? :-) ) which is, to an extent, zero-sum. As it is, I think too much else (AI, of course) will happen first.
The only acceptable answer to "But then there won't be disabled people anymore" is "Good." I can think of few better indictments of libertarianism than the idea that people should be able to choose to make their kids disabled on purpose.
That sounds uncharitable to me. If we have the technical ability to prevent certain traits that we might be considered disabilities, then *someone* has to decide which such traits (if any) are actually disabilities that should be prevented. I'm not sure that parents are always the right someones, but it's not obvious to me that government is the right someone, either.
That goes double if we can't be sure of side-effects — where we're not just preventing a putative disability but also imposing unknown and potentially unknowable risks. (E.g., maybe it'll turn out that knocking out a certain gene eliminates the risk of nearsightedness but doubles the risk of breast cancer.) *Someone* has to make that call.
It doesn't. But it's positive, not negative eugenics
It’s absolutely eugenics.
They are just going to redefine what that word means - you know that, right?
It's fairly straightforward. The word is derived from ancient Greek: eu - genes, meaning "good" and "born, come into being."
Not much to define. Unless you're talking about negative attributions which have nothing to do with actual definitions, of course.
That's not how words work. If it were, then "jaunty" and "gentle" would be synonyms (both from French "gentil"), as would "isolate" and "insulate", "diamond" and "adamant", "metal" and "mettle", etc., etc., etc.
I see. What's your definition of eugenics, if not "the process of being born well (i.e. genetic improvement)" ?
Merriam-Webster defines it as "the practice or advocacy of controlled selective breeding of human populations (as by sterilization) to improve the populations' genetic composition" (https://www.merriam-webster.com/dictionary/eugenics), which sounds about right to me.
Note that it specifies several elements not clear from the Greek roots, e.g. "human" (and not, say, selecting for meatier chickens) and "populations" (and not, say, an individual choosing not to have biological kids because they have a condition they don't want to risk passing on).
I'm not sure whether population-scale embryo selection would count as "selective breeding" for purposes of this definition, if every couple is allowed to procreate but not to choose which of their embryos is used. That technology wasn't available when the eugenics movement was in ascendance, so it didn't come up; we'll have to see if the term ends up including it. (Cf. "phone"; if you could explain all the different kinds of 2025 computer to someone living in 1940, I don't think they'd guess that any of them would be called "phones", let alone exactly which ones would and wouldn't.)
Apparently by your definition so is masturbation 2 days before trying to inceminate lol. I'm euginically eliminating the most aged sperm in favor of a fresher sample.
We're about to see some weird generational cycles.
Every generation one trait will get optimized to the point of oversaturation, and the next generation will find its comparative advantage by selecting for a different mix of traits.
Strong disagree.
If you mean "saturated" in the sense of up to the theoretical maximum, that means nobody ever gets cancer, heart disease, or obesity, and everyone's a genius. Why would you select for a different mix of traits than that next time?
If you mean "saturated" in the sense of up to some current level of technology, the current level of technology is changing much faster than the generational cycle. I expect genetic technologies to be much better than this within ten years.
I mean in things like height vs. IQ Vs. schizoid tendencies.
I know they're not up to there yet.
I'm mostly saying that I think in a world of designer babies, there won't be a static target.
One can expect that "compatibility with cybernetic implants" will be a goal too once they become mainstream.
If everyone's a genius, no-one is.
Make it so everyone gets that 6 point IQ boost, and now the 100 IQ people are 106 IQ and the 120 IQ people are 126 IQ. That sounds great, but now we have to renorm all the tests so now the 106ers are the new 100 IQ.
It might be a better world. Or it might be just more of the same, where the rich and smart get richer and the rest of us pick up the crumbs.
IIUC, the extremely rich are not generally those with extremely high IQ. It's more the "moderately rich". The extremely rich are generally those who start off rich and focus on getting richer. They may have only an average IQ, but they hire people to do things that require such a characteristic.
I think that's pretty obviously an all-boats-lifted scenario. Sure, people's individual experience of whether society treats them as "smart" is just going to do the treadmill thing you're describing, and I certainly wouldn't expect it to particularly help inequality. But, pretty much every job on the planet would get performed just a little bit better here and there.
Yeah, but we're already performing "a little bit better" compared to previous generations and eras, and we're still complaining.
The smarter world will do a little bit better than us, but that will be the new normal, so now people will complain about how it takes a whole two minutes for their customer service problems to be solved.
Are we having better lives than Homo erectus? 100=>106 is just that difference, but much smaller in size.
I recall that Damon Knight once wrote a story where this happened. One character mentions that having beautiful children is like playing the stock market, you have to guess how beauty standards will change in the future.
Imagine if we'd had this technology in the 80s and 90s. It would be a net win for society because of increased intelligence and decreased mental illness. But you'd have to take a moment to pity all the poor white women whose parents selected them to have small butts!
I hope someone can find the name of the story. It sounds more like Silverberg to me, but I'm guessing.
By any chance, John Brunner's The Stone That Never Came Down?
IIUC, that's already happened in China, where people selected for male offspring. (Perhaps that's happened elsewhere, too.) So far, however, the number who could afford the treatment has been a small enough proportion to avoid most of the bad effects.
I'm type 1, not type 2, but... I'd have to think about it, at $27,000? I mean, it's probably worth it, but I might put it off a few years.
Huh, if you don't mind me asking, what is your income/net worth? And how much do you + your insurance spend on diabetes care per year?
I don't have super consistent income yet and I could probably pull together $27K for something really important but it wouldn't be, like, remotely easy. My parents mostly pay my medical expenses but my guess is the copays are probably something on the order of $1000 per year, and the vast majority of that is continuous glucose monitors and insulin pumps (which aren't strictly necessary), the insulin itself is like $10 for a 90 day supply I think? No idea what the insurance is company is paying.
Okay, that makes sense. In the post, I said the claim required that "you’re well-off enough that these aren’t impossible sums of money for you", and I was referring to people with Type 2 who tend to be in their 40s and 50s and so probably richer than you. I'll edit "impossible" to "extraordinary".
Strongly agree. I have Type 1 as well, but am in my 40’s and fairly well off. I’d pay >>$100k to get rid of it. Not pay excruciating attention to how much I’m eating would be amazing.
FWIW, I'm type 2 diabetic, but I control it by diet so officially I'm "not currently diabetic" (which is because of a silly definition, not because of remission).
I might pay $5,000 for a complete cure. (If you'd asked me a decade ago, I would probably have been willing to pay more. If you'd asked me in the first year after diagnosis, I'd probably have agreed to the $27,000.)
I think that's fair. If you had gotten the cure for $27k as soon as you were diagnosed you would have saved yourself a huge amount of effort, time, and cost learning and working to manage the condition. Now, you would benefit from the instant cure but you'd never get that time and effort back.
The good news is that I don’t think your 50k number is correct anymore from Herasight. They might reach out and correct you, and we might see just how quickly capitalism brings down costs for technologies like this
I asked Herasight the day before yesterday. They said it was still their most recent number, but that they hope to get it lower in the near future, and that people with unusual situations could ask them for a custom quote that might be less.
Oh, I should probably tell you something: type 1 diabetes is one of the most tractable conditions with gene editing. We can lower the risk by >50% with a few edits. It's actually quite unusual in this way. Conditions like schizophrenia or ADHD involve far more genes with much smaller effects.
Not so with type 1.
And because type 1 has such an earlier age of onset compared with other "highly editable" conditions like Alzheimer's or breast cancer, it's a strong candidate to be one of the first conditions we edit for. We'll be able to demonstrate within ~5 years or so that it actually works.
I think early editing tech is probably going to be roughly on par with the cost of the more expensive embryo selection companies (possibly a bit cheaper), so it's possible cost will still be an issue.
But it's worth noting you can at least lower the cost of egg retrieval by doing it abroad. You can do egg retrieval in the Czech Republic for $3000-$4000, in Spain for $6000, and in Vietnam for $2000. And while I can't really vouch for the quality of specific clinics, the expected cost per child is almost certainly lower than for US clinics.
If you're interested in talking about embryo editing for type one, drop me an email at genesmithlesswrong@gmail.com
Scott, you say: "rich people get first crack at every new technology, but poor people usually follow close behind... Also, it would be crazy for any forward-thinking government not to cover this."
That's reassuring for the classic concern that this tech could, over generations, create genetically-stratified socioeconomic classes *within* a country.
But across countries, still seems like things are gonna get really weird, like "ok now there are *actually* significant genetic differences in IQ across countries". Seems like an important downside.
Especially because presumably this wouldn't work on unplanned pregnancies, which are strongly stratified by socioeconomic class.
You would only be one generation "behind" the rest of your country if you had an unselected pregnancy. Probably a wash if we're looking at the impact over multiple generations.
I'm actually a bit more optimistic because there are so many interventions left in developing countries that could raise IQ significantly more cost-effectively than embryo selection. The $/IQ point cost effectiveness for LEAP is orders of magnitude more than that of embryo selection. So I would actually expect to see the gap close in coming years, even if embryo selection is widely adopted.
In the long run, my hope is developing countries can just get a lot richer, and IVF + embryo selection can get cheaper. At some point those two lines will cross. Our job is to increase the slope of both curves.
Unlike wealthy countries, where correlation between fertility and IQ is zero or weak-negative, it's very negative in many poor countries and their brightest are likely to emigrate to rich countries; as environmental gap shrinks, genetic grows...
What's LEAP?
Maybe they meant LEEP, the Lead Exposure Elimination Project: https://forum.effectivealtruism.org/posts/fd96FtLFACeAshqJP/introducing-leep-lead-exposure-elimination-project
Sorry, I meant to say LEEP, the lead exposure elimination project
I am mostly of the opinion that a scenario where country A has IQ 100 and country B has IQ 120 is worse than a scenario where they're both 120 but better than a scenario where they're both 100. I think equality is nice, but not so nice that you should burn down all the houses to give everyone equally good tents.
I do have some concerns that relative advantages can be exploited to get away with evil crap, but we already have some pretty big relative advantages along other axes and the results have not so far convinced me that it's worthwhile to outright destroy value just for more equality, except in some special cases like literal weapons (e.g. treaties that put limits on how many nukes each country can have).
I think there's a somewhat wider range of scenarios where it's good to _redistribute_ some of the value (e.g. progressive tax), compared to destroying it. So hypothetically maybe it would be good if there were a global policy of higher-IQ countries paying some aid to lower-IQ countries.
("But they'd never agree to that!" "Oh, but you think they'd agree to voluntarily lowering their IQs instead?")
Agreed
All I can say is if I was inclined to do this sort of thing and I already had kids I'd be asking the company to test the existing kids too to calibrate how seriously to take the predictions.
I don't think you can get that much from this.
First, your kids probably don't have prostate cancer yet. And child IQ is pretty weakly correlated with adult IQ, and the predictors are for the latter.
Second, it's hard to get good intuitions for statistics, but I think even a somewhat valuable predictor can have a high error rate. If I'm reading the widget on Herasight's site correctly, if you have five embryos and use their IQ predictor with supposed +6 IQ point gains, there's still only a 73% chance that your child born using their predictor will be smarter than an average child born without it. So if you "test" their product and find that actually your unselected kid is smarter than your selected kid, this doesn't prove very much in a scientific sense - it's the chance of two coin flips both landing heads.
Maybe a better tactic would be for them to test you, the adult (since you might have gotten some diseases, and you know your adult IQ). Then you only have the latter problem (sample size of one). So you should get them to test lots of different adults. But then you're back at "this works in large studies", which is how they validated them in the first place!
I've had Nucleus and Herasight genetically test me. There wasn't very much useful on there. For example, I have obsessive compulsive disorder. Nucleus said that I had slightly-higher-than-average genetic risk of OCD, which put me at 2.6% compared to the overall population's 2.3%. Is this good (because it identified higher-than-average risk)? Is it bad (because it gave me only 2.6% chance of a condition which I, in fact, have)? I'm not sure, and I think the real answer is to not pretend that you can conclude much from a single data point in a heavily probabilistic domain.
Herasight did also test my two existing children, but I found I couldn't bear to look at the results. We agreed that my wife would look at the results and agreed to let me know if there was anything actionable. But AFAIK they used the information internally to get a better picture of my and my wife's DNA to predict our future children, but forgot to send it to us. I plan to remind them about it when I get around to it.
Get them to test you vs your adult siblings? lol
This whole thing really smells of “amulet to defend against leopards” - there are no leopards, so clearly it works!
It’s functionally impossible for me to confirm the service achieves what it says it does, and even in the best possible world the effects are so tiny I am unlikely to see them; and when I don’t, well, what did I expect? - the dice don’t often roll all sixes, guess I just got unlucky, LOL cha-ching.
Meanwhile, their margins will be even better if instead of doing any work they just hook an RNG up to the numbers in their report - you can’t even start to check the statistics for a couple of decades, while the half life of a startup is measured in single digit years! - and even if you literally go to their lab and look over their shoulder, since in the long run Moloch says everything that can be enshittified will be enshittified, the fact they had not yet done this when you checked has only a limited bearing on how much I can trust them by the time I get around to using the service. Net result: a generation of Silicon Valley neophiles using their cellphones to call the local shaman to help pick embryos.
Imagine, ten years from now, there are lots of these startups and also just as many of the sort that just make up numbers, and everything in between. How do you tell which one to go with, or where on that spectrum any particular one lies, given only their marketing material?
The situation feels somewhere between buying a lottery ticket and one of those deeds for land on the moon: there is some tiny chance it might pay off, but realistically, you’ve probably spent your money on a worthless bit of paper.
When does IQ change? Turn from "child" to "adult"?
Exactly on one's 18th birthday, of course. Unless you live in Alabama, then it is on the 19th birthday.
This sounds like the perfect opportunity for scammers. After all, how do you prove a negative? "Your child hasn't got prostate cancer yet". Yeah, and by the time they're in the age bracket to get it, they either do get it or they don't.
If they don't, the scammer company (which did not, in fact, select out embryos) can say "Told you it works". If they do, well, that's forty-plus years in the future. The company may not still exist. Or they can pull statistical fine print out of the air to say "Ah, we never said they'd *never* get prostate cancer, we just said this would reduce the risk. Not our fault some risk remained and they got unlucky!"
You make my stony black heart want to set up a dodgy clinic promising the gullible and desperate that I'll give them superbabies, when it's bog-standard IVF, I pocket the dough and flee to Switzerland, and it's their problem if the kids get health difficulties etc. in later life.
That's a real problem. It's a real technology, that really ought to work if done properly, but it's a lot easier to lie, and it may take multiple decades for your lies to be proven. And when the lie *is* proven, most of the evidence for "who was responsible" will be gone, so the court case may be impossible. (Or, by then, the scammer may be dead of old age.)
> that really ought to work if done properly
...where by "work" we mean if everything goes /perfectly/ we see a single digit percentage effect in statistics across the entire population measured several decades later, rather than anything specific in particular that any person actually paying for the service could point to.
Don't even need deliberate lying, as of yet there's a lot of room for "oops, didn't work out as we thought it would, genuine error, well them's the breaks, that's the way the cookie crumbles, hey at least your baby has the right number of eyes and hands, okay?"
The "oops!" is a different problem. Yeah, it's a real one, but it's a different one. (Well, except that with over-promising they blend into each other around the edges.)
Whether they select or not, very ease to prove, because they have to give customers these genomes data and after child is born you can check if child's genome matches the selected one.
And what are you gonna do if the baby doesn't match up? "Excuse me, I wish to return this deficient item, here's the receipt"?
Sure, this is what American law courts are for, but good luck arguing statistical probability there against company lawyers who are pleading "you shoulda read the fine print, there's no guarantees in biology".
Are we talking about the prostate cancer that supposedly most men develop if they live long enough?
“ Herasight did also test my two existing children, but I found I couldn't bear to look at the results.”
It would be worthwhile exploring why this is so.
(Just using 2.6% chance as a generic chance and anorexia as a generic condition) Is it because
- if you found out a child had a 2.6% chance of anorexia, you regret not having had the chance to lower it to 2.3%
- if you found out an existing child had a 2.6% chance of anorexia, it will have been in some sense unfair on them that your third child has a lower chance
- you will regret knowing that your existing child has a 2.6% chance of anorexia versus 2.3% chance on average and having to deal with the consequences of anorexia in the future
- you will regret knowing one existing child has a 2.6% chance of getting anorexia and the other has only a 2.3% chance and that will influence how you treat your two children
I can definitely imagine ‘not being able to look’ but not having kids myself I’m not sure why it is.
"Herasight did also test my two existing children, but I found I couldn't bear to look at the results."
Presumptuous of me to say this, but I am very glad you didn't. You acted like a father, not someone purchasing from a breeder who wants to check the pups are indeed the pedigree claimed.
Super interesting developments, somehow, after a slow start, the 21st century does pick up steam after all.
But with regard to
> To have a sense of why society works the way it does, instead of feeling like you’re being blown back and forth by institutions you don’t really understand?
Would not institutions become more complex at the same rate that general intelligence increases?
Good question. I don't know. Do countries with very high IQ (like China) have noticeably more complicated institutions than those with very low IQ? I'm not sure (and insofar as they do, I'm not sure if it has more to do with IQ or wealth or stability). See also my recent https://www.astralcodexten.com/p/book-review-arguments-about-aborigines post for some discussion of whether "primitive" cultures have more or less complex institutions.
Yeah that's a good point, maybe it's more about a "ceiling" of institutional complexity than about the average institution.
For the Aborigines, I wonder if they could have managed to maintain an even more complicated section system if they had an average IQ of 120 or so.
Well, the Pirahã of the Amazon don't have words for numbers, and they have no abstract terms for colors (they'd describe something red as "like blood"). Yet they seem to have words for every plant and animal species in their environment with the uses and dangers of each (if any). Considering the biodiversity of the Amazonian rain forest, that's quite an accomplishment. Probably they'd have trouble with Raven's Progressive Matrices test because they don't really have words for shapes, and they're not trained to recognize the abstractions (in fact, John C. Raven recommended that his test wouldn't be suitable for populations without formal systems of schooling—can't find the quote now, though).
Are you saying they have words of hundreds of thousands species, including those that require a microscope to tell them apart?
There's plenty of low IQ populations under obligatory schooling systems.
No. But they're foragers, and like any good naturalist, they can recognize all the plants and animals they encounter in their environment. Upon further digging, their designations seem to be general descriptions—e.g., "the plant whose bark gives visions." This is similar to the Inuit "words" for snow, which are really agglutinative descriptions, e.g., "snow that is slushy and granular".
Identifying all the plants in our environment without the help of a phone app is more than most of us can do. "Oh, that's some sort of tree. I think it's an oak."
I think you have to have a high IQ to be able to learn to read and write in a system with no alphabet and thousands of possible grammatically valid characters.
Don’t assume it, come and look! The answer is: lots of very average and dumb Chinese people are literate, so no, intelligence is not required for literacy in a non-alphabetic system.
600 characters is about what's needed for literacy. (they combine, so there's more words).
https://www.thatsmandarin.com/blog/how-many-chinese-characters-do-i-need/
Japan goes with about 2100 (these are taught in school), with around a thousand as taught in elementary school.
For what it's worth, it might be true that China has more complicated institutions than the West. When my family member emigrated from China, they were enormously relieved at the amount of bribery and favor-trading they no longer had to do to acquire basic goods and services
I suppose one could argue that the complexity of human institutions has been correlated with the general rise in IQ in all regions of the world over the past century. But I think a better explanation is that the World's GDP has multiplied 4.5x in the past 50 years. More types of gizmos are being manufactured. More companies to make the gizmos. More paperwork (now electronic) to track the gizmo logistics etc. Plus, companies have gone multinational, which requires more workers and whole new systems to deal with dozens of regulatory regimes, etc.
I joined Cisco in 1998. I was at the bottom of the organizational tree, but there were only four levels of management between me and the CEO (John Chambers). When I left Cisco in 2002, there were ten levels between me and John Chambers. The organization had burgeoned with our product space. New systems were implemented to deal with the new complexity of the organization. This definitely had nothing to do with changes in IQ and the timeline was too short to have been affected by the Flynn Effect.
Generational intelligence has been increasing by ~3 pts per decade in the US. Our institutions are definitely more complex than they were when IQ started being measured a century ago. But probably that increase can't be attributed to increased IQ. I doubt if US citizens are any smarter than they were a century ago.
That ~3 has little sense because it depends on which IQ test is used; Raven's matrices show larger improvement while arithmetic is barely up. It's also that were no large-scale studies of modern people on 100 year old tests; it's extrapolation.
Then why do they keep renormalizing them every 10 to 20 years?
But your statement is incorrect. the Flynn Effect has affected all the major IQ tests. However it seems to have plateaued in the US and Scandinavian countries in 00s and 10s.
It affected all tests, but to a wildly different degrees, the 1st picture on wikipedia page about it makes it obvious.
If population improved 10 points on test A, then improved 10 points on test B, then improved 10 points on test C, it doesn't mean it improved 30 points on test A.
Flynn compared the raw scores (not normalized IQs) of people tested in different eras but taking the *same* version of the test. Someone taking the 1950 version in 1980 would score significantly higher than the 1950 average because the test hadn't been renormed. Flynn did not claim that people were getting smarter. And as you can see from the chart you referenced, the average increase in all 5 performance subsets (across various IQ tests) was ~3 pts per decade, which also closely matches the upward trend in FSIQ. RPM is the only actual IQ test mentioned on that chart (and ironically, it improved the most, even though it's supposedly the least culturally and educationally biased measure of intelligence).
And I very much doubt that people are getting smarter, even though raw "intelligence" (whatever the f**k that is) is improving.
They might do. But (1) if you magically (or technologically) make your children a bit cleverer then the effect on their ability to navigate institutional complexity is much larger than the effect on actual institutional complexity, and (2) if it's true that societies of smarter people have more complex institutions, that's probably because they're getting benefits from those more complex institutions sufficient to justify their complexity.
Still reading but this part stands out:
"Also, it would be crazy for any forward-thinking government not to cover this; it could save hundreds of thousands of dollars in future health care expenses. In countries with public health care, this comes directly out of the government treasury; even in the US, it’s covered by Medicare after age 65. The government should be begging people to select embryos."
A good test case for how this could roll out is genetic screening for stuff like cystic fibrosis, where there's a well understood genetic cause and it can be screened in infants (1).
From a very quick review, Medicare does not cover genetic screening for embryos where there's a family history of cystic fibrosis.
It looks like private insurance does cover screening but...only for patients with fertility problems. I'm not sure I understand this policy:
https://healthy.kaiserpermanente.org/content/dam/kporg/final/documents/health-plan-documents/notice/utilization-management/preimplantation-genetic-test-mas-en.pdf
I don't know if this is a good comparison. Cystic fibrosis affects 1/5000 people, type 2 diabetes will affect about 1/3.
Is your figure for type 2 diabetes specifically for the US? A quick google suggests that around 29% of people in the US over age 65 have (diagnosed or undiagnosed) type II diabetes, but around 16% in the EU, with fairly large differences between countries. Admittedly all these numbers are higher than I would have guessed.
One can imagine a dystopian future where McDonalds offers free gene editing to improve their future market share...
As for another country, there's Britain's NHS which does offer IVF screening of embyos for CF: <https://www.cysticfibrosis.org.uk/sites/default/files/2024-10/INF8-3-C-2%20-%20FS%20-%20Cascade_Screening_October2024_RGB%20%281%29.pdf>
>Pre-implantation genetic testing (PGT)
This is a test that is used with IVF, where eggs and sperm are
fertilised to make an embryo in a laboratory. PGT is used to make
sure that only an embryo that does not have CF is put into the
womb. A couple must meet certain criteria to access NHS-funded
PGT, which your genetic counsellor will explain to you.
The other thing that stood out to me about that section is that the logic there does not quite follow. Suppose someone's child would have lived to age 65, but thanks to polygenetic embryo selection was able to live to age 66 instead. This costs the US government more money, not less, because they have to pay one year of medical expenses instead of zero.
I think where this will work, if it works at all, will be in things like "does this embryo have CF or not?" and we can then see if the babies born do or do not develop CF. It'll work where there is a discernible, measurable problem and the result can be discernible and measured: "did the baby develop CF or spina bifida or the rest of it?"
That is a case that can be argued for public health benefits.
The stuff about "taller, smarter, prettier" and the rest? Will always be pie in the sky. Maybe the selected embryo is smarter than if you had gone the natural route, maybe the unselected baby would have been smart anyway. Roll those dice and hope for the jackpot!
It seems to me that using this technology not for vanity but to eliminate such a tragic condition as CF - I recall reading an obituary once, “he was preceded in death by … (his brother, his brother, his brother, his sister)” - would be the necessary justification for selecting for IQ.
Very smart people surely made it possible, but if they can’t get this right … what good were they?
Why do you think of it as an exclusive or? People will used it BOTH ways. Some weighing one factor heavier, and others another.
So maybe we can get less CF and more rappers.
It is important to know that we have made significant advances in treatment for CF over the last few decades. In 2022, the estimated median predicted survival age for a baby born in the UK with CF was 56. In 2023, it was 64. I’d expect to hear the prediction for babies born in 2024 around October. This is extraordinary progress.
It is no longer a disease which guarantees an early death, as it was relatively recently. And the progress is continuing and rapid.
And no gene-editing required for these treatments?
Not yet! They’ve been working on gene-editing therapies for ages: some have looked promising, but I’m not aware of any success yet. It’s been the big thing people have hoped for.
That’s wonderful to hear.
I'm a parent via IVF. When we were weighing the decision to do embryo screening, our providers lightly recommended against it on the following grounds: we are relatively young (mid-30s) and thus comparatively unlikely to have significant age-based aneuploid concerns; and any genetic screening would necessarily involve removal of cells from the embryo which has a non-insignificant chance of disrupting its viability. Can you speak to that latter concern? Particularly for the current userbase of IVF, viability of pregnancy is a pretty paramount topic!
This is a good question that I don't know the answer to. I'm hoping some of the embryo selection people will read this post and share their expertise.
https://guides.orchidhealth.com/post/embryo-biopsy-for-pgta-procedure-risks-and-success-rates:
"As with any medical procedure, embryo biopsy carries some risks. There is a small chance (less than 1-2% with a skilled embryologist) that the embryo may be damaged during the biopsy, causing it to stop developing or making it unsuitable for transfer. While current data suggests that embryo biopsy has no adverse impact on the growth or medical outcomes of the resulting children, it's essential to understand that the technique is still relatively new, and there may be potential unknown consequences for live-born babies. If you're considering PGTA for your frozen embryos, our article on getting Orchid's screening on frozen embryos provides valuable information."
Obviously, they're not an unbiased source.
If embryo biopsy results in a 1-2% chance of serious enough damage that the embryo is damaged enough "to stop developing or mak[e] it unsuitable for transfer," my guess is that that's just the tip of the iceberg, and that more subtle damage is being done to a larger percentage of embryos as well.
My understanding is that a small percentage -- like 5%? -- of embryos are damaged by genetic sampling. There's a newer non-invasive way to test the genetic material in the substrate around the embryo. I bet this will become more common. (I'm no expert, just have done some reading)
One thing that concerns me is the suggestion in the research that IVF conception itself may increase the risk of certain kinds of cancers, particularly with frozen embryo transfer. I think studies say this is a pretty small risk at this point, but if there's a massive increase in people using IVF, it may become a bigger issue.
So would it be fair to effectively "nerf" all of the health benefits quoted above by 5% given that likelihood of the testing being damaging? That certainly affects the final expected value.
I don't think so because a damaged embryo won't be implanted.
Yes, obviously it wouldn't be transferred. But you've lost that embryo. For many current users of IVF, the number of viable embryos that are retrieved is worryingly small, sometimes even just one or two per egg retrieval. If the embryo is damaged, they've gone from "reasonable likelihood of a healthy child" to "no child".
I guess I'm confused because if the data is about reducing risk of conditions for babies that are actually carried to term and born, then 5% of total embryos lost are unlikely to significantly change those improved outcomes.
Average number of eggs per retrieval is 8-14, they are graded by viability. It's true some women produce fewer eggs per retrieval and in that instance may forgo genetic testing because of that, but those are a minority of current IVF patients who are mostly doing IVF because of fertility problems just like that (insufficient egg supply).
Also there is a new method for testing embryos by testing the genetic material that exists in the substrate around the embryo. It's noninvasive to the embryo, and I think we're going to see more of this technique in the coming years because of this concern.
Most current users of IVF are those struggling with fertility issues. If embryo selection becomes normalized, median user with be younger and fertile.
I'm also interested in the possible impact on the balance of men and women in a population. Most of the people I know doing IVF (which is not that many) are making choices about embryos based on sex along side genetic factors. Is anyone talking about this? Do fertility clinics have data about whether these sex-based choices are tipping in one direction or the other?
I see from reading around that IVF skews male though this seems tied to what kind of IVF technology is used rather than because people's choices are skewing outcomes. Maybe too soon to say on that.
I don’t understand this.
I’ve read that people who select for sex prefer girls. Not too hard to believe
That's my anecdotal experience as well.
People who tell you they're selecting for sex will tell you they prefer girls. I'm guessing there's substantial stigma around selecting for boys, to the point where you'd get IQ or something instead as the reason.
This may be the result of overly cautious providers, but it may be particular to your embryos. If your embryos developed slowly and were frozen at a later-than-optimal stage of development (day 5 is preferred, but many providers will use and freeze days 6 and 7) or if your embryos appeared low grade under the microscope, genetic testing could cause sufficient trauma to the embryo to lose the embryo, especially if a freeze-thaw cycle is involved. This might also be recommended if you have very few embryos.
Removal of the cells for testing has a pretty small chance of harming the embryo under standard/ideal conditions, which is to say removal of cells for PGT-A/-M after fertilization but before initial freezing.
It's best when testing embryos to know upfront (i.e. before freezing) which testing you want to do, and even then testing is not an equal risk for all embryos.
And on average, two-thirds of patients require six or more IVF cycles for a successful pregnancy. At 15-20K a pop that could make a significant cost difference if the Herasight tests reduce the viability of embryos.
So I'm not actually an expert on this, but I did work at a genetic testing company for a year and I've talked with pretty much everyone in the field.
From everything I've read and heard, a trophectoderm biopsy (the standard kind of biopsy taken in IVF) has a negligible impact on live birth rates. I wish I could give you an actual number here like "the impact on llive birth rates is <X% based on this study", but my knowledge is not that deep.
There ARE studies showing that biopsies of the inner cell mass (the part of the embryo that becomes the actual fetus) do have an impact on success rates, but those aren't commonly practiced by clinics.
No doubt I'm an idiot, but why the focus on Type 2 diabetes, which is far more lifestyle-related, rather than on Type 1, which is far more genetically based? Also, I have Type 1 in my family.
Good question. I would guess because Type 2 is about 10x more common, which means more people care about it and they have better genetic data.
T1D is easy to predict but too rare to be of much interest to many customers. Even 5 years ago, PGS AUC was ~90%. https://www.mdpi.com/1422-0067/21/5/1703
It's only 90% if you do some goofy thing where you restrict cases to people who developed it at <5 years of age. The AUC for a more realistic case cutoff (<15) is still quite good, but not that good.
Especially since we already have cure for that (GLP-1), if rumours I heard are correct.
I'm curious to know if the risks to the baby have been taken into account here. Not the risks of embryo selection, but the risks of IVF in the first place. Obviously if a couple is having IVF for (eg) fertility reasons, an embryo has to be picked, whatever metric is used. But having IVF purely for embryo selection could be counterproductive, if the process itself leads to an increased risk of some adverse effects (eg preterm birth, lower birth weight), as some studies have suggested. That said, most IVF up to now has been done for couples having fertility issues, so it's hard to disentangle causation. Is this something you have looked into, and taken into account?
I'll add a link to this review, which seems pretty thorough and is open access: https://academic.oup.com/humupd/article/25/2/137/5316072
I hope someone replies to this comment.
I concluded that IVF probably has risks to child health after reading the review linked by Sun Kitten, but I'm not an expert.
Many of these traits represent our society's current values. Even the disease ADHD is mainly a disease because of the current cultural milieu. Is it wise to increase selection pressure as much as these technologies are likely to when they trickle down, given the transitory arbitrariness of the criteria? I can see reducing Alzheimer's and cancer. But I'm uneasy about many of the other criteria.
If having ADHD (to the extent it's a disease) is advantageous in some way, then there will end up being a shortage of people who do whatever jobs people with ADHD are good at, meaning parents will stop selecting against it because they want their kids to be able to get good jobs.
I can easily imagine a tragedy-of-the-commons situation where something is beneficial to society but not to individuals : maybe the jobs ADHD people are good at are low status or otherwise undesirable in a way that can't be corrected by economics in the way you describe. Or ADHD suddenly becomes valuable because of environmental changes. But this doesn't seem terribly likely?
A future with no comedians :(
A future with a lot of things missing in people.
And a decrease in musicians, artists, actors, people who are different and contribute to the rich fabric of society …
Charles Midi’s comment should be much higher in the string of concerns I’ve seen voiced in the comments. Time to rewatch GATACA.
This is by far the least convincing argument against eugenics I've ever heard. Oh no, there'll only be one billion kinds of person instead of one hundred billion, only more worthwhile art being created every week than anyone can can consume in a lifetime instead of every day, what a tragedy.
It would be worth it just for that alone.
I worry about the environmental changes first, because our world is changing at a very rapid rate, but I think the "something is beneficial to society but not to individuals" is underrated. Imagine AIs paid in "content" -- lolcows become a very valuable commodity for a society, then. Regardless of how well they do for themselves.
re: environmental changes I'm hopeful that reduced genetic diversity won't matter a ton because humans have adapted to pretty extremely different environments in the past with changes in memes not genes. Humans *have* adapted genetically to changing environs (high altitude tibetans adapting to low oxygen, duplication of genes to metabolize grains post agricultural revolution... etc.) But these are really quite minor compared to the way humans have adapted to new environs through new technologies (consider the Inuit! Or space travel!)
Even without embryo selection, every population outside Africa is already very homogeneous due to founder effects and rapid population growth and we've done OK. So to be honest I'm not terribly worried that embryo selection will eliminate some essential diversity. Though it is a frightening idea.
You missed our evolutionary adaptations as "fire users" (which seem to be quite extreme).
Manic Depression or other psychological "brainstates" may be what's necessary to make geniuses. (Not that I'm saying every genius was the Romantic Ideal, many are workhorses -- but even they can go through cycles of "up" and "down" -- or inability to start projects because "there's so many ways to start it!")
By "evolutionary adaptations as fire users" are you referring to co-evolution with technology (e.g. we can't extract enough calories to reproduce without some kind of fire/cooking)?
Right now I'm not persuaded by the idea that genius is connected to mental illness, but if it turns out to be so that definitely would suggest embryo selection against mental illness could be pretty bad! That would be its own interesting moral dilemma. "This embryo has X% chance of debilitating mental illness but Y% chance of being a genius and benefiting lots of other people."
I worry that the equilibrium might not shake out to be that adequate. Parents make bad and silly decisions about their children's lives and future all the time, and status game spirals can often take over parts of human culture quite thoroughly.
One problem is the feedback time is too long: if there are actually jobs people with ADHD are good at (but fewer than the amount of people with ADHD today), the issue that we select too much against ADHD will only become apparent 20–30 years after embryo selection becomes widespread, so we'll have an entire generation of people too heavily selected against ADHD.
The vast majority of congenital conditions are universally considered harmful. No one's arguing that anencephaly, Proteus syndrome, Sickle cell disease, asthma, bladder exstrophy, or nearly all the others improve a society. These conditions cause great pain independent of any "cultural milieu."
The argument for preserving certain congenital conditions only applies to the tiny fraction of them that some people think provide some benefit like ADHD or high-functioning autism. These conditions won't be selected against as much since many parents will think they are fine, so I don't see this as being a problem.
Should we be maybe trying to create a database of whole-genome sequences and alleles before we start artificially changing the population distribution? Ideally one that is large enough to capture a pretty large chunk of rare alleles?
I think I'm pretty well on the side of "this is probably a good thing", but in the off chance that wide spread use of this technology causes unforseen problems, knowing what our populations genetic distribution looked like before we messed it all up seems pretty useful.
I think we have databases like this - wouldn't UK Biobank count?
It does, but only for the British population, and only for the healthier, wealthier portion at that (it's the volunteer effect).
We have one. Actually, we have lots. See, for example, gnomAD (https://gnomad.broadinstitute.org/), Ensembl (https://www.ensembl.org/Homo_sapiens/Info/Index), UCSC (https://genome.ucsc.edu/cgi-bin/hgTracks?db=hg38) or the Human Pangenome Reference Consortium (https://humanpangenome.org/).
(Admittedly, Ensembl and UCSC are genome browsers, but they include variants and their information, such as frequency, predicted impact, and known phenotypes).
Pangenome sounds pretty much like what I'm envisioning, although, if I'm understanding statements on the website correctly, they are talking about the order of hundreds of full genome sequences. That's a great start, but in my opinion, we should probably be aiming for thousands-to-millions to get a fully representative global sample that is large enough to capture non-trivial amounts of rare alleles.
It's going to be a looooong time before this tech has any noticeable impact on genetic diversity. There are about 10 million genetic variants in the human gene pool with a frequency of at least 1%. Even if you did embryo selection for ten generations, and everyone was selected for the same things, you'd barely move that number.
Frankly, even with more powerful techniques like gene editing or iterated meiotic selection you're going to need many generations before this becomes a big concern.
The only area where I'm somewhat worried about this is the human immune system. We've had crazy amounts of selection on the human immune system in the last 10,000 years, to the point where we literally have higher rates of read errors in genetic sequencing when reading regions like the HLA. It's also harder to judge the tradeoffs you're making when changing immune variants because often they're present to protect against plagues or other uncommon diseases. We can't really phenotype those effects well until there's some sort of pandemic.
HLA pops up in GWASes for every damn trait too, so alleles there might be weighted highly in the predictors these companies are using? (Or maybe they are filtered out, since as you noted it's hard to align to that region?) One can imagine a "tragedy of the commons" situation, where every individual is incentivized to select for a certain set of immune haplotypes that e.g. reduce schizophrenia risk, but this counteracts diversifying selection at that locus, making the population as a whole more vulnerable to disease. But that's just a worst-case I have no strong reason to believe currently.
> We can't really phenotype those effects well until there's some sort of pandemic.
That's a good point. It reminds me of an old saw that goes something like:
Student: The mouse knockout model has no phenotype.
Professor: Well, did you take it to the opera?"
(The joke being that you can't know that the gene doesn't produce a phenotype under opera conditions until you explicitly test it.)
Well the plague protection benefits of some HLA variant would increase as the population as a whole became more vulnerable, so if you have good enough predictors and there is at least some transparency into what other parents are selecting for, this is something that in theory we should be able to avoid.
Though it would obviously be more difficult than the usual kind of selection.
Honestly the hardest part is probably going to be figuring out which HLA or other immune variants protect against which pandemics. This is why I think by default we should be prioritizing immune diversity **relative to other people in the population**
>Well the plague protection benefits of some HLA variant would increase as the population as a whole became more vulnerable
Maybe. It also seems very possible that, rather than the main effect of HLA heterogeneity being "at least someone will have a protective allele when a pandemic hits", HLA heterogeneity is *itself* protective. If most people have the same HLA the pathogen just needs to change its protein sequences to avoid antigen presentation by that one HLA. If we all have different HLAs the pathogen needs to avoid being presented by ANY of them, which dramatically constrains it & reduces transmissibility in a population, preventing outbreaks. So your neighbor's HLA diversity could benefit you.
(This is not to mention the specter of other nasty things like transmissible cancers that can result from low genetic diversity in immune genes).
>if you have good enough predictors and there is at least some transparency into what other parents are selecting for, this is something that in theory we should be able to avoid.
Absolutely. I would bet that by the time that embryo selection became wide-spread enough to start seriously affecting allele frequencies the tech will have improved enough that we can just give kids whatever MHC to promote diversity relative to the population just as you suggest & this whole thing will be a moot point.
Who says we aren't artificially changing the population distribution? We have had significant manmade interventions on the population scale side, and they may have reproductive effects.
It seems kind of rude to say Kenyan pastoralists get weather predictions from a Shaman instead of being able to connect to the same exact up-to-the-minute weather satellite data as everyone else
I've edited, but don't those only predict for the next ten days, rather than whether it will be a good or bad season?
There are definitely 90-day weather forecasts from normal forecasting services.
I suspect these are fake, but I guess I also suspect shamans are fake, so whatever.
Meteorologists I know tell me the predictions for next week are mostly priors, i.e., what was the average weather in previous years?
I'll bet I can predict with great accuracy whether 90 days from now it will be hotter or colder than today in a given city in the world. Not perfectly (gets fuzzier in the tropics) but generally, November has different temperatures than August in most places and the direction of weather is practically always in one direcction.
ENSO is ENSO.
https://en.wikipedia.org/wiki/El_Ni%C3%B1o%E2%80%93Southern_Oscillation
It's fairly predictive into the future, when it's strong, and "interesting" when it's in the middle of the cycle.
I imagine this isn't the only "widescale weather pattern" that can be used for seasonal predictions, but it's the first that popped into my head.
Paging Keanu Reeves' agent!
Keanu has to star in the biopic "Jonny Anomaly."
Haha, very good!
(But I'd still have preferred Gene Hackman)
It's exciting to see how this field is progressing! My wife and I will likely be customers of one of these companies in the near future.
Regarding SMN and spinal muscular atrophy, the SMN gene is rather difficult to analyze because there are regions of the genome (SMN pseudogenes) that look very similar. Still, difficult is not impossible and I'd expect Nucleus to do better, given that the focus of their company is genetic testing. (See: https://www.illumina.com/science/genomics-research/articles/HBA-SMN-genotyping.html)
And regarding cost, hopefully this will come down in the near future. The cost for genotyping/sequencing should be, at most, in the low hundreds of dollars per embryo, and I think now that this kind of selection is starting to be within the Overton window, we'll see competition drive down the price. So most of the costs will end up being IVF.
Furthermore I think companies which do the actual genotyping/sequencing (GP and Orchid, right now) will end up with a competitive advantage in the long term.
Here's another concern I have about IVF becoming more widespread in order to access genetic testing... and it's about the physical and psychological impact on women. It's an unbelievably demanding process above and beyond pregnancy, childbirth, and early motherhood which are already incredibly demanding. Anyone who has done it can speak to what a very hard road it is for most women.
Last I checked, something like 70% of women going through IVF experience clinical levels of anxiety and depression. About half of women under 35 get pregnant in the first cycle, so even younger women are likely to need to go through more than one cycle and women over 35 even more.
Though I should add that most (not all) IVF right now happens to address infertility issues and that's no doubt part of the psychological anguish of the process. But the hormones, pain, unpredictability, cost, risk, etc etc even without infertility are still quite a load.
True, but women can make their own decisions. And if you consider the impact of having a child with schizophrenia, it may be a good investment.
One thing you're not taking into account here is just how much worse IVF outcomes are for women with known fertility issues.
If you're having trouble conceiving, IVF can help but your odds still won't necessarily be that great. This is especially true for women in their late 30s or early 40s. If you're under 35 and you have no fertility issues, your odds of having a kid via even just a single round of IVf are around >85%
What really makes people go crazy with IVF is multiple rounds of egg retrieval, implantation, and failure. Yes there are unpleasant side-effects of the medications you need to take, but the things that really drives people crazy is not being able to have the child they so desperately want.
Two contradictory points:
- I think that all future costs should be time discounted (the why is a whole debate over ethics and economics, but I come down strongly pro). By that argument this somehow comes out as not worth it (per your calculations), despite how great it sounds.
- I'm calling baloney on a 6iq point gain being worth that little. If we were incorporating the uncertainty about it (10% chance it works as promised otherwise you get nothing), then maybe, but an actual 6 point IQ gain can't possibly be worth only a five or six figure sum of money.
Scott cites Zagorsky (2007) for the estimate and there are a few of reasons to think the true value would be higher.
First, Zagorsky (2007) is based on a 2004 survey, but in skimming the study, it doesn't look like they adjusted for inflation between 2004 and 2007, which was about 10%. Scott seems to have adjusted for inflation since 2007, but if I'm right, the adjustment should have been about 10% higher.
Second, simply adjusting for CPI from 2007 (or even 2004) probably underestimates the modern equivalent estimate. That's because over time, people have become richer in real terms. Higher IQ presumably shifts you along the income distribution, but since the country is richer than it was in 2004 (with real GDP per capita increasing by about a third) an equivalent SD shift would have a greater impact.
Third, income stratification by IQ has increased over time, suggesting that current returns to IQ would be higher than in 2004, see here: https://x.com/cremieuxrecueil/status/1818758295856771135.
Fourth, and probably most importantly, reanalyses of the Zagorsky data show higher correlations than Zagorsky found. For example, Dalliard (2016): https://humanvarieties.org/2016/01/31/iq-and-permanent-income-sizing-up-the-iq-paradox/ found that when considering multiple years of data, the return of 1 IQ point to permanent income is $1,130 (95% CI: 1048-1213) in 2011 dollars, or $950 in 2004 dollars, more than double the Zagorsky estimate.
Using log-transformed income data to address non-linearity, he finds that the affect of 1 IQ point is an increase in income of 2.5%, such that the affect of 6 IQ points would be an increase of income of 16%.
Scott's estimate implies that 6 IQ points would boost annual income by $4,020. Per Dalliard, that would only be true if one's salary without the IQ boost would be $25,125. In other words, per Dalliard, it wildly underestimates it.
For the median salary in the US of $62,088, the annual income boost would be $9,915. For readers interested who presumably have higher than median expected earnings, the boost would be larger, yet.
And my third point applies to Dalliard as well as Zagorsky, suggesting that modern data might show an even higher estimate for Dalliard's analysis. Indeed, the Tweet above shows precisely that - for the same data being referenced above (but different years), the effect of IQ on income (for given ages) was significantly larger for the cohort born in 1997 than the cohort born in 1979.
After writing this, I see that the analysis in the Tweet came from here: https://www.sebjenseb.net/p/how-profitable-is-embryo-selection which actually addresses this very topic, estimating the effect of a 1 IQ point increase on total lifetime earnings of between $40,000 and $90,000, ignoring discounting, just as Scott did in footnote 8.
We've all met (or are) plenty of smart people who don't earn a good return on their intellect
Sure, but on the other hand we've also met people who just barely missed out on jobs that could double or triple their pay.
(There's also significant non-direct financial issues, like reduced probability of violent crime or increased marriage stability).
Damn. I was rushing to write something about this since yesterday but Scott beat me to it.
All I’ll say is that I have a feeling this will be one of those incredibly transformative technologies that change the course of human history forever. I’m somewhat of an ASI skeptic, so I imagine a relatively near future where 3 generations of this technology rapidly improving leads to a real sense of “super babies.” There’s enough sci-fi written about this idea that it’s not important for me to go into all the interesting utopias and dystopias this might lead to.
I think ASI will render this stuff moot in a few years, but if humanity had any will to not die, this is how. We'd halt the AI race, make ourselves smarter through genetic engineering, and then try our hands at solving alignment with our newfound wisdom.
My only regret is that it seems I might be a little too late to get involved with it. Maybe not though.
If we can't align AI with other AI, why should we be able to align AI with designer babies?
Because humans are aligned with us to begin with.
That's like saying that English colonizers were necessarily aligned with Sudanese tribes.
There's no reason a priori to expect designer babies to have values unaligned with humanity.
I see your point that you're building on the same architecture, and raising them to have positive feelings towards other humans. But depending on how far they are ahead of baseline humanity, they may not be interested in allowing baseline humanity to participate in collective decisions - they would see it as like allowing children to vote and bind adults to their decisions. I agree that there's a difference between that and an ASI, but I'm not convinced there isn't any overlap between the problems.
Herasight? As in Hera, casting Hephaestus from the mountain for being deformed?
She's the goddess of childbirth.
Hera is goddess of marriage. Ilithyia is the goddess of childbirth, and Hera used her to delay the birth of Heracles so his kinsman Eurystheus could be born first and become king:
"On the night Heracles and Iphicles were to be born, Hera, knowing of her husband Zeus's adultery, persuaded Zeus to swear an oath that the child born that night to a member of the House of Perseus would become High King. Hera did this knowing that while Heracles was to be born a descendant of Perseus, so too was Eurystheus. Once the oath was sworn, Hera hurried to Alcmene's dwelling and slowed the birth of Heracles and Iphicles by forcing Ilithyia, goddess of childbirth, to sit cross-legged with her clothing tied in knots, thereby causing the twins to be trapped in the womb. Meanwhile, Hera caused Eurystheus to be born prematurely, making him High King in place of Heracles. She would have permanently delayed Heracles's birth had she not been fooled by Galanthis, Alcmene's servant, who lied to Ilithyia, saying that Alcmene had already delivered the baby. Upon hearing this, she jumped in surprise, loosing the knots and inadvertently allowing Alcmene to give birth to Heracles and Iphicles."
So, um, yeah: great patron for picking and choosing who is going to be born for which traits, not so much for "child will have happy life afterwards".
Eesh, none of that is what I'd want to be invoking, but I think Immortal Lurker is right there's an intentional link.
Oh definitely, just not (it would appear, if even Scott is confused as to what Hera is the patron deity of) the associations customers think they are invoking.
"Hi! Our patron is in fact the goddess of marriage, not childbirth, and very vengeful about adultery or non-marital relationships, and very much up for revenge on the offspring of same! So let's all be certain that you're married, not to other people, and having babies under traditional norms or else there might be serpents in the cradle - so to speak" 😁
Worked out pretty well for Hephaestus. He got his own forge in a giant volcano, and married to the goddess of love/lust (even if she cucks him most of the time).
They're just poly
I really, really wish that Scott and other effective altruists took the personhood objection seriously.
I'm not talking about the primary concern that Scott mentions in the linked post, "Who Does Polygenic Selection Help?"; I'm talking about the secondary, "pro-life" concern--the concern that, beginning at fertilization, a human embryo has moral value that's similar in magnitude to a newborn baby, and so abandoning or destroying it is morally akin to abandoning or killing a newborn baby.
Although the philosophical positions that underlie that concern aren't ironclad, I think it's the most plausible account of when humans acquire moral value. At the very least, it's a *plausible* account. Consequently, that should make EAs very reluctant to advocate for a practice that might grow to result in the industrialized killing of millions (perhaps billions) of people per year.
The closest that Scott comes to addressing this is saying that if we threw out polygenic selection on these grounds, then we would have to throw out IVF too. So be it. I don't think that the moral correctness of IVF is a reasonable anchor for our intuition whatsoever.
Even Scott's claim that polygenic screening doesn't exacerbate the issue is seriously wrong. In classic IVF, the purpose of creating multiple embryos is to increase the probability of pregnancy. This number has a conservative upper bound to minimize the possibility of triplets, quadruplets, etc. In contrast, the ideal number of embryos for polygenic screening is practically unlimited, because almost none of them are intended to be implanted. Knowledgeable proponents like Gene Smith and Jonathan Anomaly advocate for batches of dozens, if not hundreds, of embryos per pregnancy. This would easily increase the impact of the ethical question by a factor of 10. Furthermore, if polygenic screening becomes the norm (as some advocate), then the impact will increase by an *extra* factor of at least 20, based on current IVF rates.
I understand the appeal of polygenic screening. I really do--and if it wasn't for the issue of embryonic rights, I would be an avid proponent. As it stands, though, we risk creating an entrenched moral catastrophe akin to the next factoring farming, and the people who pride themselves about their quantitatively informed moral sensibilities don't seem willing to properly wrestle with it.
An embryo at the stages we're talking about here is a double-digit number of cells. It doesn't even have a nervous system, and seems less conscious than an insect, let alone a newborn. If you're coming at this from the abortion debate, where we're talking about fetuses with beating hearts and semi-working brains, your intuitions will be totally off. I think if you want to argue for embryo rights, you need some non-materialist conception of what personhood is or why it matters, which is going to be a hard sell.
I don't think that embryonic rights require a non-materialist account of personhood--at least not in a way that's suspicious.
Suppose that while you're walking in the woods, you encounter a hermit who's fast asleep. If you had a way of killing him instantly, would it be morally acceptable to do so? He's not currently experiencing any desires, he wouldn't feel any pain while being killed, and nobody would miss him.
If you consider that to be murder, then what's the morally relevant difference between the hermit and an embryo? Surely it's not the nervous system itself that's morally significant; it's the capacity for consciousness that we associate with the nervous system. A human embryo also has this capacity; it's merely a step farther removed from manifesting this capacity. The embryo has to grow a nervous system, but why is that different from the hermit having to wake up?
How is this different from saying that a couple that has three kids but chooses not to have a fourth is murdering their fourth kid, and exactly as culpable as a couple that has four kids and then kills one?
I think I would solve the hermit situation by saying that the hermit has a preference against dying (ie formed this preference when he was awake), and that preference can't be ignored just because he's asleep right now.
Thanks Scott for directing me down here.
I think a good criteria for moral standing is existence and the potential for consciousness. An embryo has both, a sleeping hermit has both, an unconcieved child does not.
I don't think your hermit explanation holds up - why is an assumed past preference to live any more valuable than the assumed future preference to live that an embryo has?
I agree that the preference explanation doesn't suffice: even if we suppose that the hermit was temporarily depressed, with no desire to survive, it's still bad if they die when they would have otherwise recovered and lived a happy future life.
The better explanation is that the hermit is psychologically continuous with their (possible) future self. An embryo isn't. (See my reply to Matthew's comment for more detail.)
Could you give a more detailed account of why you think an embryo has "potential for consciousness" but (you seem to imply) an unfertilized egg does not?
The unfertilized egg requires an outside intervention to eventually become conscious, but this seems true of an embryo in a test tube as well; neither is going to develop into a healthy baby without a great deal of outside help.
to better phrase what i said above, existence as something with the potential for consciousness gives moral standing. An unfertilized egg does not meet that - it has to completely change through fertilization to have that potential. An embryo in a test tube only needs the right conditions to develop consciousness. other choices can be argued, but i think fertilization is the most reasonable as far as yielding a radical step change from no standing to standing.
An embryo has a full complement of human DNA, it is meaningfully "this person" rather than "that person".
It sounds like on your theory, sperm should have moral standing, which is pretty bad news.
Or suppose that one day robots become conscious. There is a lump of metal in the robot factory. Does it have moral standing, since it could one day be turned into a robot?
I just don't think we need to worry about things that don't have moral standing yet, just because they theoretically could in some future which is not going to happen.
I do think you need to have something like an assumption that past preferences extend into the future to value anyone's preferences at all. Suppose I don't want someone to scream in my ear. I'm probably not thinking about that at this exact moment. You need to sort of abstract my into a continuously-existing being in order to respect that preference and not scream in my ear.
(another reason not to grant embryos moral standing is that I think more than half of embryos that implant are spontaneously aborted because the body decides it doesn't like them for some reason. If we thought embryos had moral standing, this would become the worst genocide in history, and the most important cause would be inventing some kind of artificial womb to save them. But in fact I predict most people don't care about this very much.)
We used to have 50% child mortality. Most people couldn’t do anything about it and just accepted it. Not a genocide, because it’s not the intentional killing of a particular ethnic group, but definitely a mass tragedy.
For infertile couples, if we could end that spontaneous abortion, they’d be happy to have their children. For other couple as well, this kind of miscarriage is probably as sad as losing a newborn would’ve been in cultures that prevented naming until 6+ weeks to avoid being attached.
The argument from consequences isn’t a good one because humans have accepted plenty of bad consequences before we had the capability to end them.
I don’t think an individual sperm is alive in the way an embryo is. An embryo needs food, water, and shelter to grow. From the moment of conception its life plan is determined. A sperm needs an egg, and different eggs will result in different life plans (genes). Much much less deterministic. I think you can say that “you” were once an embryo. I don’t think you can say you were once that sperm, because that sperm could really have become anybody else
My line is existence as something with potential for consciousness. I don't think a sperm qualifies as a person, so doesn't qualify.
I'm going to call robot moral standing out of scope (we're having a hard enough time with people).
Your idea about a moral obligation to stop the genocide of embryos through science is very interesting though. I hadn't thought of this from an EA perspective. I think this still works though - GiveWell says right now it costs 3k to save a life in Nigeria. To argue for devoting your altruism to saving embryos, you'd have to be convinced you could find something that costs less than 3k per embryo (accounting for any suffering from the procedure) plus the up front research cost. I don't think that's likely, but if you could I think I'd be for it.
I agree with others that there is a significant difference between an embryo and a sperm. An embryo is an organism that has the complete genetic information it needs for a human body and (unless someone puts a stop to it or deprives it of a healthy environment) is already at work building it's human body. It isn't a building block in the same way that a sperm or a hot meal is. It is a living organism that is consuming energy in order to build up and develop into what it already completely *is*, not in order to become something other. Nothing further is added to it for the rest of its life other than sustenance.
A sperm isn't like that. A sperm is happy to go it's whole life swimming around and then to die. No matter how much sustenance you give it, it won't develop any further. You need to combine it with a whole different organism for it to become something other than it was, and from there it's a human, no longer a sperm or an egg. I agree that you can't grant moral standing to a lump of clay or to sperm, but that doesn't have anything to do with an embryo, which is a human at the very beginning of its development, not a building block of a human.
As for the issue of spontaneous abortions: I admit that it's pretty hard for me to feel emotions instinctively about a human that still only consists of a few cells (at least ones that I can't see--something weird triggers inside me when I see imaging of an embryo), but thinking about the fact you bring up I do think it's tragic. I think there is not much that the average person could do to solve the problem other than not to try to have kids, but I don't think they would be guilty of a death from the woman's body rejecting the embryo, just as they wouldn't be guilty if their child later died of cancer. Both are bodies not operating as they should.
Just because the average person is unable to do anything about a common cause of embryo (human) death, I don't think that justifies them in intentionally killing a different embryo.
That's a simple explanation, but I really do believe it's as simple as that. For someone like you, who has studied and thought about these kinds of things extremely extensively, I would imagine it would be incredibly difficult for you to think of it in simple terms like this. I wouldn't blame you. And if one day you change your mind and feel guilty of the murder of your own children (although if the embryos are frozen then technically they're not murdered, but that's still a really weird moral ground. I was very surprised just now when I googled it, though, to find out that there are embryos from 30+ years ago being unfrozen and allowed to develop into babies) I don't think any reasonable person would condemn you for it aside from your own self. If you don't belive in God, then what's the point anyway. And if you one day come to believe and have faith in Jesus, he will have compassion and forgiveness for you.
Please bear in mind that most of these potential consciousnesses die. Also bear in mind that something like 50% of Americans signed up for interventions which would directly lead to "the potential for consciousness" being terminated. (ahem, okay, only half of them were female, and some of not childbearing age -- you still didn't see people (who weren't Jewish and in Israel) running around issuing fatwahs about the intervention).
True, but lots of unfortunate things happen, I don't think this can be used to justify purposely causing them to happen. I'm not sure what you're referencing with 50% of Americans, but i've seen the choices approximately 50% of Americans have made, and it doesn't really inspire confidence in them being a barometer for morality.
>How is this different from saying that a couple that has three kids but chooses not to have a fourth is murdering their fourth kid, and exactly as culpable as a couple that has four kids and then kills one?
The standard answer would be that fertilised eggs, unlike hypothetical people, have at least some form of moral personhood, by virtue of having a complete complement of DNA.
If you trace a person back through time you'll find they change a lot as they regress into a child, then a baby, then a fetus, but there's one discontinuous point beyond which you can't say that "this person" still exists, and that's the point where that egg combined with _this particular_ sperm rather than any other.
That's the standard (non-religious) version of the Life Begins At Conception argument, anyway.
A sleeping hermit has a mind, even if their mental states aren't being actively processed. It's completely different from merely having a "potential" to form mental states after a bunch of further development.
I think there's an absolutely decisive argument that embryos aren't harmed by death. It's that only beings with minds can be harmed at all.
Many philosophers (from Locke through to Parfit) have been convinced that some kind of psychological continuity is what matters in survival. Here's why: If a mad scientist scanned my brain, disintegrated my body, and then wiped your brain and implanted all of my mental traits (memories, beliefs, values, personality, etc.) in its place, then it seems that I have gotten the better half of the deal. I have survived and you have not. Though your body is the one that lives on, it is our minds that matter, and it seems that your mind has been replaced by mine. In making this judgment, we implicitly judge that it is the content of a mind that matters -- the memories, beliefs, desires, and so forth -- not its "location" in a particular body, or even a particular brain.
Suppose this scenario goes ahead. I wake in your body and return to my old life (as best I can). Does this make any difference to you? Suppose you're told beforehand that after your mind is wiped and replaced with mine, I'll go on to live a happy life. Or maybe you're told that I'll be killed the next day. You might for altruistic reasons prefer the former news, but do you think that you are harmed if the latter outcome occurs instead? If not, this goes to show that preventing future pleasures from being experienced in your body is no harm to you, if you are not the one who will get to experience them.
It follows that mindless embryos cannot "survive" (in the moral sense, i.e. mentally), and hence aren't harmed or deprived of anything by their biological deaths.
I follow what you're saying about the mind being more important than the body, but I dont see how this refutes the idea that denying a pre-conscious, extant being the opportunity to become concious is morally wrong.
Call the embryo 'Emby'. If you agree that you aren't benefited by having me live on in your body, then you should also agree that Emby isn't benefited by having some future person live in their body. Emby can't "become" conscious, any more than you can give my next lecture after I take over your body. Bodies are just husks. It's minds that matter, and Emby isn't one and never will be one.
It seems the crux of our disagreement is the idea that "It's that only beings with minds can be harmed at all"
I think it's pretty common in EA to consider the welfare of future humans who don't even exist yet, much less have minds, in moral calculations. Scott even does in this article ( along the lines of think of the future person who won't have lung cancer because someone selected against it in ivf)
How do you handle this? Do you think future generations have moral standing? If so, how do you square this with them not having minds?
Merely possible people (who never end up existing with minds, e.g. because we go extinct first) can't be harmed. But harms aren't all that matter. It would be better to bring about a future with more happy, valuable lives, and I think we have moral reasons to bring about better rather than worse futures for the world, all else equal. I explain my view of existential harms and benefits in more detail here:
https://www.goodthoughts.blog/p/the-profoundest-error-in-population
Future people do, of course, have minds. They just aren't present at this moment.
I read your post. I found myself nodding along but wondering why it's reasonable to prefer Joy's existence and not Emby's, especially since Emby is already further down the path to consciousness than Joy.
I believe that you can harm a future person who will exist by making their life worse than it would otherwise have been, but you can't harm a never-existent person by making them not exist at all.
So if I use drugs while I'm pregnant, and this causes me to give birth to a baby with some sort of disability, I am culpable. But if I decide not to have a child, I'm not culpable. And this is true even though if we start with an existing person, killing them is worse than disabling them!
So it's fine to care about future generations in the sense of having good policies that protect the environment for future generations or something, because you are making the lives of people who will exist better off.
So you care about the lives of future generations of people who are theoretical and do not exist yet, but don't care about the existing lives of future people, embryos.
Go back in time to when your kids were embryos. If someone had the power to and decided to destroy them when they were still just a couple dozen cells, and replaced them with different embryos (sorry I don't know what kind of twins you have so modify as needed) would that seem wrong to you?
Now what if someone had the power to, and decided to change which sperm and egg met to create your child/children. There were never embryos of your existing kids, just different ones, but the result is the same as the above example. You wouldn't have the same kids you have, but would it still feel morally wrong?
When I consider this, I think the first scenario is something akin to murder, but the second is not.
This is good insight into EA. And quite predictable from the very observation of their concern with remote future --which does not exist.
They display quite an inversion from the normal expanding zones of moral concern. One goes out from concern with people near one and time close by, to progressively less concern about places and times distant from us.
Anything whatsoever might be justified by the inversion of this normal process.
Arguments are futile in moral matters. Since arguments can proceed to conclusion only if premises are shared. But between materialists and non-materialists key premises are not shared.
Not sure about "futile". It may be futile discussing the "morals of IVF" outside a real world. But IVF is as much real world as are IUDs, miscarriages and all people alive on this planet. 1. Sure, one can argue that condoms are ok, the pill ambiguous but IUDs 'clearly' morally abhorrent, as they keep a fertilized egg from implanting (why my teacher for Catholic religion was vs. IUDS). In real life, people call them all contraceptives and near all use them according to their preferences (STI a potential problem? Not feeling well with the hormones? Long time planning?), not to "moral" considerations. Looking at miscarriages, one may call 'our Lord: the greatest aborterOAT' - but then it may be for the better not to have all embryos become toddlers.
IVF: if one asked Emby: "Hey, you are in an IVF-clinic, together with 3 siblings: we will check you now and implant the "best one". The other go in the bin. Are you willing to take your 25% chance or shall you prefer the bin right away?" - a smart emby would say: I take my chances.
And sorry to our extreme pro-lifers, humans will not implant AND deliver all their IVF-embies. Nor will they refrain from IUDs nor will they have only 2.3 times sex in their life. Except the protestant couple after "every sperm is sacred". ;) It is selection all the way down.
Fwiw, it's not like iuds are the only option. Condoms, surgery, and the pill (depending on who you believe) solve the problem of wanting to have sex but not wanting to destroy embryos.
Just to check, do you disagree with the following premise?
> "If a mad scientist scanned my brain, disintegrated my body, and then wiped your brain and implanted all of my mental traits (memories, beliefs, values, personality, etc.) in its place, then it seems that I have gotten the better half of the deal. I have survived and you have not."
(fwiw I'm not a materialist, but I still accept a psychological theory of personal identity)
More materially (ha), I fail to appreciate the special perspectives within the embryo vs person debate in the first place. Surely these moral objections either way are themselves the product of Fisherian runaway selection - the heightened awareness of a particular trait with selective advantage. What genetics has originated, genetical intervention may cure.
Is there is materialist conception of personhood?
And as for the hard sell of non-materialist conception, I suppose it is hard sell to those to whom anything smacking of non-materialism is hard sell per se.
But I suppose, to the general population, the hardgoing materialism is a harder sell than non-material conception of personhood.
"Ethical Objection: Personhood
Is it really correct to say that you have reduced someone’s risk of breast cancer by 46%, if what you’ve really done is closer to replacing them with a different person who is 46% less likely to have breast cancer?"
I don't see this is a question. It's not "this is an existing embryo which is a potential baby and you're just picking the traits for them", it's "this is a collection of existing embryos which are potential babies and you are selecting out one of them to be born by a checklist".
By picking "embryo with the lower risk of breast cancer" you are choosing to let that one develop into a baby that will be born. The other embryos will either be re-frozen for a later attempt if needed, or destroyed. You haven't "reduced someone's risk of breast cancer" (as if, for example, they were an adult and a new drug was invented that stops cancers developing), you've killed the competitors who might have been born and left the survivor with the least risk to be implanted.
Now, if you had *one* embryo and you were doing a little snippity-snippity with CRISPR or something to go "delete this bit here, stitch in a new bit there", then yeah: you're reducing that (potential) person*'s risk of breast cancer. But that's not what this procedure is doing. The other embryos are perfectly able, if implanted, to develop into babies that could be born. You're killing those embryos or at the very least preventing their natural development by sticking them back in the freezer, so you are taking potential people out of existence.
EDIT: And no, that's not the same argument as "you have therefore to have as many babies as possible to allow all potential people to come into being". These embryos now exist in reality and in potential to be viable pregnancies. They're not theoretical or notional, they're the result of someone's sperm fertilising someone else's eggs and are now ready to follow the path of conception to birth, only you chose to do all this in a petri dish instead of by the body.
*I think embryos have personhood but yeah: dumb old believer in stupid sky fairy story here.
Thank you for saying this! I don’t think CRISPR would work on any stage but the zygote (well, in the case of something like cystic fibrosis maybe if half the person’s cells are able to produce the missing protein then the effects of CF wouldn’t be as devastating). But anyways I consider IVF immoral so what do I know. It really makes me sad to see all these rationalists’ inconsistencies in moral reasoning and reminds me that as much as I find interesting a lot of content on this blog the fundamental framework is so different we might as well be living in different galaxies (though Father Brown would remind me that the natural law applies everywhere in the universe). We can’t have honey apparently because of bee suffering but creating embryos just to let them die is a-ok!
Ah yeah, I got into trouble with a former member of the commentariat here because of my Catholic objections to IVF (it was a personal family thing with them so naturally they did feel hurt).
I mean, I think the people currently proposing this do have the best intentions about healthy babies first (get rid of the hereditary diseases) and then 'who could object to smart, happy babies?'
But it's not going to stop there, because it never stops there. The particular genetic selection company may be the first one Scott heard of offering to sell hair/eye colour selection, but a couple years back there was at least one chain of fertility/IVF clinics offering this as a deluxe option for clients.
The fertility industry is an industry, and the purpose of business is to make money. So if people want purple eyed kids with silver coloured hair and six fingers on each foot (yes, I said "foot") and are seven feet tall, and the parents have the money to pay for it, that's what they'll get.
Wait, is your conception of the abortion debate that it's a debate between two sides who both have competing materialist conceptions of personhood? This would be a surprise to me, as I don't think anybody but the most hardcore of transhumanist/rationalist/atheists have a purely materialist conception of personhood.
The anti abortion side is a pretty straightforward metaphysical claim. The soul exists, and is created at conception.
Meanwhile the pro-abortion rights side contains dozens of competing definitions and thresholds of personhood, nearly all of them with some nonmaterialist component at their heart (e.g. "personhood only applies to living things" or "personhood only applies to humans" or "personhood only applies to fetuses post-viability" or "personhood only applies after it exits the womb" or "the threshold for personhood is whether the fetus has nervous system that can respond to external pain") The purely materialist conception (one example is purely physical capacity for information processing, e.g. IIT) is hardly held by anybody except a minorities of principled philosophers and rationalists. And to be honest I think even most of these are fundamentally metaphysical claims, with the possible exception of "personhood simply doesn't exist" (which I personally disqualify on the basis that it seems to imply that ethics don't exist either, which in that case, what's the point of arguing about the ethics of killing a fetus)
Do shrimps have personhood? Because Bulldog, for one, is very much active in ending shrimp suffering and getting us to treat shrimps as if they have rights.
As humans, we have a long, long history of treating "this set of us is fine and are people, those set of them are not fine and are not people so we can do what we like to them and that's okay, we're not doing anything wrong because they're not people".
Making personhood a revocable legal status that can be granted and taken away at whim is not a good precedent.
There's already a push for miscarriage leave here in Ireland (we don't yet have a legal entitlement to such, although there is for a stillbirth). So is a seven week old or twelve week old foetus a baby or not? If you want it, it's a baby and you are a grieving expectant mother if it dies. If you don't want it, it's a clump of cells and you are entitled to terminate that pregnancy and expel the contents of the uterus.
"Maybe baby" is not a great way to think about this entire topic, because if we're worrying about AI rights as future sentient beings, we're already laying the legal and social groundwork for "aw, that ain't no person if it's inconvenient for you for them to be a person".
And I don't think it'll stick at the foetus, either. We've done the "adult human but not a person" thing already in several instances, why not in the Brave New World now the unedited, unselected, stupid (because everyone is supposed to be 115 IQ which is the new normal, not a dumb old 105 IQ clod) human-looking animals are not in possession of personhood as an innate property of their humanity, so we don't have to worry about them. In fact, the kindest thing to do is put them down humanely, like we do with feral animal populations. They just can't have a decent quality of life under the circumstances that society demands of its citizens, they're too lacking.
I can't remotely join you on the idea that eight cells has the moral personhood of a newborn baby. I can clarify that IVF doesn't generally any longer involve implanting more than one embryo at a time.
I can also add the reality check that women have to still biologically produce the eggs to be fertilized and that the multiple and high levels of hormones used to produce that many eggs at once is an intense and painful experience and the numbers of eggs that can be produced in any one cycle is quite limited. The average number retrieved in each cycle is 8-14.
"Retrieval" is a surgical procedure under general anesthesia and the process leading up to it is hard and prolonged and demanding. Not all the eggs retrieved will be viable on their own (ie, before being fertilized), others will not grow after being fertilized. It's a long, slow process in other words, and one that does not yield up bunches and bunches viable eggs or embryos. It's extremely hard on women in particular, physically and psychologically.
After you replied, I addressed the question of the embryo's moral value in response to Scott's comment.
As for IVF and multiple implantation: noted, but I hope you realize that this reinforces my point about the seriousness of PS relative to standard IVF.
Regarding retrieval: yes, that's how it currently works, but there's a lot of effort going toward widening that bottleneck. For example, Jonathan Anomaly is banking on in-vitro gametogenesis so that "tens of thousands" of embryos can be created from a single vial of blood.
https://youtu.be/uvPDbOHSS4M?t=679
And will we then have "tens of thousands" of births? I'm thinking "Hell, no!" so we now have the potential for massacres in a test tube.
Ain't science wonderful! Shame our moral intuitions can't remotely keep up with our technological ability.
See how easily we've moved from "prevent preventable hereditary diseases" to "mass produce humans, mass murder humans, but all in the clean scientific way not the messy old King Herod way".
A poem by Charles Causley:
https://allpoetry.com/Innocent's-Song
Innocent's Song
Who's that knocking on the window,
Who's that standing at the door,
What are all those presents
Lying on the kitchen floor?
Who is the smiling stranger
With hair as white as gin,
What is he doing with the children
And who could have let him in?
Why has he rubies on his fingers,
A cold, cold crown on his head,
Why, when he caws his carol,
Does the salty snow run red?
Why does he ferry my fireside
As a spider on a thread,
His fingers made of fuses
And his tongue of gingerbread?
Why does the world before him
Melt in a million suns,
Why do his yellow, yearning eyes
Burn like saffron buns?
Watch where he comes walking
Out of the Christmas flame,
Dancing, double-talking:
Herod is his name.
Mostly orthogonal to your point, but I’ve met some couples who use polygenic risk scoring without destroying any embryos. They use the risk scores to inform the ordering of their IVF implantations.
One of these couples reasons along the lines of: “If my child has an elevated chance of breast cancer, then it’s better for them to be born later and hence closer to its cure.“
That's funny (but reasonable!). Do they actually plan to implant all of them? I've always wondered if there are ivf clinics (or patients) that are able to factor in the worldview that embryos have moral standing
For one of the two couples, I overheard them saying that they will implant all embryos at an Orchid Health presentation, because it’s important for IVF to be compatible with their religious beliefs.
The other couple is Malcolm and Simone Collins, they just posted about this on X recently.
beginning at fertilization, a human embryo has moral value that's similar in magnitude to a newborn baby […]
Consequently, that should make EAs very reluctant to advocate [embryonic destruction]
I don’t envision any reluctance: most Effective Altruists are hedonic utilitarians and I think would very straight forwardly reject your argument using their moral framework.
Hedonic utilitarians believe that moral value lies in the sum total of pleasure experienced minus pain experienced across time and space.
Early stage embryos (dozen cell structure) don’t feel pain or pleasure so under this moral framework destroying them is not directly immoral. In fact, hedonic utilitarians will reason that under resource constraints, it’s better to selectively choose the embryos that create people likely to have higher hedonic utility.
(disclaimer: I’m not a hedonic utilitarian)
ref your disclaimer, I'm curious, what are your personal views?
I disagree with hedonic utilitarianism because people and other beings can have lives that are net negative from a hedonic perspective, but they still find to be meaningful and worthwhile. It would be wrong to tell these people that it would be better that they were never born (they would disagree, even if a negative utilitarian would tell them they are “delusional”). Furthermore, I am even skeptical about whether pain and pleasure are inherently meaningful. After all, these are “arbitrarily” evolved signals from an ancestral environment we no longer inhabit. When we have the technology to edit such signals, should we in fact coerce all people into wire-heading in a pleasure pod?
Haven’t deeply thought through my moral philosophy, but I’m certainly a consequentialist, I highly value “survival”, and I consider “myself” to also be my descendants (e.g. my descendants are my New Game+, and I am a New Game+ for my ancestors).
I was wondering about coercive wire-heading, that seems to be the most obvious issue with hedonic utilitarianism.
There is also the question of what this evolutionary signal is _to_. Consciousness? If causing harm to others is bad in part because of this pain signal, it seems there is _some_ inherent meaning, although I wouldn't count myself as a pro wire-heading hedonic utilitarian either.
Survival for survival's sake is what cancer is though, I feel like this has to bottom out in "meaning" somehow. But maybe that is a hard-problem-of-conscioussness-complete problem.
"cancer == survival for survival's sake" is a fun analogy haha
I'll keep that in mind as I flesh this system out.
I don't think most EAs are hedonic utilitarians - most are utilitarian or utilitarian-leaning consequentialists (I think the 2019 survey was the last one to have this data, where just under 70% were utilitarian consequentialists), but I think further qualifying with 'hedonic' would cut that down a lot, at least in the traditional sense where experience is the *only* thing that matters (and even more so in the pleasure/pain sense you talk about below). I'm not aware of any solid data, but my guess would be that less than half of utilitarian or utilitarian-leaning EA consequentialists are pure hedonic utilitarians.
Thanks for your correction, and for pulling up the EA survey stat. Do you know how typical EA utilitarianians add more nuance beyond hedonic utility in this context? Words that come to mind include "flourishing", "personal preferences", and QALYs but not sure if this is how people would typically frame their utility functions.
I've never heard of the names Gene Smith or Jonathan Anomaly. Are those people's actual names or is it the title of some blogs? If the former, that's some serious nominative determinism.
Gene Smith is a fake blog name. IIRC, Jonathan legally changed his name to Anomaly but it is not his birth name.
I think it's worth mentioning that a very large number of embryos fail to develop even in natural pregnancy. I've seen people argue about the figures, so I'm not sure what the exact percentage is, but it's probably pretty high (https://pubmed.ncbi.nlm.nih.gov/28580126/).
This doesn't directly contradict your point, (embryo selection could still theoretically increase the number of discarded embryos) but it does confuse me. Would moral standing at conception suggest that we should *only* perform IVF and ban regular conception to avoid the associated death? (Given the current popularity of non-IVF conception surely it is the greater cause of embryo death?) Would embryonic personhood then be comparable not with selection, but CRISPR modification?
Right, it's repugnant conclusions like this - that in order to avoid creating naturally doomed embryos, the only ethical choice is invasive and alienating procedures like IVF - that are my strongest reason for not being too serious about embryonic value/personhood/rights. That's crazy.
I can see both sides of the argument otherwise. I suppose I come down in a gray area, where I'm uncomfortable enough with IVF for embryonic personhood reasons (and miserable maternal procedure reasons) that I probably wouldn't do it myself unless I were desperate - but there are plenty of people who are understandably desperate enough that I think the value they gain from it outweighs the small value/chance of personhood of the embryos lost.
Yes, this exactly. To me, the personhood argument is the ballgame.
And also, have you seen the movie Gattaca? Because that's the dystopian future we're taking about.
The movie Gattaca was a utopia for everybody except the protagonist.
If the protagonist's parents had been responsible like apparently everybody else and had a genetically screened baby, then he wouldn't have had any problems.
The movie focuses on his problem that he can't get the job he wants, but the underlying far bigger problem is that he has a genetic defect which may cause him to drop dead in his thirties.
The solution to "it sucks if some people aren't genetically engineered" isn't "nobody should be genetically engineered", it's "everybody should be genetically engineered".
The intuition that personhood begins at conception is one I can understand, even if I don't personally share it.
But it does run into the empirical observation that north of 50% of conceptions are spontaneously aborted (a point also raised by Charles Midi).
I would genuinely welcome some exposition on how you reconcile that with your feeling that conception is personhood.
At what point do identical twins become different people?
I asked ChatGPT o3 about it:
Q: Hi ChatGPT! I hope your early morning is going well. Can you please tell me how many clearly distinct arguments there are in the interminable abortion debate? I realize that these can be combined into a (roughly) exponential number of stances, but, for _just_ the distinct arguments, how many are there? Many Thanks!
A: There isn’t a canonical count—philosophers and legal theorists carve the terrain differently—but you can compress the debate into ~18 argument families. Most specific positions are combinations of these.
Core families (one-liners):
1. Biological humanity → full moral status. Human life begins at conception; killing it is wrong. (SEP overview.)
Stanford Encyclopedia of Philosophy
2. Psychological personhood criteria. Rights track capacities (consciousness, interests, rationality); early fetuses lack them. (Tooley.)
JSTOR
3. Potentiality / “future like ours.” What’s wrong with killing is depriving a valuable future; abortion (usually) does that. (Marquis.)
edX
4. Bodily autonomy. Even if the fetus is a person, no one may conscript another’s body (the “violinist” case). (Thomson.)
rintintin.colorado.edu
5. Responsibility for pregnancy. Voluntary risk-taking (sex) can create obligations that limit abortion claims; dissenters dispute this. (Discussion in SEP.)
Stanford Encyclopedia of Philosophy
6. Self-defense / double effect. Life- or health-saving interventions can be justified even if fetal death is a foreseen side-effect. (Foot; SEP.)
PhilPapers
Stanford Encyclopedia of Philosophy
7. Gradualism & thresholds. Moral status may increase with development (heartbeat, brain activity, consciousness, viability). (IEP/SEP surveys.)
Internet Encyclopedia of Philosophy
Stanford Encyclopedia of Philosophy
8. Viability as a legal/moral line. Before viability state power is weaker; after, stronger (contested and technology-dependent). (Roe/Casey/Dobbs.)
Justia Law
Library of Congress Tile
Supreme Court
9. Equality/anti-subordination. Compulsory pregnancy undermines women’s equal citizenship and life-planning. (Casey’s reliance interests; commentary.)
Library of Congress Tile
10. Privacy/liberty vs. state interests. Liberty (privacy) grounds a right to decide; opponents stress the state’s interest in “potential life.” (Roe/Casey/Dobbs.)
Justia Law
Library of Congress Tile
Supreme Court
11. Public-health / harm-reduction. Restricting access doesn’t end abortion; it makes it less safe and worsens health outcomes. (WHO, CDC, ACOG.)
World Health Organization
CDC
ACOG
12. Safety comparison. Legal abortion (especially early) is far safer than childbirth; policy should reflect that. (ACOG.)
ACOG
13. Moral uncertainty / risk. Given deep disagreement, precaution (for or against abortion) may supply defeasible reasons. (Moller.)
Squarespace
14. Fetal pain. Some argue pain perception grounds later limits; mainstream reviews place the onset ≥24–25 weeks and dispute earlier claims. (RCOG/ACOG.)
rcog.org.uk
ACOG
15. Disability-rights (“expressivist”) critique. Selective abortion can convey harmful social messages about disabled lives. (Parens & Asch and successors.)
PubMed
16. Sex-selection / demographic harms. Even if abortion is generally permitted, selection for sex is challenged on justice/inequality grounds. (Ethics literature reviews.)
University of Birmingham
17. Socio-economic consequences. Denial of wanted abortion increases financial distress and harms health; access improves life outcomes. (Turnaway Study, NBER.)
NBER
ANSIRH
18. Paternal interests & parental obligations. Disputes over the moral/legal claims of putative fathers and over duties to gestate/adopt-out rather than abort. (Surveyed in SEP.)
Stanford Encyclopedia of Philosophy
If you like, I can map these into a matrix showing how different policy packages (e.g., “early liberal, late restrictive,” “health-only exceptions,” etc.) combine subsets of these arguments.
Three possible objections I can think of that you don't discuss:
1. You mention how IVF is expensive and very unpleasant, but it might also just have negative effects on the embryo by itself (I know zeta_globin on Twitter is always banging on about this). If we're lucky the selection might yield a net improvement? I don't know, I don't really have enough knowledge here to make this argument or know how strong it is, but I've seen it out there so I thought I should mention it.
2. This is another argument I'm stealing from *someone*, but I don't know who (I remember asking in the comments here once and someone said it was the Unabomber, but I could not in fact find it in the Unabomber manifesto). So far it doesn't sound like this is allowing any sort of selection on personality traits (unless one considers IQ a personality trait). I'm mildly worried about the possibility of selection for agreeableness. That is, I'm worried that parents might select for agreeableness in the hopes of having an easier-to-raise kid, but that this occurring on a large scale might lead to increased conformism that holds back humanity. (Though selection on some other personality traits sounds pretty nice -- who doesn't want lower neuroticism or higher conscientiousness?) (I guess arguably this is covered under the "Selection" category, but it's a little different, I think, in that it's not purely individual.)
3. I can't find it at the moment, and this is even more arguably already covered under the "Selection" category, but I remember an old blog post by Ozy about the possibility of an even more extreme form of the downsides under "Selection" -- i.e., that awful parents might completely disregard their child's welfare in their selection (think of what's happened to certain dog breeds, bred for aesthetics). Like, abusive parents exist, so! Not sure how great that is as an argument because it strikes me as perhaps a bit too "criminals use cars", but thought it was worth mentioning (can anyone find the post? For some reason I can't find it right now!).
For 2, there are parents who would select both ways there: low agreeableness means rejecting groupthink and may favor economic success.
I expect extreme forms of 3 would be banned, or be covered by existing laws against child abuse.
Yes, #3 is a good point -- if this becomes generally used/accepted (and hopefully not banned), laws around it will (hopefully) be made to cover this sort of thing.
For #2 that is a counterbalancing factor, but if you get clustering then maybe there's a problem? OTOH, selection on higher IQ -- which is generally desirable -- would probably do a bit to mitigate that...
Natural forms of #3 are, to my knowledge, not illegal currently: AFAIK you are fully within your legal rights to have a child even if you know they will (or likely will) have a serious disability or fatal disease. For example, the couple in this article: https://www.nytimes.com/2022/11/09/health/pompe-disease-treatment.html
I was playing with the Herasight widget for various conditions (plus IQ), and it tells me the expected outcomes for 5, 10, or 20 embryos estimated by their genetic predictors. But I don't know what the implications of these values are. Does this imply they can select the best out of 5, 10, or 20 embryos to obtain the lowest liability (R-squared for risk variance) for that condition? I assume that's what they're claiming, but without a comparison of liability scores between the embryo counts. If not, what are they trying to prove with this widget.
They're not claiming that they can select the best with certainty, they're making a probabilistic claim that they'll do, on average, that much better than average.
If you haven't given them a real set of embryos, then they're simulating embryos with a normal distribution. For example, if two IQ 100 people marry each other, then on average their children will center around IQ 100 with some variance. I don't know exactly what that variance is, but they do and they've calculated it. Maybe on average, five embryos will range from 90 - 110. So if they could always choose perfectly, they could pick the 110 one and get you 10 points above average. But since they can't choose perfectly, they only expect to get you 5 points above average.
That's what I would have expected. But the widget tells me that with 20 embryos from 2 European parents whose IQs are 100, we can expect their child's IQ to be plus or minus half a standard deviation above or below the parents. The kid could have an IQ of between 93 and 108. I don't think they're claiming that, but somebody needs to talk to marketing about improving their value proposition. ;-)
> What if they selected your brother for everything great, then had you naturally?
This is basically the plot of Gattaca, except they had the first child naturally, regretted it, then used genetic selection for their second.
I was playing with their widget. It displays the values of the expected outcomes for 5, 10, or 20 embryos estimated by their genetic predictors. I presume they're claiming they can select the best of 5, 10, or 20 embryos—but it doesn't give me a readout of liability scores. So I can't really make a risk/benefit calculation. Am I missing something obvious here?
Another (tongue in cheek) objection. Your parents will be absolute morons compared to you at some point. You’ll be a mopey 14 year old screaming at your parents for not making obvious stock market picks and running their own companies.
On average, smart parents' children will be less smart than they are, because of regression to the mean. Currently, I think this technology just helps you tread water. I agree that this will be true in 10 or 20 years when the technology is much better.
I’m not sure what the overlap between below 100 IQ and IVF is, but regression to the mean + IVF would be a boon for a lot of people
People with 150 IQs and 100 IQ parents usually still love/like their parents and dont care much about their intelligence.
My understanding is that nobody actually has 150 IQ
150 IQ is 1 in 2000.
150 IQ is 3.3 SD above the mean. From an eyeball at a statistical table, 1 in 2000 people are this smart or smarter. And several of them may be reading this blog :)
see also: https://www.astralcodexten.com/p/464f7a04-1147-4eeb-8550-1608fcc78d92
Typically children of this sort are much more interesting in doing things in their own lives rather than controlling what they do with their parents lives. They're building full-flow 2 stage model rockets with landing capability rather than crying to their parents that they aren't supermen.
Teenagers feel this way about their parents regardless of the real relationship between their levels of intelligence, it'll be nice for them to actually be right about being smarter for a change.
The main thing I'm hoping for from this research is that it gets us closer to adult genome editing. It'll be really awkward if I'm in my 40s and there are 10 year olds smarter than me (Assuming AGI doesn't happen by then and render genetics completely moot).
> "but say that it most people gain 1-4 years"
The extra "it" seems to be a mistake.
If we're really concerned about improving our future children's genes, it's worth noting that compared to IVF and embryo selection choosing a partner with strong genetic traits (e.g. intelligence, health, height) would have a far larger impact.
The genetic variation between potential partners is much wider than the small variation among embryos from the same couple. So if you believe improving genetic outcomes is a moral good, does choosing a partner for love or chemistry over heritable traits amount to prioritizing our own happiness over our future child's potential?
I feel like some people already do this! Prioritizing someone hot/smart/healthy/wealthy over other people that might be nicer to you seems like a commonly discussed phenomenon
When i was dating many of my peers got genetic testing to screen for heritable disease that they might be a carrier for. There was a service that saved the results and you could then check potential partners for the same recessive disease. It would tell if you are both carriers but not what specific disease.
I think those traits are typically considered for their first-order effects on the relationship, or *maybe* for their relevance for providing during the reproductive years, not for their genetic effects on the offspring.
I don't think choosing partners for the genetic traits we want our kids to have is at all a commonly discussed phenomenon.
I sure never thought about it. Both of my serious relationships were with people who had siblings with relatively common heritable disorders. I was basically like "huh weird, I guess that's cool/sucks," and not once did it cross my mind that it could affect my potential kids.
Arranged marriages have been the norm throughout history. And still are the norm in many nations.
Yes, I can see how parents could be more scientific about what is best for their child and children's children
Even less data on multi-racial / mixed-race genomes. Multi-racial couples are out of distribution for these machine learning models and plausibly will get the least value from these embyronic screening services.
We don't currently understand much about genetic interactions, either because they rarely happen or because our current tools aren't good enough to detect them. Since embryo selection can only use the genetics we understand, interactions aren't relevant here, and predicting mixed race couples works fine using what we know about the component races.
So for example, on Herasight's calculator, an Asian couple with five embryos can gain 4.1 IQ points, a white couple can gain 5.3 IQ points, and a mixed white-Asian couple can gain 4.8 IQ points - right in between.
Can't see this leading to anything but dystopia if it pans out. Saying that about a lot of new developments nowadays.
https://dresdencodak.com/2009/09/22/caveman-science-fiction/
This is very horrible. Ghastly!
Warning, commenting policy here is that you have to provide arguments for an assertion, please do this in the future or don't comment.
I don’t care what your policy is. You posted on a public forum. I’m free to respond. If you don’t like others’ opinions, don’t post.
And he's free to ban you for being a stupid jerk. My argument for the assertion that you are a stupid jerk is that you didn't provide arguments for your assertion, refused to do so when he asked you to, and said you don't care about his preferences.
It’s an OPINION.
I didn’t make a factual claim. Jesus, how can people not distinguish between an opinion and a factual claim?
You can’t cite research to prove whether something is ghastly/ghoulish.
I think it’s ghoulish. I have a right to think that, and to say that.
He has a right to block me, if he dislikes my opinion. I really don’t care. I made my statement.
I think making designer babies is a very bad thing that will lead humanity into a very troubling future.
I don’t know if you’ve ever read Brave New World, but perhaps you should.
Opinions still have an internal logic to them, unless you are claiming to have arrived at yours by rolling dice. If you hope to convince people to adopt yours, you will generally need to engage with that internal logic. For example, you could say *why* you think this will lead to "a very troubling future", or what you think will be troubling about it.
As it happens, I *have* read Brave New World, but to my recollection it objects to nonintellectual hedonism and the creation of a low-intelligence underclass. If you also object to creating healthier, more intelligent people, and think Brave New World is criticising that goal, them I think you have misread the text.
>the creation of a low-intelligence underclass.
The creation of a high-intelligence upperclass implies the creation of a low-intelligence underclass, even if no active measures are taken towards the latter. Implying it's an intentional, active process may be a POSIWID fallacy because nobody is proposing to bathe embryos in an alcoholic solution or whatever exact mechanism BNW offered, but the overall result is bound to happen anyway.
It turns humans into products. It creates rigid castes divided by intelligence level. It eliminates family structure. It’s a horrifying dystopian world that we are running to as fast as we can.
If you do not understand that BNW was critical of producing babies as a product, instead of as a part of a family, I think you misread the text.
How do you feel about selecting for implantation, out of two embryos, the one which does not have trisomy 21?
How would you feel about magically removing the third copy of 21 instead?
You're missing the point. People who need to use IVF because they cannot naturally conceive a baby will of course seek to have a healthy child if possible. But that's not what this whole piece is about: It is about people deciding to put a woman through the painful and difficult IVF process just so they can get an athletic kid with blue eyes or whatever.
It's really disingenuous to pretend trying to have a healthy child free of a terrible disease is the same thing as choosing a kid who is tall.
And to use IVF tech when there's no infertility problem is ethically questionable.
There are dozens of people with your opinion participating in the conversation here. Nobody cares that you have this opinion, but this space is used for discussion. The reaction you're getting is not because anyone disagrees with you or feels you shouldn't be allowed a voice, but because you're treating this like a facebook post. You can think and say a lot more here than most other places; you're just expected to contribute signal, not meaningless noise.
From your comment, I infer that you don't support a woman's right to choose which embryo is implanted into her body... is that correct?
You went off into that direction all by yourself.
but you still didn't answer the question... if you don't support a woman's right to choose what goes into her body, just say so.
(in case it's not clear, this is, indeed, a rhetorical device, pitting one purported value against another in an attempt to get you to make a coherent argument, rather than just being enraged)
An issue that I have yet to see addressed well enough is how sure are we that there is no added developmental risk from the procedure itself that would counterweight the benefits, especially when the benefits are "add an expected 4 IQ points" rather than "not have CF". The risks I'm thinking are mostly:
1) Stress to the embryo from the sampling, either in inducing developmental mispatterning (more likely) or de novo mutations
2) Stress to the embryo and the mother from the IVF (development/mutations for the embryo, hormonal and immune signaling mess for the mom), if comparing against in vivo fertilization
There just seem to be so many risks there that even if each one is very tiny change in odds because doctors optimized it excellently, they should add up to more than 4 IQ points.
Yes, these are real life risks. I don't know how big they are.
Polygenic selection isn’t just a tweak to medicine or parenting. It’s a decision about who gets to exist, and why. But this conversation keeps retreading familiar grooves without confronting the ontological vertigo this new technology introduces.
We’re not choosing a school or a tutor; we’re choosing which possible person becomes our child. That crosses a metaphysical threshold. Potential humans are now compared probabilistically—not for life or death, but for entrance into life itself. And this comparison is based on traits that increasingly resemble KPIs: IQ, BMI, schizophrenia risk, even eye color. That's person-as-product, full stop.
We need plainer language for what’s happening. The core ethical shift isn’t “screening embryos” or “preventing disease.” It’s that the value of a life becomes a function of its forecasted traits. And you know once that logic takes hold, it won’t stay confined to embryos.
The risk isn't Gattaca-style coercion. I'm more worried about soft cultural convergence. No laws required. Just dashboards, rankings, and best practices. And because it feels like choice, not imposition, no one will notice the narrowing until it’s a fait accompli.
Maybe this tech does increase aggregate well-being. But it also reshapes the moral substrate beneath our choices. When you start selecting children the way you select college applicants, your deciding what kind of soul you think is worth betting on.
Jesus H. Christ.
Is this also true when you take prenatal vitamins to prevent spinal deformity? Isn't that selecting for children who don't have spinal deformity over those who do? Or that the value of a life without spinal deformity is more than one with it? Should we ban prenatal vitamins?
I think there’s a category error here. Taking prenatal vitamins supports the development of a child who already exists. Polygenic selection chooses which possible child gets to exist at all. The former is about nurturing; the latter is about filtering.
Once we’re filtering not for catastrophic disease, but for probabilistic forecasts of intelligence, temperament, or BMI, we’ve crossed from medicine into metaphysics -- defining what kinds of human potential are “worthwhile.” That ain't just banning vitamins.
What happens when the logic of actuarial triage becomes our primary mode of making people?
What you call "catastrophic disease" is also a value judgement. It's a very common value, but saying it's better for a child not to have a spinal deformity is just as much a decision about what kind of lives should be lived as selecting based on predicted IQ.
Not all value judgments are equally loaded. Shifting from healing a child to curating one begins to apply a logic of comparative worth to potential lives. If we’re going to do that, we should at least be honest about what it is.
I'm not sure what you mean by "loaded", but keeping your child from being born with a spinal deformity is absolutely applying a logic of comparative worth to potential lives. At the time you make the decision both lives (with spinal deformity and without) are merely potentials, and your decisions change their relative likelihood of being realised. People act on this because they think the life without the deformity is better than the life with.
Yes, choosing to avoid spinal deformity involves a value judgment. But there's a difference between reducing the risk of profound suffering and optimizing for predicted traits like IQ or BMI. One stems from a widely shared ethic of care; the other starts to encode comparative desirability into existence itself.
That’s what I mean by "loaded": some judgments operate within a shared moral horizon. Others begin to redraw the boundaries of which potential lives are seen as worth bringing into the world.
Scott, the problem is lower intelligence is not a disease.
I don’t think the argument is that the lives of children with spinal deformities are worth less than other children. I think it’s that it would be preferable to live in a world where children with spinal deformities can receive appropriate treatment and accommodation, as opposed to a world where those children simply do not exist.
This tech proposes to “solve” a “problem” by engineering a situation where the “problem” doesn’t exist. That’s fine for some things, but I don’t think it’s crazy to get uneasy when the “things” in question are people.
What’s weird is that this tech simply lets parents choose one randomly generated embryo over another. If we assume there’s no a priori reason to choose one over the other, it doesn’t seem crazy to let the parents’ preferences break the tie. And right now, these preferences are things like “maximize the chance of a pregnancy that’s successful and safe for the mom” and “minimize the expected health care costs over a lifetime”. Not unreasonable!
I do think parents could be a little more honest with themselves about how much these choices are being made for their own sake, rather than their prospective child’s. It’s fine not to want to go through the gauntlet of raising a kid with a rare genetic disorder: I image it’s a huge emotional and financial weight. But probably don’t say the choice is for the child’s sake, because it’s not hard to see how that could shade into saying it’s best for such children not to exist at all.
In my opinion everything gets way more fraught if you start designing embryos from scratch. That’s not what this tech does, so I’m mostly agnostic about it. But I do get worried about slippery slopes, because humans haven’t had the greatest track record in this domain.
>I think it’s that it would be preferable to live in a world where children with spinal deformities can receive appropriate treatment and accommodation, as opposed to a world where those children simply do not exist.
FWIW, I have the opposite preference. Treatment and accommodation have their limits.
Indeed, there is a spectrum of conditions along which I think reasonable people can have differing preferences. A spinal deformity is perhaps not the most interesting example for considering the ethical consequences of this tech, due to the nature of the associated treatments/accommodations. But what about deafness? Or ADHD? Or left-handedness?
(To be clear, I’m not trying to play “gotcha” here, I’m just trying my best to pinpoint the crux of the disagreement on this topic.)
Many Thanks!
( FWIW, I, personally, happen to be childfree and long past my potential reproductive years, so I don't actually have a dog in this fight... )
I think the best public policy course of action is to leave these decisions up to the prospective parents. As you said
>Indeed, there is a spectrum of conditions along which I think reasonable people can have differing preferences.
Hopefully, the parents have their child's best interests at heart, and will weigh all the possible factors, from catastrophic ones like severe spina bifida to near-neutral ones like left-handedness, in that light. If they _don't_ , the child is screwed anyway, and there is very little that the State can do to remedy the situation. ( Yeah, child protection services exists - and regularly screws up in both directions. )
"And you know once that logic takes hold, it won’t stay confined to embryos."
Why should it? It's factually correct. This is the logic used by the economic system which has produced prosperity unimaginable for the vast majority of human history.
Market logic has produced staggering material gains. It’s also optimized away biodiversity, cultural memory, and non-commercial forms of meaning. Useful doesn’t mean neutral. When that same logic starts governing which kinds of people get to exist, we’re not just scaling efficiency; we’re rewriting the criteria for entrance into life.
If you think the logic that brought you cheaper iPhones should also dictate the selection of human souls, that’s…a position. But let’s not pretend nothing changed.
In any case, the point wasn’t that the logic is flawed. The point was: once it’s normalized at the level of reproduction, it will reshape what we mean by a “worthwhile” human being.
Perhaps we ought to mull that over.
>it will reshape what we mean by a “worthwhile” human being
Sure, like it has been continuously happening for centuries.
>Perhaps we ought to mull that over.
OK, what is your positive vision for the future (which presumably excludes this particular development)?
True, what counts as a “worthwhile” human has always been contested and evolving, which is precisely why we should be alert when a new mechanism emerges that may lock in a narrower definition under the banner of optimization.
As for a “positive vision”: I don’t want to halt innovation, and I’m not against reducing suffering or helping parents make informed decisions. But we need to preserve a moral space where unpredictable, unoptimized, even “low-probability” human lives are seen as fully valid -- not just tolerated, but valued. In other words: a future where technology supports life, rather than presorts it.
That will require drawing some philosophical lines before the dashboards fill in automatically.
>which is precisely why we should be alert when a new mechanism emerges that may lock in a narrower definition under the banner of optimization
Well, there are few stronger taboos in the modern world than against "eugenics", so I very much doubt that if this manages to squeak through it would be through lack of alertness.
>That will require drawing some philosophical lines before the dashboards fill in automatically.
Philosophy's track record doesn't inspire much confidence in any lines it would produce, but of course that has never stopped philosophers.
The taboo against “eugenics” exists -- mostly in name. What we’re watching now is its rebranding as choice, optimization, and personal empowerment. Norms don’t tend to collapse through brute force, but through a gradual shift in what feels reasonable.
If your view is that ethics can’t help and vigilance doesn’t matter, then there’s nothing to discuss. That’s a shrug, not a position.
I expect to see capitalism apologia on here as a matter of course, but sweet Jesus do some of you lay it on thick.
I mean you already can make a decision about the characteristics of your baby that goes much further than polygenic selection: By selecting your partner out of 50% of the world's population. By choosing to have a child with one person, you deny existence to all the babies that could have been born had you chosen someone else. Of course partner choice is probably rarely made just based on gene-based traits one wants to have in one's offspring, but it certainly plays a role.
However I can see where you're coming from. One parallel might be pets, e.g. dogs: Anyone can, and most do, choose a dog breed they like from a wide range available. And I guess dog owners are really happy with that and love their dog and want it to be happy and healthy, nothing wrong with that. However, the problems start when dog owners have such a strong preference in characterisics that they choose breeds that have well known health problems, like french bulldogs or whatever. There the characteristics that the owners like trump the health concerns of the dog.
What if deaf parents lived happily in a deaf community, and select the one embryo out of five that is also deaf, so their child can experience life as they do? Would that be morally right, as most people would percieve deafness as a severe disadvantage in life? Or if the parents really, really want a blue-eyed baby, and choose that one blue-eyed embryo despite a 50% chance of cancer? Also looking at the parents that send their kids to those extreme child beauty pageants, strong guardrails will probably be have to be implemented should this tech get any further, to prevent a lot of suffering in those edge cases.
Your buffet of analogies sidesteps the categorical difference between influencing an existing person’s life and choosing which possible person gets to exist based on ranked traits.
Yes, selection happens at every level of reproduction, partner choice included. But polygenic selection crosses a new threshold by ranking potential humans by algorithmic desirability before they’ve been allowed to exist.
This has now become an existential decision based on probabilistic predictions of cognitive, behavioral, and physical profiles. Let’s at least call it what it is, rather than normalizing it through analogies that don’t fit. We're not choosing a dog breed.
> But polygenic selection crosses a new threshold by ranking potential humans by algorithmic desirability before they’ve been allowed to exist.
Do you mean polygenic selection would lead to the single alogrithm that gets applied to all humans, and only those with the maximum number of points would be allowed to exist? It's the parents who make the choice, and they do so based on their own personality and preferences. They get presented with a table of x embryos vs y probabilities of characteristics from disease to IQ, and make a choice based on their perferences. I'd say it's very much like selecting a dog breed.
The parents make the choice, but the fact that it's individual choice doesn’t negate the nature of the decision being made.
We're no longer just nurturing a child or even choosing to conceive at a particular time. We're comparing multiple potential humans based on statistical forecasts and deciding which one gets to exist -- not based on catastrophic disease, but on probabilistic desirability across a range of traits. That’s fundamentally new.
It may feel like choosing a dog breed, but dogs aren’t people. They’re not moral agents. They’re not asked to find meaning in lives they were selected into based on marginally higher predicted IQ or lower BMI.
So no, I don’t think “it's just like picking a dog breed” holds up once you follow the implications through. It’s a flattening analogy that deflects from the deeper shift underway: from accepting the unpredictability of human life to curating it in advance based on optimized projections.
This redefines what it means to begin life.
What about deaf parents who refuse the MMR vaccine for their child, hoping that Rubella renders them deaf?
That’s a category error dressed as a gotcha. Pass.
Did I misunderstand your objection? Vaccination is a decision which influences an existing child.
Personally I think it's morally objectionable to deny a child vaccination because the possible outcome is one which I would call worse, even though the parents would call it better. My understanding is that in your framework, that would be okay (I'm pretty sure I disagree with that, but maybe you have a point I haven't considered).
But I certainly don't share this intuition that excluding one embryo amongst 10 is any worse of an intervention than vaccination.
I guess I can understand the reservations around genetic diversity, and I can understand the reservations for folks who believe that life begins at conception, but your reservation seems to be based on personal outrage, and I'm not sure where it's coming from.
Appreciate the clarifying effort. But no, I’m not defending MMR refusal. I flagged your example as a category error because it conflates influencing an existing child’s health (via action or omission) with choosing which potential child gets to exist in the first place.
My concern isn’t rooted in “life begins at conception” theology, or in squeamishness about medical influence. It’s in the shift from nurturing unpredictable lives to selecting among predicted ones based on ranked desirability.
Vaccination aims to protect a child already in the world. Polygenic selection, at scale, begins to define what kinds of humans are worth bringing into it. That may not feel morally weightier to you, but to me it marks a foundational change in how we relate to human life: not just caring for it, but pre-sorting it.
Except that "ranking potential humans by algorithmic desirability before they’ve been allowed to exist," does not meaningfully differ from "partner choice based on gene-based traits one wants to have in one's offspring." They're different phrases that mean the same thing: deciding beforehand what kind of children you want.
We are, in fact, choosing our 'breed' of offspring every time we decide to have children with one person and not another. We don't always know what that strain will produce, but we definitely make inferences on the basis of pre-existing traits.
The reason we currently accept the unpredictability of human life is because we cannot predict it, not because it should be fundamentally unpredictable in order to remain human. However objectionable you may find the mechanism, a change in kind has not taken place - only one of degree.
A more pertinent question would be whether we have the right to deny people the foreknowledge of genetic information, if it could be used to improve their children's lives. I don't think we do, but I'd be interested to hear your take.
Choosing a partner reflects implicit preferences, including genetic ones, but that’s not the same as explicitly ranking embryos by predicted traits and selecting accordingly. One is inference under uncertainty, the other is filtration by forecast. That’s not just a shift in degree. It’s a shift in kind.
I don’t object to foreknowledge, but to what happens when it’s used to filter human lives before they begin. The right to know doesn’t automatically imply the right to act on that knowledge in ways that redefine who gets to exist.
But people absolutely DO rank their partners on the basis of genetically-transmitted traits, with predictable outcomes for children. Tall, smart, blue eyes, etc. A shift from uncertain conditions to more certain ones is not a shift in kind, because in either case we try to make the same selections in advance.
We act on limited knowledge to redefine who gets to exist every day. The alternative is refusing screening for horrible congenital diseases because we believe that a random affliction is more moral than a conscious prevention, or actively acting against our own trait preferences for sexual partners. That's not a reality I'm eager to indulge.
When you fall in love, or even just choose a partner, you’re not running a regression on their genome. You’re drawn to a person, a whole person, whose traits emerge through culture, chance, hardship, and a lifetime of adaptation. If you hope your future child inherits their curiosity, resilience, or kindness, that’s not “ranking potential humans.” That’s loving someone and hoping what you love continues.
Embryo screening is something else entirely. It takes a cluster of possible humans and scores them on a grid of forecasted traits—IQ, BMI, mental health risk, even “educational attainment” -- and selects accordingly. That’s not preference under uncertainty. That’s predictive exclusion.
Let’s call it what it is: a shift from attraction to optimization, from courtship to computation. It doesn’t just fine-tune a familiar process; it industrializes the logic of better and worse humans.
And yes, it will change us. The tools we use to make people will reshape what we expect from them and what we think counts as a “good” life. If you think that’s just a difference in degree, I don’t think you’ve really looked.
This reads as pure nonsense to me even as someone who has misgivings about this. If you jerk off one additional time before impregnating your wife you're choosing not to let certain "potential humans" be born based on no useful criteria at all. There cannot be value to "potential humans" without ending up with completely ridiculous moral conclusions. Humans who never got past being a clump of cells are not and never were people, and have no moral value.
Libertarian eugenics can create serious issues just like authoritarian eugenics is almost certain to, but the issues arise from increased inequality and unforeseen consequences from removing certain parts of genetic diversity.
Not exactly opening with Socratic grace, are we?
The point isn’t that every potential human has moral status. It’s that the criteria we use to decide which potential humans become actual ones reflect and gradually reshape our values.
Jerking off doesn’t involve a value-laden decision between ranked embryos scored on predicted IQ, BMI, or psychiatric risk. Polygenic selection does. The novelty isn’t selection per se -- it’s the filtration of future lives through actuarial models of desirability.
You don’t need to believe embryos have full moral status to see that shifting our selection logic from randomness or love to statistical optimization exerts long-term pressure on social norms, parental expectations, and even our idea of what counts as a full human life. The stakes aren’t just biological. They’re civilizational.
“I’ve decided to use this technology” Bros gonna reroll his children 😭🙏
Typo: "well, there’s still an outside change that" - probably meant chance?
I think I agree that the immediate first-order benefits of IVF embryo selection (and by extension the type of more targeted gene editing discussed in the linked post by Gene Smith) are likely to be extremely positive, considering how much low-hanging fruit there is, even if considering only the disease risk reduction.
Still, I find myself more concerned about the long-term consequences. Gene's post on LW makes an analogy to the development of the modern broiler chicken, which seems to me more alarming than he lets on. Inevitably, if these techniques develop as Scott projects, it will lead to a narrowing of the human gene pool over time (though probably not to a single point, as I imagine there would be some clusters of coordinated traits that are favorable among different values systems).
Slightly closer to home, I look at my purebred yellow Labradors. Fantastic companion animals, and certainly not the least practical breed out there. But I cannot imagine a population of labs surviving in the wild under more adverse conditions. How many generations of hardcore gene selection does it take until we're the same? How robust is that population (to X-risks) compared to the current, naturally-selected order?
Reading some of those objections I'm reminded of that old tagline from Gattaca "There is no gene for the human spirit." A sentiment to which we can now reply "No, but there's probably about 10 thousand SNPs correlated with it."
I've never really understood the objection that with the development of AI there's no point in trying to improve human abilities. Setting aside the fact that I'm very much not convinced that AI is just around the corner, even if it is I still see the value in humans. A diversity of minds has the same benefits as a diverse ecosystem, each organisms filling and exploiting its niche. It would be incredibly short-sighted of us to neglect the development of different mindtypes.
While there's lots to get to, I'm disappointed you didn't even brush up against my main ethical concern about this technology, which is diversity. No, not the race thing -- I mean real diversity. I like living in a world full of vastly different people, with different body types, different tempers and intellects, different tastes in ice cream. A democratization of designer-babies would bring with it a homogenization of the human race towards a consensus 'optimal' form, and that just creeps me the hell out, like a lower-rate version of tiling the universe with paperclips. The natural endpoint of this technology seems to be a world populated entirely with tall, mathy, conventionally handsome, allistic, sociable people. Depending on the way society's blowing they'll either all be heterosexual, or all bisexual. By god, what a waste of a planet.
When I imagine a glorious transhumanist deep utopia, it looks more like Wonderland than Heaven; it's a world like the Culture; it's a world of endless variety, where hardly any two humans have the same skin color and the person next to you on the bus might not even have the same package of *senses*. Granted, modulo AGI which gives adult humans the ability to shapeshift, this sort of technology would be necessary to get something like that. But I don't think the humanity of 2025 can be trusted with it yet. If this goes big before something else changes the course of destiny, I think it's all too likely we'll erase all the interesting but unconventional or stigmatised variations, the left-handedness and the autism and the obligate monosexuality and the unibrows, from the genome long before we grow wise and realise how stultifyingly *boring* it's all become now that everyone is a sporty Von Neumann.
And there's another, related fear. I don't want to go extinct. Maybe *you* want your kids to be smarter than you, but I don't, particularly. There are simple pleasures I wish I could enjoy that bore me to death, because my mind won't sit still; jokes I don't laugh at because I can see the punchline coming a mile away. And sure, I don't know the thrill of winning a high-level chess game against a grandmaster; but that's just it. It's all a treadmill. I don't want our supergenius children to find all our jokes boring, our music simplistic and our novels obvious. I want children to whom we can pass our culture on into eternity; children we can laugh with. Not children to whom everything that defines who we are will look like Dora the Explorer. I think it would do them, and us, a great disservice.
> I like living in a world full of vastly different people, with different body types, different tempers and intellects, different tastes in ice cream. A democratization of designer-babies would bring with it a homogenization of the human race towards a consensus 'optimal' form, and that just creeps me the hell out, like a lower-rate version of tiling the universe with paperclips.
Thank you! I completely agree. There are many traits that probably aren’t “optimal” for maximizing socio-economic status or whatever, but society needs some % of them to thrive long-term. Tortured poets, revolutionaries, etc. Anything that reduces genetic diversity risks “optimizing” these people away. (I’m a diagnosed autist, and I don’t want to “optimize” *me* away!)
How do these concerns not apply equally to giving babies prenatal vitamins?
You're going from a more diverse state (some babies born with deformities from lack of vitamins, some healthy) to a less diverse state (all babies born healthy). And if you have a minor vitamin deficiency, then moving all kids to the no-minor-vitamin-deficiency state is driving "your type" extinct.
I think I just decide that diversity in whether people have deformities and diseases isn't the kind of diversity I care about. I care about the kind of diversity where some people are Einstein and some are Taylor Swift and some are MLK. But all of those are healthy, high-IQ people who didn't die of spina bifida at age 4.
And how, exactly, do you intend to have a glorious transhumanist future where everyone is a different type without any genetic technologies? I'm actually more concerned about this than excited - I worry at some point they'll come up with a predictor for extraversion, and half of parents will select in favor and half against, and then things will get pretty weird.
My main answer is that this technology at its current level of efficacy/cost isn't enough to cause broad social changes. And if you extend it outward until it is, we're already in the glorious transhumanist future (or at most one generation away from it, after the hyperselected people grow up and invent transhuman technologies) and we'll pass off the problem to them.
I still think the equation of raising IQ with curing obvious awful diseases is fallacious. It's not obvious to me that, past a certain threshold enabling e.g. literacy, smarter people's lives are better or more worth living; you're just moving the hedonistic treadmill around, while reducing the variety of minds that exist.
I also want to stress that my two points are separate, even if they come from similar places, and that my concern about "my type going extinct" was specifically cerebral (i.e. "I want there to always be people who think sufficiently like me to appreciate the same kind of art I liked and created"), as distinct from my worry about from the overall loss of variety incurred by erasing pattern baldness or whatever.
Relatedly, you seem to have side-stepped the risk of traits which are not objectively harmful, but are socially stigmatized, being selected against. It's one thing to be maximally spicy and say "we shouldn't select dwarfism and deafness out of existence", but my worries apply even without that sort of thing. e.g. unibrows, (non-debilitating) autism, etc. There are plenty of idiosyncrasies that are not deformities or disabilities, but which parents, for various, shallow reasons, might disproportionately choose against passing on.
> And how, exactly, do you intend to have a glorious transhumanist future where everyone is a different type without any genetic technologies?
I did raise that very point in the first post. You're right, of course. It's just that, like that *other* technology that could bring about a glorious transhumanist future, it could ruin the world as much as it could save it, and I'm not sure current humans are wise enough to use it well yet. If embryo selection remains the *only* transhuman tech for the foreseeable future, I don't think "well, that'll be our transhuman children's problem, they've got this" is a get-out-of-jail-free card. I wouldn't particularly expect the hyperselected transhumans to be wiser just because they're smarter in terms of raw IQ. Potentially they just get stuck in a bootstrap loop where the kind of types that are selected for feel increasingly icky about people who display non-selected traits, and have progressively more homogenized children until the idiosyncratic genes are lost - with the unselected humans who felt the skin-crawling horror I feel now dying of old age before they can grind the process to a halt.
Obviously the "we'll hit AGI first" argument defeats this concern, but then, as you said, it kinda defeats all concerns. I think conversations like this one are best carried out with the understanding that we're talking about the "for some reason AGI takes another hundred years/isn't possible/etc." scenario, not one where extending this particular technology to infinity has to also interact with other technological breakthroughs that would happen in the meantime.
>I think conversations like this one are best carried out with the understanding that we're talking about the "for some reason AGI takes another hundred years/isn't possible/etc." scenario, not one where extending this particular technology to infinity has to also interact with other technological breakthroughs that would happen in the meantime.
Agreed. AGI (or at least ASI, if possible) overwhelms virtually all other concerns.
If embryo selection (and, possibly, genetic engineering of some sort) becomes pervasive, for generations, I share your concern about diversity being selected out.
I agree with Scott's
>I care about the kind of diversity where some people are Einstein and some are Taylor Swift and some are MLK. But all of those are healthy, high-IQ people who didn't die of spina bifida at age 4.
I expect that the long run effect of parental genetic selection wouldn't preserve that degree of diversity. I'd expect that, e.g. we would get _good_ predictors of the big 5 personality traits. And there would be a point in that 5-D space which predicts good life outcomes. And the large majority of parents would choose genes which predict close to that outcome. And the distribution of traits would narrow in the direction of a delta function at that point.
My main reason for not being concerned about this is (a) it will take multiple generations, perhaps even centuries (b) yeah, I really _do_ expect AGI to happen first, maybe 2-10 years, barring an antarctic level AI winter.
EDIT: Re personality traits, currently the effective dimensionality of the 5-D space, while not quite the full 5-D, is a bit over 4 (from ChatGPT, Claude agrees)
https://chatgpt.com/share/688dc25b-6940-8006-8483-bef9271cb775
Even if pressure towards the optimal choice doesn't push it to a 0-D point (there are a range of socioeconomic niches...), I would be surprised if it doesn't at least collapse to a lower-dimensional attractor.
Your mistake here lies in predicting that choice in heritability leads to sameness. The genetic variance of a given population will not necessarily decrease with a loss of deleterious mutations, and traits that influence genetic variety are found within particular groups, so there's no worry about the loss of unibrows any time soon.
Additionally, you seem to be assuming that every trait will be optimized and selected for, even though we don't find this in practice. Genetic variation is hugely influenced by selective choice, and there is some evidence to suggest that this is genetically-determined and therefore resilient. The fact that you or I might want a tall, mathy, conventionally handsome, bisexual god for a partner doesn't mean that someone else might not have a strong preference for a medium-height, artistic, conventionally handsome (this probably won't change), heterosexual. So long as choice remains free, beneficial variation will persist.
Hi Scott, thanks for the thoughtful post. I was surprised to find myself very convinced, but then reflecting more on my un-ease, I think there is an ethical dilemma that haunts me, and I was curious your thoughts.
I'll assume that in a couple decades we'll have much stronger predictive power than today on a wider range of traits, in order to dispel objections around the efficacy or current menu of options.
Ethical Objection: Arms Race
Diving into "Ethical Objection: Cost," an underlying assumption seems to be that people's nominal choice is indeed a true and freely made choice. If we give people the means to choose, then they can choose freely whether or not to choose IQ of their embryo. But, given how competition is fundamental to human life, if your child's peers are getting ahead with genetic enhancements, then do you really have a choice? This looks a lot to me like an unavoidabe start of an arms race? Those who choose not to participate will be dooming their progeny to failure competing in society. To exacerbate the issue, I would imagine that the positive traits can compound, generation over generation.
I am sold that the category of traits control mere preferences (say, eye color) seem harmless, and traits that fight disease are good. Consider the category of "enhancements" that will make your child more competitive in society (whatever societal context you participate in). Does this lead to an arms race? If so, is that a good/bad/neutral thing?
If productivity increases enough, for this or other reasons, we'll be able to afford UBI for all anyway, and so the need to compete is reduced.
Keynes thought that we'd develop so much productivity his grandchildren would work 15 hours a day: https://direct.mit.edu/books/edited-volume/2712/Revisiting-KeynesEconomic-Possibilities-for-Our
Karl Marx argues that the surplus labor should be socialized back to leisure. This did not come to be, and China is one of the most competitive societies in the world now.
Deep thinkers from every end of the spectrum of economic theory have predicted that we'll have reduced competition, and more leisure for generations.
I'm starting to think it ain't gonna happen.
Evolutionary biology seems to have woven the need to compete and progress and race into our fibers.
By most accounts, hunter-gatherer societies had a lot of leisure time. We already are plenty productive, that's not what's stopping us from achieving this!
You could imagine some government presiding over a single country and limiting leisure, but as long as some people want a leg-up, or to get ahead, or to push their social status, or to indulge their ego, there will be competition, domestically. Internationally gets even crazier -- Without a credible threat of force from a governing body, no society would feel comfortable "being lazy" while another potentially maniacal society (bloodthirsty dictator, democracy with ethnonationalist ambitions, etc) can out-compete and threaten you.
On a related note, collectivist societies are more competitive rather than less https://x.com/robinhanson/status/1949283054058443218
> If this were a single-use medical treatment, delivered by a doctor after someone got the relevant condition, it would be one of the biggest advances of the decade - imagine a drug that cures 10 - 40% of breast cancers with no side effects! But in fact, it works for breast cancer, and schizophrenia, and heart attacks, and approximately everything else. The only things comparable are antibiotics and GLP-1RAs.
Vaccines too, aren't they?
Also, not to minimize the benefit of embryo selection if it works, but one difference from a drug that cures 10 - 40% of breast cancers (and other diseases...) is that it's like a drug that has to be given to the entire population, including those who don't have an won't go on to have breast cancer, in order to cure 10–40% of those who do get it, so this increases the cost (and, if you wouldn't otherwise use IVF, discomfort) per case cured — kind of like vaccines. Actually this may also be a reason there's more controversy around vaccines than other drugs: they are given to healthy people, whom it's not yet known if the vaccine will benefit, so people tolerate less risk of side-effects than someone who already has a disease and wants to get rid of it.
----
((EDIT: already mentioned by other commenters:) One thing I didn't see mentioned: last time I've read about it, IVF itself had some moderate risks to the embryo; what's the situation with that now, and how does it compare to the benefits of embryo selection? This is relevant to those who consider IVF only in order to do embryo selection.)
Another minor point: Do the companies pay attention to avoiding inadvertent sex selection? If, say, breast cancer is more common than prostate cancer, or if they are worse at selecting against it, that could accidentally lead to the embryo with the lowest risk of the diseases they select against disproportionately being male. Widespread sex selection could cause problems (some people of the more common sex won't find a partner); here in the West I don't think most people have a strong sex preference about their children, but if we don't consider and avoid it, embryo selection could accidentally lead to sex selection even if nobody wants it.
> It depends on the number of embryos they have to work with; the more embryos, the better you can do by selecting the best.
You can do even better by selecting gametes from individuals who have the traits desired. If if you want to have a kid who is good at tennis then get sperm and egg donations from professional tennis players.
In general, reading a phenotypic effect off an actual phenotype is always going to be easier and more accurate than trying to do it from genes.
I'm sure some people will say "but I want my children to pass on my genes"; and fine, those people who want that can do it. But the government should also facilitate prospective parents who want make use of the best genomes out there.
> It’s probably even obvious for governments to pay for all citizens to get these, given how much they’d save on health care costs.
IMO it's a no-brainer for governments to do this, and offer genetic counselling including sperm/egg donation, for those people who want that. Governments would not only save on health care costs, they'd also benefit from a more intelligent and conscientious population with people less likely to commit crimes.
Yes, this technology is most helpful for people who want to have biological children with their life partner, and/or can't get the greatest genius in the world to donate a gamete to them. But I think that's 99% of people.
> and/or can't get the greatest genius in the world to donate a gamete to them
I envisage government-run gamete donor banks where donors are selected for things like: high IQ, high conscientiousness, high achievement, in good general health, not having committed serious crimes, being physically attractive, etc.
Because it's a lot easier to get sperm out of a man than eggs out of a woman, that would probably be what was mostly done.
> But I think that's 99% of people.
I agree with you that it's most people, but I would guess it's more like 70-80% than 99%. My reasons for this are:
(1) If I ever had kids I would want to go this route, as I have been held back all my life by poor mental health, and I wouldn't want the same to happen to my children.
(2) Most parents want their kids to have happy successful lives and if this helps, many would go for it.
(3) Never in my life has any parent bragged to me about how ugly or stupid their child is, but many times have they done the opposite.
(4) As well as optimising for being clever, this process can also optimise for children who're good looking or have pleasant personalities. I feel this last one might be important to many parents considering they will be in close proximity to their kid for the next 18 or so years.
(5) Adoption is a thing in many cultures. In some cultures you could even adopt people after you were dead!
Obviously if 20-30% of kids were created by this route and so had high genetic quality, that would improve things in any country.
This is fascinating and IN PRINCIPLE sounds like great things happening fast, but I have an incredibly strong feeling that the whole text seems to devote much more attention to comparatively lesser concerns ("we don't have this as good for non white people yet", "what if I still get x cancer even if I had a 5 vs 10% chance) and only fairly briefly cover the huge, glaring, elephant in the room: implementing this on anything even remotely societal scale (even within the top 25%) would completely, possibly irrevocably change how one of the key human life-form / species - relevant functions is carried out. That this is "IVF only" option seems to be huge. Unless you have a massive genetic burden (and even then testing and aborting if detected seems better), this seems a real hurdle?
On the other hand, expanding to individual pregnancies would massively up abortion rate in women "trying for optimal" or at least "trying for good enough", which would create its own issues (and a big outcry in groups already opposed to abortion or on the fence regarding it).
> Selecting the best of five or ten embryos is not a very effective way to get the genes you want
I appreciate that Scott mentions this primarily because it is quite common in popular culture to go around claiming that PGS embryo screening will give people “superbabies” and this claim is obviously a marketing gimmick.
The within-family validation that Herasight has done is great and shows that their screening can help parents increase the chance that their kids will lead healthier lives. I think that’s great! But we shouldn’t oversell the technology because that just breeds mistrust and resentment.
If you've already decided to do IVF, I think the rest is footnotes. One has to manually select somehow in any case. (I disregard scare talk using thought-stop words like "eugenics." If someone has a particular objection, I'm happy to consider it.)
For my part, I strongly favor making babies the natural way.
Arguments about the genetic components of heritability (or lack thereof) often seem sort of unresolvable with existing techniques since there is no independent variable to manipulate when you're dealing with humans. e.g. If you hypothesize an allele causes black fur on an animal, you can manipulate it (by breeding or CRISPRing) & see if you're right : a proper attempt at falsification. In humans, there's no manipulation so you have to all kinds of correlative stuff, which kind of can't make causal inference. (Theoretical considerations aside, not being able to directly test stuff seems to make progress way, way slower.)
I have my reservations about current PRS, but I am hopeful the application of these sorts of technologies could produce exactly the sort of data that could resolve some of these arguments & increase predictive power: by actually *testing* trait predictions, not on a held-out set (which could have the same confounders as the training data) but with a real experiment (selecting or CRISPRing an embryo). The resulting data could have long-term benefits. Even in the unrealistic extreme case where everything is population stratification and selection on PRS didn't work at all, that would suck for the people who paid for the service but would be a very valuable result scientifically & medically.
(Of course this assumes that there is enough high quality follow-up work measuring the traits people selected in this way, and enough transparency on the part of the firms offering this service about their predictive algorithms. Also practical considerations mean that this won't exactly be a controlled trial & there will probably be dramatic selection effects based on the sort of people who pay for it. But still better than just running regressions on biobank data.)
Modifying genes was impossible for most time and still very restricted, but many people tried creating environments to create geniuses and barely found anything; most educational interventions produce IQ gains in children that fade in adulthood.
"Confounders" here are not environment variables (like nitrogen in soil for plants); if anti-hereditarians determine that a trait is "confounded" they don't learn how to modify environment to change a trait, but they actually find that people like to self-segregate and the rich will pay money so their child wouldn't attend same school as children of poor. In some sense, when people can migrate where they want, unlike North Korea, is very good!
Some of "confounding" might be womb environment rather than "school quality", did they test that?
But hey, we could just clone a few geniuses instead of trying to analyze what of 6 billion bp makes them, and who is against that the most?
All the companies you've mentioned are going off of public bio-bank data from what I can tell. Is there a reason they haven't tried to prove out their technology by raising some genetically predicted super mice, and provide empirical evidence (through enough overly healthy and smart mice) that this technology works?
I absolutely approve of reducing the risk of disease. It also seems profoundly stupid to spend lots of money on determining hair color when hair dye is a thing. Most people don't even know Christina Hendricks isn't a natural redhead.
>Mark my words, one day there will be entire subfields of therapy dedicated to these issues.
I feel like this is the core of the issue, really. When people say "I am more than my genes," they're probably not saying "I deny that statistical correlations between genes and outcomes exist," they're saying "I do not want my kids to grow up being told they're Genetically Superior/Inferior." Which is a correct thing to say!
> Somebody in Macaronesia in 3525 AD will avoid getting breast cancer because of you (if there is still cancer; if there are still breasts).
Big breasts are a burden (literally), genetically engineer flat-chested women and men who are not attracted to breasts.
I wanted to say that you can go even further, do parallel gender selection away from the accidents of evolutionary attraction toward more functional bodies that we would consider ugly now, but... in reality the fertility crisis will likely result in artificial wombs/vat babies/cloning much sooner than 3525, decoupling procreation from attraction and eventually making sexual attraction a weird rudiment that the enlightened aces (and eventually everyone) would sneer at.
The Selection objection I find strange. If you had been selected for pretty eyes regardless of cancer risk, you still won the billion:one lottery! The remarkable odds against any of us existing at all makes this a silly objection. If they had made a "better" choice regarding cancer, you wouldn't exist!
Doesn't this prove too much? If your parents beat you as a kid, you can't complain, at least you exist at all!
I'd complain, but they'd have to beat me pretty damn hard to make me wish I had never existed
Yes, and I think people will also complain about their parents selecting them for eyes rather than cancer risk (while still not wishing they didn't exist).
I agree there's a philosophical sense where if their parents had done anything differently, they wouldn't exist, but I think it's a very weak sense. See https://www.astralcodexten.com/p/who-does-polygenic-selection-help . Or compare to the example of someone whose grandparents (before they were born) left their parents money for their college education, but their parents gambled it away and lost everything. Now, there is some sense in which if the parents hadn't gone to the casino, it would have jostled the sperm in their testicles slightly differently, and a different child would have eventually been born instead of them. But I think it would still be reasonable to resent your parents for this, even though in some weak sense it caused you to exist.
Sure, but I think there's a tangible difference in that in the selection case the parents made a conscious choice of me vs another, and presumably I know that. In the gambled away example it *feels* more random, disconnecting the choice from the existence in one's mind. Even though it could be argued that the end result is the same, complaining in the first case seems more unreasonable and less defendable.
Don't IVF babies have modestly higher baseline rates of birth defects and other health problems?
If you already need to do IVF, sure, adding genetic selection largely makes sense. But if you don't need IVF, it would be important to consider not just the health downsides for the mother of doing IVF but also the health tradeoffs for the baby, right?
> Although GP would continue to only test for health outcomes, you could forward the raw data from GP to Nucleus, and Nucleus would predict extra traits, including height, BMI, eye color, hair color, ADHD, IQ, and even handedness.
It does seem important to ensure your kids have an even number of hands.
Nah, don't you want them to grow up to be Alexander the Great? https://people.brandeis.edu/~mob/puns.htm
"Suppose you invest the $3,250 in an index fund that makes 7% over inflation, then give it to your future child when they turn 45 (average age of type 2 diabetes diagnoses). Now it’s worth $75,000. Is this the “true” cost of the intervention?". no, $75k in 46 years is less than the true cost, which is $3250 today. it is less because $3250 can buy $75k in 46 years or it can buy other things which are likely more valuable. we know they're likely to be more valuable because people use their marginal $3250s to buy them and not $75k in 46 years
of course sometimes people do trade $3250 today for $75k in 46 years, usually when saving for retirement. so sometimes the two are the same. but your post seemed to imply that this was not the next-best use of the money
So much rigor here and then "AGI is going to cure all disease and I know that with overwhelming confidence despite the entire history of human folly regarding such predictions"
Eventually we are going to automate scientific research through artificial intelligence. Not clear when. Soon after that happens, I think it’s reasonable to assume that most diseases will be cured, just as we might assume that most diseases will be cured with enough scientific study
Well, if we're all turned into paperclips, that sure will cure all disease!
I'm less sanguine about what will happen to medicine, but I _do_ expect that, barring an antarctic level AI winter, AI will radically change the whole game.
I think the main worry I have about this technology, (and about selective reproduction programs like ones attempted by Musk or things like The Repository for Germinal Choice) is some variation of "what gets measured gets managed".
That is, we are going to use this technology to optimize things that we can measure. These things that we can measure, like IQ, are imperfect heuristics for things that we actually care about (intelligence or mental capacity or academic potential I guess?). And even those are simpler than the things that we actually-actually care about which are at the level of our society and culture - overall standard of living, kindness, freedom, etc...
So if we gain the power to optimize for things, and we start optimizing for the heuristics that we have available to us, like IQ, we may face an alignment problem. Maybe the things that we can measure right now (like our current conception of IQ), or the things that are in vogue right now (height, particular shades of eye color), are misaligned with the things that we might ultimately care about.
So maybe IQ is negatively correlated with some other important set of traits that are murky and difficult to define and we can't measure very well (say, creativity or being an empathetic human being, or "research taste" or etc...). Just because these things are hard to define and difficult to measure doesn't mean they're any less important.
Furthermore, traits that may be beneficial for individuals may not be beneficial for society. So maybe an individual does better with higher IQ within our current society, but a society where a significant number of individual have a higher IQ undergoes a phase transition which is worse in some way. Maybe it's an overall higher level of cynicism, or a breakdown in the useful fictions that create some form of social cohesion and makes economic cooperation possible. Maybe it's a loss of overall diversity of the population that makes the overall population more fragile.
I guess my overall feeling on this is that we are so primitive in our understanding about these things, and our power to make changes to our genome may advance much faster than our understanding about the societal consequences of those changes. And part of me thinks "but I suppose that's just science in general", and part of me thinks "no, that's just technological determinism!".
This post made me think a lot about the silo series by Hugh Howey. Is humankind doomed because the advances in our power to change the world and ourselves outpaces the wisdom required to wield it without destroying ourselves?
This sounds like the antagonistic pleiotropy suggestion, but see my response above. In practice most good things are correlated with one another.
I am also very worried about our power to change the world outpacing our wisdom, which is why I want to intelligence-enhance children as soon as possible! I agree intelligence isn't exactly the same as wisdom, but it's probably pretty correlated. If it isn't, feel free to submit an ACX Grants proposal to develop a genetic predictor for wisdom!
the Doom of the Noldor - they were intelligent but they were not wise.
It seems to me the most important ethical consideration is the one you treat the most glibly, that parents will be making a specific choice about what their kids should be like, that their kids DNA will effectively become a message vector in our society. This isn't just a matter for therapy, this means inviting all the most irrational cultural pathologies into our genetic code, and basically making future generations subject to a certain biological totalitarianism by their parents. It seems to me an exceptional important ethical consideration to standardize any selection process so that it ceases to be a potential message vector.
Stuff like this: "How much would you pay to give your children an extra 1-4 years of healthy life?" and framing things in terms of how much better you're making your child's life is misleading. You're not making anyone live longer, you're bringing into being someone who lives a longer life over someone who lives a shorter life. I might pay a lot to make my actual dog live four years longer, but that reason doesn't automatically translate to a reason of the same strength to buy or breed a chihuaha rather than a great dane, even though in the same sense I am "giving my dog a longer life".
If you're a repugnant-conclusion bullet-biting totalist utilitarian then this difference doesn't matter but for most other plausible moral perspectives (especially those on which your reasons regarding your children are stronger than those regarding strangers) it does. That's not to say you don't have any reason to bring about a longer-lived child, but it's very plausibly not a reason of nearly the same strength as your reason to cure your child of a disease which would end their life early.
I've had 8 children, three miscarried.
No....just no. This is not a good path.
Why not?
It isn't going to end with just wanting to make sure your kid doesn't have diabetes. Genetic disease is the bogus gateway drug. This is a frankensteinian future with a million unintended consequences. No one is going to think about, or care about those unintended consequences until it is too late.
This isn't a spinning jenny moment. This is playing God. Historically, philosophically, morally - that never ends well.
What are the historical precedents?
Playing God has gone pretty well for us so far.
> unintended consequences
The problem with the "unintended consequences" argument is it's far too broad. When people invented agriculture, writing, metalworking, or movable type, all those had unintended consequences.
If we ban everything that might have unintended consequences, we ban all progress.
The argument about Dude, Genetic determinism has a fine side. Basically, not that this is a reason that we should not do this. But basically, if you really think about the stength of these predictions, they are not very strong. After use for selection, there are very few situations where the information about future traits would be useful compared to not knowing (mostly test order for diagnosis). If these techniques get broader, they are still weak evidence about what someone will become. These companies do not publish the remaining variance from the expected values published (especially for IQ and healthspans), and they should. If you really internalize those, doing that behavior is obviously dumb. The variance remaining (either from what a 3 percent chance means, or the remaining lifespan variance, or the IQ variance not explained), is often high enough that bets against the low probability even need to make sense, and having strong expectations about how children will end up are in general, low to negative EV even if your are expectations are right, and even decisions based on that expectation with bad payoffs in the low probabilty case should be avoided, for obvious reasons. That is, the fact that the obvious case is so obvious is tripping up the complainers.
> It’s probably even obvious for governments to pay for all citizens to get these, given how much they’d save on health care costs.
Not a criticism of the idea, but I was under the impression that healthier citizens cost more, because they live longer - is this not true?
I don't expect the average ACX reader to be persuaded by my thoughts on the immorality of IVF in general, let alone embryo selection. Those arguments are out there if you want to look for them.
But perhaps you feel a bit of disgust at the thought of people picking out children like they'd pick out puppies from a breeder. If you can have a labradoodle, why not a blue-eyed blond baby with ten more IQ points and a 32% lower risk of cancer?
I also think this post glosses over the cost/inequality aspect of it. There are billions of people on earth who are never going to be able to afford this. Their descendants in the next few generations won't be able to, either (barring some weird post-scarcity techno-utopia AI-world, which I'm not counting on). What does the world look like in 2125 after a few generations of elites selecting embryos (with the technology presumably improving along the way) while 4 billion or so Africans can't or won't? As embryo selection gets better, people will be picking for everything you can imagine: looks, athletic ability, artistic talent, etc. You'd be on the way to having two totally different human species. Doesn't this sound like a dystopian sci-fi novel to y'all?
“Doesn't this sound like a dystopian sci-fi novel to y'all?”
Yes, but I think you’re guilty of status quo bias here. The world looks like a dystopian novel to me now. It’s not clear that your image is any worse.
What's the state of polygenetic prediction for Big 5 personality traits? Supposing you think AI will make IQ and "boring" illnesses like cancer, T2D, and heart disease irrelevant. You might reasonably conclude that you should focus your efforts on having kids who are kind, thoughtful, creative, and emotionally stable (or at least not unhinged assholes). Would love to know how many SDs you could move on various Big 5 traits as a function of number of embryos chosen from.
Has anyone written at any length about the potential for personality selection?
Seems like there are 4 big categories of traits ordered by controversy:
1. Clearly negative health based traits (genetic diseases, propensity for high cholesterol, etc)
2. Clearly positive traits, not clearly health-related (IQ, balance, etc)
3. Preference based aesthetic traits (eye color, height)
4. Personality traits (neuroticism, openness to experience, etc)
Of course, many traits will blend between multiple categories. IQ is correlated with mortality, tons of psychiatric conditions are associated with big 5, etc. By and large, though, I’d guess that there are more who would object to eye color than to IQ and more to IQ than to cancer risk.
Personality traits seem to be objectionable on entirely new criteria than the rest. It’s one thing to charitably select for lower incidence of cancer in your kids and a whole other thing to select for kids that are likely to have a charitable attitude or a certain belief structure. This seems inevitable, though. If we can select against depression we will eventually be able to select for gregariousness, for ambition, for any number of other traits. There’s the question of if this is ethical to do at all and then the secondary question of, if it is, is it ethical to select for a personality for your kids that is, for example, less likely to be happy but more likely to be successful?
"Most genetic data available for research/commercial use comes from the UK, US, and Europe - areas which are mostly white." All three had been working on this problem by making their populations less white, although the US has shifted on that in the past six months.
>the US has shifted on that in the past six months
Has it? I've seen much wailing and gnashing of teeth, but few stats.
I think people in the rationalism community should definitely not select for higher IQ fetuses, because they likely already carry risk factors for autism. In another population it'd be fine; in this one it's riskier.
Isn’t autism negatively correlated with IQ
Some guy with a similarly sounding blog wrote a whole post about this! But to sum up, *actual* autism is negatively correlated, but SNPs for IQ are positively correlated.
So if you have two parents that are on the BAP - broader autism phenotype - they already have a much higher risk for having an autistic child. If you do embryo selection on IQ you're also much more likely to select the autistic child because there's overlap between IQ SNPs and autism SNPs; some of these are the same SNPs.
https://slatestarcodex.com/2019/11/13/autism-and-intelligence-much-more-than-you-wanted-to-know/
> Ethical Objection: Personhood
I'm glad you have addressed this concern, but you have not really internalized it properly.
> Ask anybody dealing with regular insulin injections (let alone limb amputations) whether it would be worth $27,000 to wave a magic wand and not have Type 2 diabetes! It’s not a hard question!
The proper question is, whether it would be worth $27,000 to wave a magic wand and be replaced by your sibling who doesn't have Type 2 diabetes. I agree this is not a hard question.
Also, while I follow and agree with all your points intellectually, I just can't share your seeming enthusiasm. "romantic and aesthetic"? Yeah, nothing says romantic like selecting an embryo based on the polygenic score calculations.
Thank you! This technology is not "curing" anything, it's just throwing people in the garbage who have undesirable genetics.
Only if you consider a zygote to be a person, which I don't.
I read recently about multi-PGS where they take PGS for seemingly unrelated traits and make composite predictor (which might be non-linear) with better score. But I found very little about it.
Eugenics carries a deeply negative legacy, for reasons any student of history will understand. That said, it’s worth distinguishing modern genetic screening from its past abuses.
Preventing heritable diseases is widely considered a positive goal. Reducing the burden of serious medical conditions can improve individual lives and contribute to public health. Using technology to improve the overall condition of the human species is a valid long-term goal.
Moreover, I submit that the defining feature of our species is intelligence. In today’s society, lower cognitive functioning can be a barrier to employment and independence, while higher cognitive ability is linked to greater productivity. If reproductive technology allows parents to select for intelligence, does it really matter if they, at the same time, choose their children's eye color?
You go on to mention race. While obvious characteristics like skin color are irrelevant, science has repeatedly observed group-level differences in cognitive outcomes across populations. These differences correlate strongly with the prosperity of the respective societies. At the same time, fertility tends to decline as societies become more prosperous - something noticed even in antiquity. This is, effectively, a brake on human progress. Whether this can, in the long-term, be addressed by genetic screening?
If a company is selling a product that costs tens of thousands, but there's not really any way to verify the product worked in practice without doing a study across large numbers of the product's customers over 80 years... then I'd have a very high prior that it doesn't work as advertised. The potential for fraud or mistakes is too high.
If it doesn't work for whatever reason, by the time this can be proven to a legal standard (particularly the parts relating to the longevity of the selected children, and their disease frequency in later life), the customers and the founders will be dead, and the company may no longer exist for your grandchildren to sue.
Or, to put it another way, if I wanted an unethical way to get rich, I could sell expensive products whose efficacy couldn't be disproven until after my death.
I think this is a concern, but we already have a culturally normal practice for something similar: the construction business. We regularly contract companies to build houses and bridges and so on intended to last for 100+ years. I think many of the same ways we assess the reliability of construction can be applied to embryo selection, albeit with a lot more uncertainty at this stage.
Because the selection involves a variety of traits that become evident in different years after the purchase, we can assess how reliable the company is to some extent long before we get to the 80 years out threshold: firstly you can genotype the baby to see if the claimed genes are actually what's in the kid. Secondly, there are traits like eye color that show up within a few years, traits like height that show up within a few decades etc.
In that time, we also get more data on the genetics of cancer and other long-term risks in the population in general, and we can compare the results of the embryo selected vs the baseline.
I will never get tired of linking https://www.astralcodexten.com/p/the-buying-things-from-a-store-faq
That really doesn't seem to apply to the case where a store sells you an item with a promise that it will or won't do a thing 40 years later. Or that it's 10% less likely to do a thing 40 years later. How much class action do you need for a bunch of customers to get together 40 years later and say "hey, this thing you sold me wasn't actually any more likely to do the thing than the version we could have got for free!"
If I go into a store to buy a stand mixer, and they promise to sell me a stand mixer, and then wheel out a concrete mixer - well yeah technically it's a mixer but come off it, mate.
At the moment, I think the technology is at the level of "we can tell you if the embryo has Down's syndrome or CF" and they're promising (but unlikely to deliver on) "if millions of people select on these criteria we supply, then in twenty years time out of the hundreds of thousands of babies born, there may well be an overall 1 point increase in population IQ" but leaving the clients to parse that as "my baby might be 6 points smarter!"
You and your wife are reasonably smart people, she's going to take care of herself during pregnancy, and you as parents are not going to fold, spindle or mutilate any kids you have. So there's a good chance any future babies will turn out okay. Will that be due to the genetic selection? God alone knows! But I'm fairly sure the company does not.
They're selling hope in a bucket and badmouthing their competitors in the effort to grab market share, market advantage, and those sweet, sweet, aspirant dollars before they go as bust as 70s cryonics companies.
I'm guessing (I haven't read any PGS company's offers) that they are promising to do only what they can do - to offer customers the choice of embryos to implant based on the polygenic risk scores they have scored the embryos for. The actual outcome will be indistinguishable from chance on an individual basis, as you say. It's only in retrospect, across hundreds of thousands of people, that any impact will be noticed.
I was trying to think of an analogy, and I think it's like an impressionist painting. Up close, every splash of colour is just a splash of colour. You have to take a good step back to appreciate the image. Up close, each individual is just that, and, except for some Mendelian single-variant conditions, it's impossible to pin down a talent, ability, condition or illness to any one cause. It's only from a distance of hundreds of thousands of people that you can see a pattern.
EDIT: And for that reason, I rather think that what these companies are really selling is a chance for parents to believe they have chosen the best they can for their child.
Here are some potential objections:
1. This allow parents to select sex of a child as well, which could move us away from natural balance in which for every 100 girls 105 boys are born.
2. Selecting for obvious traits like risk for certain diseases, IQ, etc, might also unintentionally select for some and against some other traits that weren't even considered. This too might skew the natural frequency of some traits... making some more prevalent than is optimal, and some less prevalent than is optimal.
3. This opens the can of worms which will allow kids to blame their parents for pretty much any negative outcome in their lives - whatever they did: if they used this technology and kid turns out to be unhappy or to have some disease, they can blame parents for wrong choice; if they didn't use this technology and conceived naturally, kids can blame parents for NOT using this.
4. In general kids lose some of their dignity points by not just coming into their parent's life, but becoming in some ways their explicitly engineered product. Imagine a situation in which a kid is selected for high IQ, and then is pushed by ambitious parents to become some wunderkind or to excel academically, while the kid itself turns out to be completely uninterested in such a thing and would prefer to be a soccer player or a plumber. Imagine the kind of arguments that could ensue. "Did I in vain select you to be so smart? Now you want to waste all that potential?!
- F**k you and your embryo selection! I'm my own person!"
5. If eventually personality traits become a part of the equation, so we can select for them as well, there is a risk that humanity could become much more homogeneous, we could become a society in which popular traits are very frequent, even if "popular" doesn't necessarily mean optimal, especially not on a level of whole population. (For example, for a single person, it might be optimal to have some "good" personality, but for the entire population it might be more optimal if not all people have such "good" personality, and we also have a fair share of quirky people, neurotics, introverts, etc... For example high neuroticism might be required for taking seriously X-risks)
6. Achieving genetic perfection could put more pressure on people to perform, because now they have no excuse. If you fail in life, it's only you to blame.
7. In general it interferes with natural selection and introduces elements of artificial selection. A thing that we do to dogs, for example. Many of the breeds created by this process actually have worse health and fitness than mongrels.
8. There is a open question about how many embryos to select from. It could be 5, 10, 50, 100... or more? Each choice could be criticized.
1 is already an issue. Parents using IVF in the US can select by gender, and apparently it's socially acceptable in educated circles to do so, as long as you are selecting for a girl. In the EU, it's illegal to select by gender (though I'm not sure of how this can be tightly enforced).
Gender-selective abortion bans came about in the EU because, surprise surprise, turns out parents (particularly those of certain ethnicities whose culture place much more value on sons) wanted boys and were aborting girls, even if that was already illegal since they could find a compliant doctor if necessary, although it is a complex question - some studies find a skewed ratio in live births, others don't, and there is a natural ration of 105 male to 100 female live births anyhow.
https://www.bbc.co.uk/ethics/abortion/medical/selective_1.shtml
But yes - see the Boaz article I linked to.
Not sure if you, or the companies, have answered this elsewhere, but I can think of two more issues. You've discussed the relative risk of developing the condition, but not the absolute risk, nor the severity of the condition. Some of the conditions tested are exceeding rare, and so even a 90% reduction might not mean much. And to pick breast cancer survival for example, while apparently the risk is rather high for the average woman (about 14%), the outcomes if not diagnosed late are actually pretty good. And you don't know the severity of those who will get the disorder. Maybe the ~7% who do get breast cancer will have a kind that is more aggressive or treatment resistant, which means the the relative risk is misleading and this process is actually not worth it.
The other issue is that in removing these genetics from the gene pool, you're removing beneficial traits that are harder to identify. Lets say that people with schizophrenia genetics are more... adept at being altruistic or something like that. You could develop a society that's perfectly healthy but but there's a statistical predisposition to selfishness that changes the whole culture.
About action when the future is unknown: 500 years ago, when Martin Luther was asked what he would do if the world were to end tomorrow, he answered, “I would plant an apple tree today." That story got rephrased into "Even if I knew that tomorrow the world would go to pieces, I would still plant my apple tree."
3 fun facts: 1. there is an eco-bestseller in Germany titled: "Let us plant an apple-tree! It is about time." 2. The original quote might be Islamic: “If the Final Day comes upon one of you while he's in the middle of planting a sapling, let him finish planting it.” -Prophet Muhammad ﷺ
3. Some mural-artist in Washington mixed up Martin Luther and MLK https://reformedjournal.com/2013/01/28/martins-and-luthers-and-misquotes/
"suppose that I had a natural, unselected child, and that child became high school valedictorian and got into Harvard. Would my first reaction be “Oh no! This slightly raises her risk of anorexia!”? If not, why should this be our reaction to artificially increasing IQ?"
With natural, unselected children, you don't probably think about their traits (might be using "trait" wrong here) in terms of tradeoffs, because you never opted for anything in the first place. Like I don't actively think about stuff like "Gawd, I wish I was born with worse eyesight so I had more voluminous hair" or such correlations there might be between different traits, because they're not commonsensical and, more importantly, you don't opt for them.
There's no sense of responsibility or regret over traits we didn’t select, while selecting involves intentionality and responsibility.
I think, though, that the question is fair to present just as "If I could be smarter but have anorexia, would I still want to be smarter?" assuming the elevated risk of anorexia matches the risk of being smarter.
I would probably select for non-anorexia, but my guess is that this community would tend to prefer the IQ? Does this point out a flaw in the selection process? I freely admit I would probably be bad at making these kinds of decisions for my baby.
Assuming the selection only yields potential positive effects and no potential negative effects, it's of course a no-brainer that this would be good for everyone with any selection. Assuming there are tradeoffs, it raises legitimate ethical concerns about whose values guide selection and whether those values are always in the best interest of the child.
Also, let's assume the tradeoffs aren't that clear. Having lesser risk to heart disease has a slight potential risk of heightened aggressiveness? I assume this community would baulk at aggressiveness as an undesirable trait. But assume other communities you can probably name, where aggressiveness is desirable, where people think that it's a dog-eat-dog world and being big and intimidating and a bit of a rabid dog gets you places and means you can protect your family better and so on.
I guess I am essentially making the slippery slope argument and highlighting the vast differences between cultural preferences for desirable and undesirable traits.
I think if you asked parents of anorexics if they would prefer their child were a little dumber but healthy, they'd pick that.
"Smarter, but a risk of anorexia" sounds great when it's a theoretical discussion. Who wouldn't want their child to have the advantage of being that bit smarter than their peers? And my kid will be too smart to fall into the trap of self-starvation! When it's your teenage child starving herself down to skin and bones, you may wish you hadn't ticked that box in the clinic, because intelligence has little sway over compulsion:
https://www2.hse.ie/conditions/anorexia-nervosa/
"Anorexia can lead to severe health problems associated with malnutrition (not eating enough healthy food). But these will usually start to improve once your eating habits return to normal.
Possible complications include:
- problems with muscles and bones - feeling tired and weak, osteoporosis or physical development
- fertility problems
- loss of sex drive
- problems with the heart and blood vessels
- problems with the brain and nerves - such as fits, difficulties with concentration and memory
- kidney or bowel problems
- having a weakened immune system or anaemia
Anorexia can also put your life at risk. It's one of the leading causes of deaths related to mental health problems. Deaths from anorexia may be due to physical complications or suicide."
"It’s hard not to love this technology."
Oh no, it's not hard at all.
I'm extremely sceptical, not to mention cynical, about promises of earth-shattering benefit to all humankind. So this makes me murmur "Oh, really?" very strongly.
First and foremost, this whole industry is a business. Yeah, yeah, healthy babies is the selling point. That's not what the share price is about, or even the business (else a company wouldn't be promising 'pick hair colour and eye colour').
So one company takes a swingeing criticism brickbat to its competitors? Well, colour me surprised that a business is telling potential customers "ignore those fake lying snake oil purveyors, spend that 30 grand on *us*, the only true genuine real deal snake oil!"
Second, as pointed out, we'll only know how all this works once once many genetically selected babies are born and are observed as they grow up. Did you get that 6 point IQ increase after all, or would your kid have been smart anyways? Oh, Junior *is* smarter, but they're also troubled?
What we're forgetting is that we're playing dice with real beings, the babies who will be born and grow up after this process. I'm happy if they aren't going to get cancer or whatever, but I remain dubious.
(As an aside Type II diabetes doesn't necessarily need insulin, other medications and lifestyle changes are tried first and it's only if they're not working that you're put on insulin).
Thirdly, I think the potential for more family conflict this opens up is horrendous. It's bad enough when parents compare you to other family members/classmates/the neighbour's kid, but now you'll be competing in an unwinnable contest against an unbeatable rival: the ghost of the ideal you that your parents planned out when they selected for and against traits.
You're not *quite* as smart as you should have been (that 6 point IQ gain was in a range, in the small print, and you only got the 3 point gain). Not quite as tall. Not quite as attractive (your hair isn't the exact shade of blond/blonde, your eyes not that bright enough blue). Not quite as athletic, or academic, or musically/artistically talented as the brochure promised.
You can pick up that your parents are quietly disappointed that you're not as good as you could have been, even if they hide it (and as for the parents who don't bother hiding it? yeah that's gonna be a fun childhood).
And fourthly, I damn well could see people selecting for anorexia. Not the full "starve yourself to death" kind, no, but a stronger inclination towards it? A kid who is constantly dieting, exercising, calorie-counting, and reliably only ever eats exactly six peas, two carrot batons, and a square of tofu for every meal - well, they're going to hit that BMI target as promised by the company, aren't they? No dangers of obesity here, no disgusting bucket of lard boy or girl to shame you in public for raising a fat kid and being unworthy parents! Couple that with the tallness genes, and if you get a tall, skinny, weight-obsessed kid, there's a good chance of a modelling career which will more than repay you for that extra expense in gene-selecting! And if a little IQ bump comes along too, that's just gravy.
Good luck to those that choose this road of embryonic selection, I know you all mean well (in this generation and context, at least). All ye holy men and women, pray for us.
https://www.youtube.com/watch?v=6lFkebgycxo&list=RD6lFkebgycxo&start_radio=1
I share your concerns and I'll add another -- the dynamic you describe could be reversed, and parents may end up blaming themselves if their kids don't realize the advertised benefits. Channeling that disappointment inward, on the decision maker, is less bad than channeling it outward, but it's still bad and can have spillover effects on the children.
Generally a good comment. Yes, I agree with the hazards you cite. One quibble:
>What we're forgetting is that we're playing dice with real beings, the babies who will be born and grow up after this process.
Embryo selection is _reducing_ the role of chance in reproducing. The dice are already there, in the natural process. Selection makes it _less_ of a dice roll, not more.
But we are still rolling the dice, only this time we've loaded the dice without being sure which side is going to come up. We're hoping for sixes, we might in fact be setting ourselves up for snake eyes.
Many Thanks!
>only this time we've loaded the dice
Exactly agreed. On each individual roll, embryo selection is reducing the fraction of the variation due to chance.
>without being sure which side is going to come up. We're hoping for sixes, we might in fact be setting ourselves up for snake eyes.
Yes, you are right that we are introducing a systemic risk that the selection may have unknown deleterious effects. I think _most_ of that is covered by Scott's section on correlations. E.g. generally speaking, higher IQ correlates with a bunch of _better_ outcomes in general health.
Admittedly, we _never_ know the long term effects of any innovation when we introduce it. If it had turned out exposure to the 120 Hz flicker in fluorescent lights (unprecedented in our evolution) mades people go stark staring bonkers after exactly 30 years of exposure, we would have turned every office worker into a maniac a generation after those tubes started flickering. If we wish to innovate _at all_, we have to accept that kind of risk. Mostly, we monitor short term side effects, and, mostly, that works well enough.
A moral premise implicit in trait-based embryo selection is that a reduction of difficulty leads to better lives: that traits associated with struggle (lower predicted IQ, higher BMI risk, neurodivergence, etc.) are inherently undesirable, and that fewer of them means progress.
But not all difficulties are defects. Some challenges are formative. Some constraints shape the trajectory of a person’s development in ways that aren’t easily captured by metrics of “health” or “success.” Many people living with lifelong limitations -- not just physical, but cognitive, social, or emotional -- report that their identity, moral insight, and empathy were forged through precisely the conditions others might now screen out.
I don’t like suffering, and I don’t romanticize it. But I reject the idea that a desirable life is necessarily a frictionless one. Yet, that’s the vision this technology seems to be nudging us toward, without much pause.
One thing I find confusing about Scott's stance on eugenics is that he takes Chesterton fences and the idea of cultural evolution seriously in other contexts. His stance on central planning, at least for the pre-1970s era, seems to be that the theoretical case was strong but conservatives were right to resist for illegible reasons instilled by what made societies successful in the past.
Normal selective breeding has been known about since the Neolithic and could change a society's population genetics much faster than embryo selection, and having better genes would have been really advantageous for any society. But in 10,000 years no society has had a serious eugenics program. Isn't the fact we aren't already the descendants of Bronze age eugenicists who took over the world evidence against eugenics? If you believe in cultural evolution, it must really hate eugenics for some unknown reason.
It doesn't seem consistent with his caution for other kinds of social progress.
Eugenics has been suggested since our earliest philosophical writings (Plato and Socrates both recommend it in the ideal society). Beyond philosophical writings, the knowledge or ability to influence population-wide genes simply didn't exist. You can't control who has children, especially with more primitive social structures, and people were mostly having as many as they could to start with.
The closest example in ancient societies is probably the Spartans, which effectively practiced eugenics, but they also had a lot of other cultural baggage which made them more stable, and more effective in war, but ultimately lead them to have population issues and eventual collapse.
I don't think it's right at all to claim that normal selective breeding could change a populations genetics faster than embryo selection. When I've seen people model it, it requires massive changes in the demographic of who's having children and how many, only for tangible benefits to be realized after 4-5 generations at immense cost. I don't think there's a society on earth that has enough excess power and foresight to make it happen.
>Eugenics has been suggested since our earliest philosophical writings
Yeah, I think once you know about selectively breeding animals it quite an obvious idea, which makes it more surprising that it's not already the norm..
>I don't think there's a society on earth that has enough excess power and foresight to make it happen.
Possibly, but there have just been so many societies since 8k BC. Autocratic ones, God-king cults, ones with lots of different sexual norms, slave societies etc. None of them could get a eugenics program going?
>only for tangible benefits to be realized after 4-5 generations at immense cost.
I think this applies to all arguments from cultural evolution. There are also strong immediate pressures against eg. monogamy, private property, state subjugation etc. which cultural evolution proponents generally accept. And eugenics seems like it could potentially give a bigger competitive edge than those things.
I'm not personally that convinced by Chesterton fence/cultural evolution arguments, but if you do accept them it seems you should at least consider eugenics through that lens.
> Beyond philosophical writings, the knowledge or ability to influence population-wide genes simply didn't exist.
The concept of genes didn't exist until the mid 19th century, but people were deliberately breeding plants and animals for millenia before that.
> I don't think there's a society on earth that has enough excess power and foresight to make it happen
I think you're right -- the reason it hasn't happened with humans is a mixture of the long time necessary and lack of state capacity.
On the ethical objection of selection:
This is a chimerical issue, if only because of human selfishness. We appear to have no difficulty dealing with the same question in other respects. I know of several people whose parents squandered their inheritances, for example, and went on expensive trips or spent it all in other creative ways. They garner no sympathy from anyone if this is ever mentioned, even though it materially affects their lives and possibly the lives of their children. Why should we imagine that the new genetic selection will be treated any differently?
(I agree that helping people to be born healthy is important, and I don't understand the technology enough to see its potential faults, this is just a side note that I thought might be useful):
This seems to give parents an even greater amount of control over the bodies of their children than they currently have. This can do great good in the hands of parents who are altruistic toward their children, but having seen my share of dysfunctional families, and parents who are actively harming their children, I'm moved to wonder what this technology might do in the hands of the worst abusers. A controlling parent selecting for a disease that would make emancipation difficult — selecting for *lower* IQ, perhaps? Or a molester selecting for a child to look to their liking, the better to eat away their childhood?
Perhaps my own issues are showing, noticing this amount of control being given up and flagging that there's potential for even greater abuse in the hands of the minority of villains, but I wanted to bring up that this would also be one side of the question to keep in mind.
The eugenics performed by the blind idiot god already maims us sometimes, and so does the negligence and abuse in our childhood, but being able to just add some problem right into the DNA is an interesting new situation.
***
Now that I think about it, I wonder if opponents of homeschooling might be driven by something like this, knowing how horrible parents sometimes may be. I. e. someone like Scott and me might say, "the school is a horrible experience for many children, it's insane that our culture gives them no choice and doesn't listen to them when they say they're miserable and forces them to suffer for years on end", and at the same time, the opponent would be saying "being at home is a horrible experience for many children, it's insane to force them to be there with no alternative" etc. But I guess this is less applicable in a country with any kind of social services?
Go tell the dying person that the reason they can't be genetically enhanced is because you have "ethical concerns" about eugenics.
There is no inherent difference between enhancement and healthcare and that is fine.
People who have a problem with non-violent eugenics are motivated by envy and not some fear of the second coming of Hitler. It's really gross and disgusting to say to someone that they have to live with a disability because you have an ethical problem with enhancement.
Literally millions of people have died because of the ethical concerns of left-wing academics.
Everyone you know who has ever gotten cancer probably only got it because leftists have opposed genetic enhancement on the basis of arguing against eugenics.
I believe that a person should support a world where everyone is better off even if they themselves would not be better off in that world. If a person would oppose other people having greater health because of concerns over inequality or eugenics that person has a low and disgusting soul and is simply not worthy of any respect at all.
The only acceptable equality is the equality of spreading privilege around. To advocate for the deconstruction of other people's privilege is to advocate for a worse life for other people. That's disgusting logic, and the people who believe it are gross and disgusting people.
I can't respect them at all. I can't respect anyone who opposes making the world better on the basis of so-called fairness, equality, or equity.
People who oppose non-violent eugenics are the worst human beings on Earth. They are all literally mass murderers with the blood of millions on their hands. Everyone they know who ever got cancer or diabetes only died because of them.
“Everyone they know who ever got cancer or diabetes only died because of them.”
The alternative isn’t for those people to have not died. The alternative is for those people to have never existed in the first place.
The reason the dying person can't be genetically enhanced is because if they're an adult, it's too fucking late (to be coarse about it). The genes are already expressed and you have the deficient body, you can't just use an eraser and wipe out the little mistakes then. "Oh no, if only my parents had done embryo selection before I as born!" First, the tech didn't exist then and second, had they done so, the resulting baby would not have been the you that exists right now to complain about it.
Very different from picking an embryo on the grounds that "we think this one will have a lower risk". And of course the lower risk one if it survives implantation etc. will not be your dying person.
You're trying emotional manipulation here which is not going to work.
The time horizons involved in practically testing these claims are so long as to make verification impossible. Without verification, the field is going to be extremely vulnerable to motivated reasoning/wishful thinking/publication bias/etc. from the various companies offering this service.
The description of type 2 diabetes isn't true to my experience, but I have no idea whether I'm very unusual.
My blood sugar is well controlled with metformin, two pills per day. I test my blood glucose daily. I can eat fairly freely, though with some limits on carbs. It's been over five years, My eyes and feet haven't taken obvious damage.
I'd rather not have it, but the biggest downside is being dependent on the insurance/prescription system.
Unrelated, but anorexia does seem to be strongly correlated with OCD, bipolar, etc. Possibly the real solution is selection against mental problems.
On the other hand, selecting against anxiety might be out there in fantasy land at this point.
And select against anxiety: It turns out there are kids who don't eat-- sometimes so much anxiety about choking or other problems that they can't/won't swallow anything. https://archive.is/iZRcs
Discussion: https://www.metafilter.com/209770/The-Monster-at-the-Dinner-Table
I think there's another risk you're not considering -- systemic risks and pathologies that come from lack of diversity.
There's a common trope in disaster movies where a subset of the population rides out the apocalypse in a bunker so they can rebuild civilization. The lucky few are inevitably selected for being really smart or skilled in something that seems important for rebuilding civilization, at least to the people in power (who often just so happen to have many of those highly valued traits).
That's always seemed like a recipe for disaster to me. By decreasing diversity (genetic, behavioral, cultural, etc.) you increase the chance that some vulnerability that would normally only impact a subset of the population becomes a systemic, existential risk. For example, imagine a society that is entirely dependent on a single crop that experiences famine when that crop is blighted. Or a society where everyone gets the news from one or two organizations and are thus extremely susceptible to propaganda. Diversity is important for what Taleb calls anti-fragility, while homogeneity tends to increase fragility.
I also think you run the risk of creating dysfunctional societies. Imagine a world made from only the hardest working and most ambitious among us. They could put that drive to use creating all sorts of wonders! Maybe...but I suspect they'd first spend a lot of time fighting and maybe even killing one another because everyone would want to be at or near the top of the hierarchy.
It takes all kinds to make a healthy society. We need ruthlessly ambitious people who will stop at (almost) nothing to fulfill their dreams, because that's how seemingly impossible dreams become reality. But we also need empathetic people to prevent the ruthlessly ambitious from running amok. We need introspective, philosophical people to come up with institutions that can check our worst impulses. And we need a lot of people who are just chill and average and who aren't worried about "fully comprehend[ing] what we know of nature, and be[ing] able to fully appreciate the mystery of the rest", because those are the people who do all the boring things that make the world work. My grandparents were pretty simple people -- they weren't all that ambitious and they weren't particularly deep thinkers. But holy shit! -- they could put their noses to the grindstone and get monotonous but necessary stuff done in ways that I just don't seem capable of replicating, even though I can spend way too much time writing a long comment on a blog post about polygenic embryo selection. There have been more than a few occasions on which I've envied them.
Now, I will readily admit that I can't point to a particular correlation between, say, IQ, or ambition, and some other high-risk or anti-social characteristic. But I'm a Hayekian and I worry a lot about about the pitfalls of trying to design or otherwise control complex systems with lots of emergent, difficult to discern but absolutely critical characteristics. There's also an evolutionary argument that the characteristics that people are are using to select their embryos would already be the norm if they are such an unalloyed good. And finally, I worry that we won't appreciate some of the characteristics that people have until we've bred them out of existence in the pursuit of a better polygenic success score.
At the end of the day, I just get really nervous about all the subtle, unforeseen side effects of trying to optimize on a small number of traits.
Footnote, and disclaimer: My view is very likely biased by ethical objections to IVF itself, so consider that when evaluating my arguments.
"Suppose you invest the $3,250 in an index fund that makes 7% over inflation, then give it to your future child when they turn 45 . . ."
Good for you, Scott! You thought about discounting. As an economist, I was working it out in my head and then saw you had the idea. But you didn't do it right. Here's my back-of-the envelope thoughts:
Suppose we can add 2 extra years of life by spending $10,000 today, and we can get 4% if we invest the money instead. By the Rule of 72, the investment will double every 72/4 = 16 years. Average longevity is about 80 years, so we get 5 doublings, so the $10,000 would be $320,000 at age 80. Would someone rather have $320,000 at age 80, or 2 extra years of life? (Or, if you don't care about heirs, $160,000 at age 64 or 2 extra years of life at age 80?)
The average American would prefer the money, I think. But the person thinking about spending the $10,000 is surely top 10%, and they might well prefer the extra years. Except, for them average longevity is higher, maybe 90, so it would be maybe $500,000 for two years. Keep in mind also that for anybody, their last two years of life are apt to be feeble, even though most rich people are doing pretty well at age 80 because they're living 10 more years.
> It’s hard not to love this technology.
I don't know, man, the vibes on this are fucking awful. And while I acknowledge how weak an argument that is, the post ends on this note:
> At some point your answer has to be romantic and aesthetic
My romantic, aesthetic self is screaming "bad idea" while my rationalist self makes fussy arguments about how this is fine, actually.
The trouble so much in the current incarnation of these technologies, which really do seem at worst likely to be harmless and at best to meaningfully improve some people's lives. It just seems somewhat obvious that there is an arm's race dynamic to this that won't permit us to get off this ride even if we don't like where it leads.
It depends on how much sense people have. Discussions below about animal breeders breeding to standards which aren't designed for the benefit of the animal give a hint about the risks.
For example, height seems like a purely competitive trait. It's not like there's something intrinsically good about being tall, but there are advantages to being taller than other people. There comes a point where being taller leads to health problems.
It's possible that there'd be selection for girls with very slender builds. Osteoporosis later in life? Who cares?
I wouldn't be surprised if some parents select for children who are extremely "typey" for their ethnicity. Very light, very dark, facial features on the extreme end of their ethnicity.
The net effect might still be good, but considering how obsessive people are about competition, I don't think trait selection can be reliably centrally controlled.
>For example, height seems like a purely competitive trait. It's not like there's something intrinsically good about being tall, but there are advantages to being taller than other people. There comes a point where being taller leads to health problems.
Yup! I inadvertently echoed you before seeing your comment here.
"I care about IQ because I bang up against the limits of my own a thousand times a day, and I hate it." I'll take Scott's word for it, but can't it also be the case that what Scott truly excels at - let us call it syndetic intelligence - isn't covered by IQ tests?
IMHO, the right thing to do when getting into a discussion that brings about the comment "how exactly does this differ from eugenics?" is not to either run for the hills or dig in, but to reread Franz Boas's classic article
https://www.jstor.org/stable/6055?seq=6
The passage most relevant here is the one starts in page 476. Sure, you can say some of these things about human educational systems, but, say whatever you want about them, they are tolerant enough to allow for all of the following: (a) Scotts, Asimovs, etc., (b) great researchers, (c) great artists, etc. (Of course only a fraction of humanity gets the chance to truly develop its talents, and that's intensely shameful, but it's a different issue.)
Lots of words to say "kill defective children" and yes, the unborn child is a child.
Hard no.
This. Per Wikipedia, only 30% of natural conceptions result in pregnancies, the rest mostly fail to implant. [1]
Because, as you say, unborn children are still children, I consider couples who try to get pregnant through PIV sex to be monsters who are willing to statistically murder an expected two children to get one life birth.
</sarcasm>
[1] https://en.wikipedia.org/wiki/Implantation_(embryology)#Clinical_significance
Are you arguing that having a miscarriage (something you can't control) is equivalent to creating excess embryos with the PURPOSE of doing so being to only select the best one, and to destroy the others?
It's the pro-abortion set's attempt at false equivalence: an accident is exactly the same thing as deliberate murder, because death is the result in both cases, so murder is not in fact murder.
So for instance, speaking purely hypothetically, if you laid in wait for quiet_NaN and beat them to death with a baseball bat, that is *precisely* the same thing as if they slipped on the wet street and fell in front of a bus that ran over them. You can't be blamed for killing them any more than the bus can! (hence, I suppose, all the news reports about "cars running into crowds" and not mentioning the driver who rammed his car into that crowd, unless the driver was a white conservative guy).
Intention? Deliberate action? One was caused in order to bring about death, the other was an accident NaN might have survived even if severely injured? La-la-la, can't hear you, sticking my fingers in my ears, abortion is the exact same thing as miscarriage or stillbirth!
Indeed. I have seen a lot of false equivalences made in this comment section. Frustrating to see Scott arguing that choosing to take prenatal vitamins to prevent spinal bifida (nourishing an existing human life) is equivalent to selecting one embryo with a lower chance of spinal bifida out of a pool of embryos and destroying the others (deliberately killing embryos)
If this technology didn't require killing unwanted embryos and somehow was only ensuring only very healthy embryos are created, that would be a different story and I wouldn't have this particular objection to it.
A miscarriage is something which happens after implantation. We are talking about the worth of an embryo before implantation, which you assert is equal to that of a child, because "the unborn child is still a child".
I will grant @Deiseach that my comparison was a bit inexact. On your "embryos are kids" basis, the couple using IVF (with or without embryo selection) is basically creating a bunch of kids, then picking (either at random or through careful selection) one to keep and murders or freezes (which comes to the same thing for most of them) the others.
The couple trying to conceive naturally is doing the equivalent of creating a baby and then trying to smuggle it through the cargo hold of an airplane, where it will suffocate 70% of the time and be fine 30% of the time, until they succeed in getting one to survive. Sure, they do not want any of their smuggled babies to die, but they are also willing to accept that risk to get a kid, in a way which is not entirely dissimilar to the feeding shelter cats to coyotes meme. [1]
I will grant you that in this hypothetical, the IVF equivalent is likely slightly worse, deliberate murder, while the smuggling might be depraved-heart murder or manslaughter through gross negligence depending on whom you ask.
Furthermore, I will grant you that there is already historically established precedent that it is fine to bring kids in the world even though they have a high chance to die in infancy, because that was the case throughout most of human existence and is considered acceptable in all cultures. (But then again, we should be careful to conclude "something is ok" from "something is widespread", for example, some forms of rape were rather widespread historically and yet I am still opposed to it.)
However, even if I were to accept that these gambles are necessary sometimes, I would still expect the parents to do their utmost to optimize their kids chances to survive. A mother who refuses a supplement for a child which would raise its survival chances from 50% to 55% because she believes that it is up to god to decide the survival commits 5% of a depraved heart murder. With the implantation probability, I do not think anyone tries very hard to optimize it.
So, in short, my problem with religious right generally is that while I recognize that the idea of protecting the unborn is ur-Christian, they do not generally act in a way which is consistent with what they claim they value.
Take abortion. If you consider abortion to be equivalent to murder, then you would obviously want every girl to get on birth control before menarche. Sure, this might enable women to commit all sorts of lesser sins, like fornication or having-sex-on-birth-control, but surely a single act of filicide would stain her soul more than a lifetime of sinful sexual encounters?
Instead, the religious right is also bearish generally on birth control, which should be the lesser evil by orders of magnitude. From this, I conclude that the main goal is to coerce women not to have sex outside committed relationships on pain of being a single mother. If half of the unwanted pregnancies in Texas are terminated out-of-state and the other half are carried to term and serf as a cautionary tale for the other woman, that is clearly preferable to trying to avoid most of these pregnancies through state-sponsored contraceptives, even if that option would be much preferable to a utilitarian Christian.
With the status of the unimplanted embryo, I observe the same, but on steroids if anything. While I am sure that the onset of menstruation will make a couple trying to conceive very sad, I am also sure that they will be orders of magnitude more devastated if their two-year-old drowned in a pool.
From IVF, we know that implantation rate goes down with age, from 37% at 34 to 12% at 41. This is likely similar when trying to conceive naturally. So even if you consider it acceptable to create "kids" where two are expected to die for every one which gets born, if you care about about unimplanted embryos, you should probably stop trying at age 35. At age 41, you might make as many non-implanting embryos per birth through PIV sex as a 34yo undergoing IVF.
Again, if the religious right genuinely cared about the fate of unimplanted embryos, they would tell woman to be celibate or at least use birth control between age 35 and menopause. I mean, unlike the drowned 2yo, these are human beings who die before being baptized!
It feels a bit like the US caring a whole lot about the human rights situation in Iran (which is their enemy) and not a whole lot about the human rights situation in Saudi Arabia (which is their ally), so that one might conclude that human rights are mostly of instrumental value to them as a way to bash their enemies.
[1] https://ifunny.co/picture/my-neighbor-told-me-coyotes-keep-eating-his-outdoor-cats-zvaaz5r47
Leaving aside the political and religious right discussion, I'll say that embryos should have status as the earliest stage of persons and the decision to use the technology Scott is describing should include consideration of their status. Destroying an embryo is not morally neutral.
The topic at hand is whether it is good to use technology to select from multiple embryos that are created for the purpose of being weighed against each other, and most destined to destruction.
I am just arguing that giving unimplanted embryos the same rights as born children, as the OP was heavily implying, would have a lot of unintended consequences as far as normal human behavior is concerned.
Personally, if I was an embryo created through IVF, I would very much prefer to be created, even if there is a 90% chance that I am discarded and another 50% chance that I will die through natural causes. Suppose that the IVF technician already has picked up a sperm and an ovum an just as they move one to the other, they get a phone call "the implantation was successful, we will not need another embryo", and they proceed to terminate both cells separately. For the "don't destroy embryos" perspective, that would be better, but from the perspective of the prospective embryo, getting created, put in the freezer and then (very likely) destroyed would still seem like a better deal. It is not like things made out of a few cells can suffer, any plant will likely outperform them cognitively by a mile.
My very argument is that creating and destroying an embryo is morally neutral in itself (apart from being a waste of time and resources which could be better used elsewhere).
One way it would make a difference is if the embryo was ensouled at the moment of fertilization, and souls were either in limited supply or would suffer. But again, if we suppose that the human reproduction system is divinely inspired and not designed to cause a lot of unnecessary suffering (which seems a bit unlikely to me given what I know from history, but theodicy is another problem), then we have to conclude that during a woman's life, a few dozen embryos which never implant are created and starve was fine with god, and he certainly did not give us commandments which could have easily reduced their numbers.
"I care about IQ because I bang up against the limits of my own a thousand times a day, and I hate it."
From what I understand of very intelligent individuals who have made major achievements, I think this might mischaracterize what intelligence means. I think very intelligent individuals "bang up against their limits" all the time. In general they don't find it simple to do challenging and complex things, they spend tons of time and effort developing their skills in a domain and often feel like they aren't able to surpass a perceived limit. It often seems like obsession with something is a key characteristic to achievements that cause someone to be perceived as a genius.
I'm not saying "your limits aren't real." There are also people who feel like they are banging up against their limits and never surpass them. I'm arguing that, qualitatively, "intelligent person who feels like they are hitting their limit and never surpasses that limit" and "slightly more intelligent person who feels like they are hitting their limit but persevere and eventually achieve at higher and higher levels" probably feel about the same in the moment. And polygenic selection isn't going to change that basic phenomenon.
I might even argue that "banging up against limits" might be more of an intelligent-person phenomenon rather than an unintelligent-person phenomenon. It's not hard to live a life where you never challenge your mind to do new things. Continuing to challenge yourself and seek your limits is something I associate with intelligence.
Yeah, IDK Scott. Maybe you get 10 embryos, but they might not all stick, and many families need every embryo they get for successive tries. IVF communities are turning away even from conventional PGT-A testing, which just looks at whether there is a typical number of chromosomes. The accuracy behind PGT-A is being questioned, and with good reason, as there are some people having normal healthy successes with embryos that were deemed abnormal or mosaic. Because each embryo chance is so valued, people can't afford to throw any of them away, and the more you can retain the better. For these reasons, people are trending away from testing at all; most people just want 1 healthy baby, and my experiences along with participation is many IVF communities suggests that's all you can really hope for.
The personhood argument is well-worn territory, but apparently "proper thematic understanding of Gattaca" is not.
It's been some years since I watched it, so I'm sketchy on a few details, but from what I remember:
- It's not like "everyone else" genetically engineered their kid. The implication is that there's a large underclass of people whose parents couldn't or wouldn't genetically engineer them.
- Then there's Jude Law, who becomes disabled for non-genetic reasons, and realizes that he's suddenly in a world that only values you if you're a perfect Adonis.
- There's the doctor at the end of the movie who <SPOILER ALERT> lies to get Ethan Hawke into the rocket because he too has a non-genetically engineered kid who's been denied opportunity because of his genes.
The dystopia is that in the world of Gattaca, your genes are what define who you are, from career opportunities to what you pay for insurance. The system goes far enough down this rabbit hole that genetic potential replaces revealed achievement in a kind of false meritocracy.
At least one of the themes is, struggling against our imperfections is part of what makes us human. On, as Scott would say, an aesthetic level. And there's something insulting to our humanity about the idea that we could select away our imperfections.
This seems like "I have read a dystopic novel where doctors are trying to cure broken legs and depression with antibiotics and cause a lot of suffering, therefore I am against antibiotics."
If you hire an employee for a non-reproductive role, their actual IQ will obviously predict their performance much better than a genetic prediction of their IQ.
I do not see why this would run into a false meritocracy when a real meritocracy is so much more profitable for everyone involved.
> At least one of the themes is, struggling against our imperfections is part of what makes us human. On, as Scott would say, an aesthetic level. And there's something insulting to our humanity about the idea that we could select away our imperfections.
Tell that to the woman who decides to gets her breasts replaced preemptively because she has a family history of breast cancer, and to her sister who does not and dies of cancer at age 40.
Tell it to the IQ 75 person who will likely never achieve a lot of things most people here would take for granted.
Tell it to me, who has spent most of his adult life dealing with depression.
Are our struggles aesthetically pleasing to you?
While struggling is human, it is also very human to improve ourselves where we can. If a human is looking for a reproductive mate, they will not pick one at random but carefully consider which person likely would pass beneficial genes to their offspring. And it is bloody human to cheat where we can, too. If a human gets attacked by a coyote, the human will not try to fight with tooth and claw, he will use a club or a gun when he can. Do you really expect us not to cheat at genetics now that we can?
FWIW, using embryo screening, humans will likely still remain diverse and imperfect for the foreseeable future.
Is selection preferring boys or girls embryos? If none then how and why it is tweaked this way?
> Here’s the same table for other conditions, courtesy of Genomic Prediction (except I flipped the colors from the original, to match the one above):
I do not think that this is the "same table". The psych table is clearly symmetrical. For that one, I think I at least have an intuitive understanding, roughly "the Pearson correlation coefficient between the alleles which contribute to PTSD and the alleles for ADHD is 0.7".
The other one is much weirder. First off, there is likely a typo in the caption (4000 base pairs, not kBPs).
I can kinda make sense of figure 6 in the preprint. There is a bit of asymmetry, but this comes from different normalization: if you have 100 predictors for asthma and 10k predictors for basal cell carcinoma, and the overlap (within 4kBP) is 10 predictors, then you might have a correlatedless of 10 (implied: %) in the one direction and 0.1% in the other, fine.
But I am still puzzled by figure 7, where they put in the sign of the correlation somehow. First off, I would say that if your method says that genes for height are 83% correlated with genes for height, your assumption that anything within 4kBP is the same is probably not totally solid.
Systolic and Diastolic blood pressure shared 35% or 38% of their predictors in figure 6. (Only in the text, they say 18% and 19% instead.) In figure 7, their not-quite-correlations are 50% and 51%. I interpret this as "the most important predictors for systolic are also the most important predictors for diastolic, and point in the same direction".
The most significant negative correlation is between gallstones and hypertension in one direction, -80%. In the other direction, the effect is much smaller (likely because the figure 6 relatedness in one direction is 11%, but in the other direction it is 0.18%). But it is also positive: 0.067%.
Or take pulse rate and systolic blood pressure, which I picked because they share the same correlatedness in figure 6 (12%) and are also next to each other in the (for some unfathomable reason alphabetically ordered) list of measurements and conditions. At least, for them figure 7 has a positive sign in both cases, which seems sane. But why is the magnitude so different? 7.7% vs 18%.
Ah, I think that the answer is in equation 5, where beta_j (the correlatedness with the SNP which is currently not normalized) is only used directionally, in sgn(beta_i*beta_j). So if you have a correlation where beta_j is positive but very small, that will yield a large correlation in that direction (because the amplitude of beta_j does not matter), but a small correlation in the other direction.
For that matter, how do you even define the direction of correlations for single nucleotide permutations? Say that at a certain point in their genome (site X), half the population has an A and the other half has a G, (with C and T being directly lethal, for simplicity). The A variant has an expected systolic blood pressure which is 2 mmHg higher. 200 base-pairs downstream (site Y), you have another SNP, where you have either A or T, with T having an expected height which is 1cm higher. Due to our "less than 4kBP apart is the same thing" approach, we need to lump them together. But should we say "X=A and Y=A are basically the same thing" or "X=G and Y=A are basically the same thing"? This will affect the direction of our correlation! If on both sites, one variant is dominant, we might call the other variant "the weird one" and lump the weird ones together.
Finally, I do not like the overall smell of the data. For example, consider the field "education years". From priors, you would not expect that to be strongly correlated with most clinical diagnoses, and likely rather negatively than positively. Well, in the one direction, education length variants are positively correlated with everything except for "CAD 20 k" and T1 Diabetes, the largest effects are for two types of cancer (>30%). In the other direction, it is positively correlated with everything except that mysterious "CAD 20 k" again, but mostly on a level of a few percent. On priors, I would have expected that the correlation between education years and genetic diseases is mostly negative, because I would expect most variants which increase the disease risk not to have a positive side effect (if you count a longer education as a positive side effect).
In conclusion, I support superbabies so that future generations will have more brainpower to understand such papers.
Anorexia seems like a strange example to use for the More IQ Costs argument when there's a dozen alternative examples. Isn't anorexia much more prevalent in girls or is there an understanding this is only a bias in treatment?
If you ask almost every parent who is about to have a child, “What is your greatest wish?” You will most likely hear “I just want my child to be healthy.” If you would ask a public health professional, “What is the most important factor that determines whether a child is born healthy?” You will probably hear: “The age of the parents.” Consequently, on a population-wide basis, the most effective way to have a higher percentage of the children born healthy is to decrease parental ages. This is opposite of the current trend for increases in parental ages, especially among people who are likely to achieve graduate-level education. Restructuring the education system to accommodate parents better, including undergraduate parents is probably the best solution to this problem. It is extremely unfair that so many women need to wait so many years to have children and then feel compelled to go through IVF. I say women, because they are the ones who have to struggle through it physically. I also recognize that men are also extremely worried about having healthy children and may not feel confident enough to take on the responsibility of fathering children until they have enough income to provide for their families. So, it this way, they are under pressure too. Ultimately, a better public policy solution will bring better medical outcomes, as well.
Importantly, I think most of the age effect comes from the age of the genetic material, not the age of the (surrogate) mother having the baby.
So a kind of high tech solution would be to just put gametes of everyone into the freezer after puberty -- which admittedly will be a much more pleasurable process for men than women. Then 30 years later, instead of going through the process only to get gametes which have accumulated decades of mutational load, you just open your freezer and take the material from when you were 15.
On priors, I would expect that sperm would benefit more from early extraction and cryo-conservation as opposed to ova, because the oocyte is basically just passively waiting while the spermatogenic stem cells probably pass through a lot of generations during the life of a man. (Hm, a computer scientist could easily create a scheme where the generation count since conception of sperm cells would only scale logarithmically with the total amount produced, but knowing evolution no computer scientist was consulted when implementing the system which decides which primary spermatocytes undergo meiosis and which are kept for mitosis as required.)
The developments you describe are very positive for prospective parents in the future. I hope that most people will have such choices to help them if they find themselves in situations when they have to have children later in life. According to your example, the biological mother will be 45 years old when her first child is born. But this is not a viable solution on a population-wide basis. The 45 year old mother would have difficulty baring her own child, even if she wants to. She would not be able to breastfeed her own child, even if she knows that this is what’s best for the health of her child. Psychologically, she would have suffered through suppressing her natural instinct to have a child, which is a heavy emotional burden for years. Plus, the constant fear that the eggs that were stored for her have not been damaged, somehow. There are many issues regarding surrogacy as well. So, I still feel that a better public policy solution is preferable for the general population.
Waiting until ~after 30 as a matter of course really is an awful trend; my wife and I woke up to it juuust in time to manage a normal healthy family.
Reforming the education system as you describe would be a good start, but then there's the more subtle problem of career advancement in the workplace/job market. That's much harder to fix without some pretty heavy-handed intervention that would seriously clash with the notion of free enterprise. And, ultimately you're either spending a given daylight hour on accomplishing stuff at work, or on childcare: even if you force the structure to not penalize child-havers in terms of material rewards, there's still the issue of wanting the feelings of accomplishment that working hard in an interesting job can bring.
I really think this is the single most difficult and important issue in terms of how modern society is shaped. If it somehow gets solved, life in developed countries will become much less dysfunctional for everyone, and more just for women. I so far can't even in my imagination think of a good solution. At the very least, though, we're at least clearly on our way out of the woods on "more👏45-year-old👏unmarried👏girlbosses👏having👏IVF👏babies". Now here's hoping the tradwife meme doesn't take us too too far in the other direction...
I applaud all these developments. I have a cousin who has an advanced degree and was able to bare her youngest child with the help of IVF and without the help of a surrogate at the age of 52. I am very happy for her! I had another cousin, who got married after age 30, went through a few rounds of IVF unsuccessfully. She had serious complications after her IVF treatment that required major surgery. Unable to cope with that psychologically, she committed suicide at age 45. I still talk to her bereaved mother every day to give her some emotional support after the death of her only child. In sum, I definitely support IVF. I think it should be widely available. At the same time, it is not a viable solution for many women. We still need a better a better public policy to support young families. Yes, it is very complicated, but we must think about workable solutions.
Sasha Gusev has waded in on X with some questions about the lack of transparency in Herasight's methodology—"while clearly conceived and marketed around the idea of IQ prediction, has also elected to hide all the relevant validation until some date in the future."
He also lambastes Orchid, Nucleus, and Genomic Prediction for their sloppy methodologies: "People who care about this technology should be furious at Nucleus and their collaborators (as well as Orchid and Genomic Prediction for their own errors). Finding such flaws should not require reverse-engineering by a competitor. These products clearly need independent audits."
And there's a cast of sketchy promoters of race-based genetics who are in the Herasight orbit.
https://x.com/SashaGusevPosts/status/1951085088290029845
Threadreaderapp link for those of you without X accounts...
https://t.co/CEtrXCW3GK
As for the risk-benefit calculus of using this product, Gusev wrote: "I personally think the benefits don't outweigh the risks and am genuinely bothered by the dodgy roll-out and affiliations. But I don't have any special insight on the risk/benefit calculation and I think people should be in charge of their own reproductive choices."
With regard to claims of accuracy, it seems to me that this should be something which should be easy enough to benchmark.
Find elderly sibling volunteers who are not in any genetic database. Sequence them (there genomes, not the humans), and hand the data to the prediction companies. Ask them to predict the relative genetic outcomes ("sibling A will have five more IQ points", "sibling B will have a +2% absolute risk to develop breast cancer", etc).
Of course, the benchmarking has to be done by a neutral third party. Preferably, have the predictors run their models on an offline laptop on-site, so that nobody will be tempted to id the volunteers and google them.
In the long run, I think we should have open source predictors, so that independent researchers can examine how they come to their prophecies and how much they are just testing for confounders.
Number-crunching through genetic databases is likely not something I would want to try on my laptop, but it also does not seem like the kind of thing which is on par with training LLMs. Basically academics should be happy to do this for publications ("we found a SNP which explains 0.01 IQ points between siblings"), and then you "just" need someone to sanity-check and curate all these findings into a model.
>Everyone else probably wants a generically healthy embryo with low risk of all conditions. Exactly how this works depends on the customer’s own values - would they prefer an embryo with lower cancer risk to one who will have fewer heart attacks? - and the exact benefits will depend on how parents make that decision.
"Good problem to have" though it may be, this level of information/control sounds like an absolute nightmare. I need to make a choice among several points on a many-dimensional pareto front? And if I choose wrong my children suffer? Wow, sign me up!!!!
Actually, I *would* sign up, but I would be absolutely miserable over it.
I see huge risks in wide adoption of such selection. If evolution hasn’t weed out some genetic variants over millions of years it means that they were useful at some point in time in some context. Even if we ignore possible interactions between genes (which can at least potentially produce unique traits), there are still environmental factors making some genes more or less important. We can expect that some genetic variants widely presented and seen negative by scoring was positive some time ago so collective evolution process “decided” to keep it.
You can say “evolution fails at fast changing world like ours, we are no longer exposed to lots of past environment factors”, but are you sure we are good enough at determining variants no longer relevant? Will our quite simple polygenic models and weighting of diseases hold in 20 years? Are we sure there wouldn’t be local famine where genes giving us diabetes become critical for survival? We once thought appendix is something unnecessary and dangerous and blamed evolution for it, but now we know it’s important. Are we at point where we can outplay evolution at world population scale?
Of course we can say that this scores are best estimations for now, but I haven’t heard about any adaptation of models for expected advances in medicine for example.
Speaking about IQ specifically. We know it depends on so many genes and we know that they were under high selection pressure. Have we checked that good variants from 20 years ago are still good today?
Polygenic scores or other models might be overfitted for current state of the world and it’s not just calories intake. Imagine gene that allows child pay 30% more attention to face on screen. Probably negative in 90s (more TV), probably positive during Covid. Or imagine gene reducing long term effects of Covid on brain but lowering IQ one point. It probably kept in gene pull by evolution because we get pandemic infections once in a while but of course it’s negative for carriers between them. Since IQ is depending on so many factors and so sensitive to them I expect any model not taking it into account to lose accuracy fairly quickly (of course there will be more robust genetic marker). It is not a problem for personal decision, you just expect less accuracy, but it will have unintended consequences if half of population adopt such models.
Also wide adoption of technology will significantly reduce variance of population and we can end up with higher average but lower let’s say 90th percentile of IQ. At least at this point of time I see 90th percentile more important for technology and quality of life progress
This frustrates me immensely, because you are focus on material issues when cultural issues will be a far bigger threat.
The big issue is that children routinely struggle against being dominated by their parents's wishes and values. Currently the idea of serendipity helps keep this in check: while children are begotten from parents's genetic material, they are a blending affected by time, chance, and many other factors.
If you select for IQ, you are going beyond just giving them material advantage; you are changing the relationship between parent and child into parent and product. It's bad enough that parents expected their kids to be little geniuses and get mad they underperform or struggle, what happens when they bring out the invoice from the company saying they paid for you to be built better? Psychic damage will be immense, because products are defective.
Plus you always push values when you do this. The focus on IQ is mostly due to upper-middle class competitiveness in seeking power and status. It's not always about the kids because the kids may be happier with less status and less pressure. But you don't want to enshrine values through genetic selection of traits because a lot of them are full of shit. That kid may be happier not trying to be a master of the universe, and maybe we need less jockeying for power and status overall.
>changing the relationship between parent and child into parent and product.
<mildSnark>
( turn on the liturgical music )
( tweaking the quote )
...created not begotten...
</mildSnark>
I appreciate the joke, Mr. Soreff 😁
I think "Frankenstein" is the cautionary tale here; Victor sets out to create his idea of the perfect being, achieves it, and is then horrified by the result and abandons the child he made, with bad results.
"Begotten, not made" may be good enough for the Creator of the Universe, but not for middle-to-upper middle class white Westerners competing in the ceaseless strife to get Junior into the best preschool onwards.
my favorite example in media is an anime called Geneshaft, which is 50% silly anime tropes like cute girls and giant robots and 50% weirdly hard science fiction about this.
Like the spunky protagonist gets down about her ability to pilot the robot and calls her "mother," who assures her she hand picked the necessary traits to succeed in her role, talking as if she is a robot.
Or how with genetic engineering society does prefer girls, with boys mostly there only as needed and with a monitor to keep them in check. Girls are generally more stable and reliable in that future, and in a weird way our own.
anime and manga are odd. Oh there is now a rationalist manga lol, Qualia the Purple. You want to explain what a p-zombie is that one is a terrifying example of it.
Many Thanks Deiseach!
>I think "Frankenstein" is the cautionary tale here; Victor sets out to create his idea of the perfect being, achieves it, and is then horrified by the result and abandons the child he made, with bad results.
Yup, oops! As a cautionary tale (and cautionary tales, both as fiction and as poor predictors of reality, themselves need to be treated with caution...) it is indeed a good match.
>"Begotten, not made" may be good enough for the Creator of the Universe, but not for middle-to-upper middle class white Westerners competing in the ceaseless strife to get Junior into the best preschool onwards.
Hubris? Who, us? :-)
( Admittedly, as a childfree man well past his potentially reproductive years, I don't actually have a dog in this fight... ).
Not sure I get the snark, as that's my whole point: we don't relate to our kids as a creator does to his creations, we beget them. God can make but He is assumed to always be good, and also far superior to us in nature. But we are neither.
The "only begotten" is theological in part to show why Jesus can be the sacrificial lamb, and also to prevent more divine sons of God popping up to claim the mantle. We are sons of God by adoption, and also the bride of Christ in analogy. we pointedly don't relate to God only as things to fulfill a purpose, in Christian theology at least. The more strains of it think that, the more inhuman they become (predestined vessels of wrath)
the angels were created, and christians don't know what to make of them.
Many Thanks!
>Not sure I get the snark
Well, I was poking fun at the theological distinction, and agreeing that embryo selection (and genetic engineering, if it becomes an option) does have some resemblance to the inverse of the theology (modulo the Nicene Creed's "begotten" isn't _really_ the same as human begetting...)
Now, I, personally, don't see this change as all that much of a big deal. Parents try to mold their sons and daughters in many ways today, so I see this as an incremental change.
( Admittedly, as a childfree man well past his potentially reproductive years, I don't actually have a dog in this fight... )
>The "only begotten" is theological in part to show why Jesus can be the sacrificial lamb
I've always found the force of the distinction to be odd (admittedly looking at the whole thing as an atheist/agnostic). An omnipotent deity could do whatever it wants - making a sharp distinction between the possible _mechanisms_ by which it generates a more-or-less related deity, or anything else, seems like a strange thing to care about to me. Maybe for a deity with limited powers we'd care about side effects of different mechanisms, but that isn't the model in Christianity, as far as I know.
> What if they selected your brother for everything great, then had you naturally?
Presuming your brother is older (because "then had you"), there are two possibilities for change: they could have avoided selection of traits for the first child, or they could have selected traits for the second child. Considering the first case, we're just envious, and somewhat despicably so. Considering the second case, that would only improve things for this child if they believe it was better had they never been born. There's no "you but with better genetics," because that "other you" is actually someone else, which is more transparently obvious in this case than the usual "put yourself in someone else's place" thought experiments, since it's about embryos that already literally exist.
Creating dozens, or hundreds of embryos with the purpose of destroying all but the "best", "best" being determined not by viability but by subjectively desirable traits, is unethical. Two reasons off the top of my head.
1. embryos have moral value because under natural circumstances (regular ol sex resulting in conception) they will likely develop into people. Yes miscarriages occur, this doesn't change the argument. No, sperm and egg separately do not have the same moral value.
2. It devalues human life in general by making the explicit determination that someone with blue eyes is more worthy of existence than someone with brown. Or that someone with a higher IQ is more worthy of life than someone with a lower IQ. The foreseeable downstream cultural effects of this are frightening.
Yeah, there appears to be an incentive to create as many embryos as possible.
The upper limit is just the point at which the woman becomes too physically uncomfortable for additional retrievals and the pool of trait combinations tapers off. The more traits get selected for, the more embryos would need to be created to get a desired combo.
Embryos created via IVF do not "likely develop into people". You have to go through the active step of carefully inserting the embryo into the uterine lining of the woman for this to happen.
Yes, that's true, but my point was that the embryo has a purpose and a potential. The purpose and potential is to develop into a baby and eventually into an adult human. The fact that an embryo is an early stage adult human warrants consideration before creating them and destroying them.
Many parents might feel bummed if their first choice of embryo doesn't take.
Welcome saudade, my old friend.
Vaccines are one of the most amazing human inventions and have saved countless lives, beyond any statistical doubt and with a very large effect size. Yet there are countless anti-vaxxers who say "my cousin got the flu vaccine and then felt sick, vaccines don't work, I'm letting my kid grow up naturally."
The effect sizes for these effects are far smaller. If this becomes popular, millions of families will have a designer baby that grows up to get lower grades than a non-designer baby and they will say "that designer baby thing doesn't work."
Humans have a very strong naturalistic fallacy, that anything natural is good and unnatural is bad. Obviously these aren't ironclad -- not everyone believe the naturalistic fallacy, most people give their kids vaccines. But still, I predict those forces will put a lower ceiling than optimists might like on rapid adoption of this technology (assuming it works as well as claimed to begin with.)
The naturalistic fallacy is not a fallacy. It is a fundamental human value.
If your view is that unnaturally screening embryos for higher IQ is a necessary evil to combat the challenges of this world, then your opinion is worthy of a discussion. But if your view is that screening embryos for IQ has no moral problems whatsoever, even though it goes against nature in an unprecedented way, then your morals are alien to mine.
>The naturalistic fallacy is not a fallacy. It is a fundamental human value.
Yet you type that on a machine with transistors of elemental silicon, a material found nowhere in nature.
And I'll bet that your lifestyle is as far from our environment of evolutionary adaptation as mine or Dylan Kane's is.
The naturalistic fallacy is indeed a fallacy. "Natural is good" is profoundly mistaken.
EDIT: A point you know, but perhaps have forgotten:
_LITERACY_ is unnatural
Reading and writing are _not_ part of our environment of evolutionary adaptation. We have been human for around 300,000 years. Writing was invented only about 6,000 years ago. Children acquire spoken language spontaneously in a family. Reading and writing have to be actively taught to them. If you want to reject the unnatural, you must reject the written word.
There are decent arguments for being skeptical about embryo selection. Scott covered a number of them. The naturalistic fallacy is not one of them.
> “Although in general higher scores mean you are more likely to have a condition, many healthy people will have high scores; others might develop the condition even with a low score”, says the Society for Psychiatric Genetics, as if they have just blown the lid off some dastardly conspiracy.
I know that many people here are used to being surrounded by intellectuals trained to reflexively do Bayesian statistics, but you have to remember that *this is not at all obvious to the average person/most people*. In fact, there's a significant portion of the population who I've come to realize might actually be incapable of understanding that.
For someone who interacts with the public and is aware of the existence of these sorts of people, you can see how someone might come to the conclusion that PGES companies are trying to mislead people about efficacy with the claims they're making (even if it's almost certainly those companies just speaking from a place of the curse of knowledge).
" I care about IQ because I bang up against the limits of my own a thousand times a day, and I hate it. I fantasize about ways to make my children smarter than I am for the same reason a dog confined in a tiny crate might fantasize about getting her puppies adopted out to a nice house with a big grassy yard.)"
Well said. There's another reason, though. High-IQ people can be evil, but they're not stupid. They don't shoot their best friend by accident when they get angry. They understand that someone who said something offensive didn't mean to offend them. They know how to have good manners that please other people at low cost to themsevles. In short, they have positive externalities. And that't even without talking about innovation or artistic creation.
an average person will cheat on his wife, a smart person will cheat on his wife, write a book about how cheating is morally justified, and convince others to cheat too.
Intelligence is just a force multiplier, it isn't any better in a moral sense. There's a bad idea that evil is stupid, but no, smart people commit it too, and can even systemize it.
"High-IQ people can be evil, but they're not stupid."
No, they don't shoot someone on impulse when angry; they go off and plot a tidy murder. As to "understanding someone didn't mean to offend them", please tell the wider world that. There's plenty of smart people (or people who like to think they're smart) getting very angry over offence even if it was not intentional.
I have to quote Auden's little verse here 😁
Note on intellectuals
To the man-in-the-street, who, I'm sorry to say
Is a keen observer of life,
The word Intellectual suggests straight away
A man who's untrue to his wife.
This article is deeply concerning. It fails to address the most obvious argument against selecting embryos for IQ or appearance: that it breaks from nature in a way no other technology has. Selecting embryos for these traits disregards the natural process by which humans are created—something no previous intervention has done so thoroughly. Once we begin down this path, we risk losing what it even means to be human. What do you truly value? Aren’t your values rooted in the human condition itself? Isn't "the natural way humans are produced" an essential value? Some problems are so ugly that they are worth betraying nature to solve. Increasing IQ is not one of them.
You say that selecting embryos for IQ or appearance breaks from nature in a way no other technology has. I disagree. Birth control pills are a technology that separate sex from reproduction. 14% of women in the USA aged 15-49 take birth control pills. Only 2% of children are conceived using IVF. This new polygenic scoring technology provides parents with additional information. IVF parents are going to choose which embryos to implant, should they make the choice randomly? Are you against PGT-A screening for down syndrome? Hiding data from parents is not the answer. Humans are diverse with diverse values and will make diverse decisions when it comes to embryo selection. This technology is inevitable.
It's unprecedented not because it's the first technology to "separate sex from reproduction," but because it's the first technology that allows parents to make value judgements about which types of people get to live, beyond screening for debilitating diseases. Intelligence may be a valuable quality in the abstract, but is a smarter child more deserving of life than a less intelligent one? Are intelligent people worth more than less intelligent people? I assume you would say no to these questions. In short, it's immoral to choose who gets to live based on subjective values. I'm not sure if uptake is inevitable, but even if it is, that obviously doesn't mean I need to support it.
Decisions based on "subjective values" happen all the time in mate selection. Talk to people about how they chose their partners and you’ll hear stories like “I noticed her because she wore a red dress to the dance,” or “he looked cool smoking outside.” Subjective value judgements in mate selection are far more impactful than embryo selection. Consider sperm donor selection: a woman flipping through donor profiles is making decisions of who is more "deserving of life" based on appearance, education, health, and personality all before conception even occurs. In comparison, choosing among a small number of very similar embryos is relatively inconsequential. And you asked if an intelligent child is "worth more" than a less intelligent one. Does society value intelligence? Yes, it often does. Intelligence correlates with prosocial behavior and the ability to make meaningful contributions, and society tends to reward it. It is innately human and natural to want healthy, happy & smart children. Breast cancer risk, diabetes risk, IQ, risk of Alzheimer's... I wouldn't exactly call these "subjective values." But, I respect that you don't support it. Society will ultimately decide whether or not to accept this new embryo selection technology.
Good point. You’ve brought me closer to your perspective. That said, I’ve always been uneasy with the idea of mate selection as a moral framework. It feels wrong to judge someone’s worthiness as a partner based on objective traits like IQ, rather than personal, indexical ones like how deeply you connect, or the good they’ve done for you.
Still, mate selection is a natural process; embryo selection is not. And I don’t think the appeal to nature is a fallacy here. Valuing what’s natural is a deep human intuition. Part of me feels that using artificial technology to amplify natural ugliness crosses a moral line.
" I assume you would say no to these questions."
Don't assume that. We've plenty of people who have commented on here that yes, in fact they *do* think smarter people are better and more deserving, and that Huxley's "Brave New World" with embryos treated and decanted and conditioned to fit into certain niches is something we should be striving for, not reading as a cautionary tale: but just think of the soma and the feelies! the dumb people do the dirty dangerous work without complaint and are no threat to us smart types at the top of the pyramid!
Yup! Though my generally positive view on BNW is more complex than that https://www.astralcodexten.com/p/profile-the-far-out-initiative/comment/56490956
Summarizing the value of most responses to this, it's exactly why we need IQ screening of embryos lol.
This cannot be the right framing:
“Consider e.g. Genomic Prediction, which costs $3,250 for five embryos and claims to lower absolute risk of Type 2 diabetes by 12%. That implies that not getting Type 2 diabetes is worth $27,000. Ask anybody dealing with regular insulin injections (let alone limb amputations) whether it would be worth $27,000 to wave a magic wand and not have Type 2 diabetes! It’s not a hard question!”
Linear extrapolation of the utility function to zero Type 2 diabetes risk is wrong. The PRS scores have a floor (well) above zero risk. While I applaud an attempt to quantify what these screens may be worth, a linear utility function can’t be the right way to think about this. It will always be biased towards making this technology seem far more effective per dollar than it is.
Ok, two things:
First of all the floor for type 2 diabetes is pretty damn close to zero if you can modify enough genetic variants. In theory you could get up a ~98% risk reduction if you could change enough genes (though you're not going to be able to get anywhere close to that with embryo selection)
But second of all... I don't understand your objection here.
If I can reduce the odds of getting diabetes by 12%, then surely that is worth 12% of the value of not getting type 2 diabetes, right? Maybe you want to argue we should do some time discounting (frankly I would agree with that)
Scott's 12% reduction is absolute risk. It's obviously true that if the average risk among a family's embryos is lower, the intervention on that disease will be worth less.
You can reduce your risk of getting Type II diabetes by being damn careful about what you eat in what amounts, getting exercise, and getting regular health checkups.
The irony here would be that the genetically selected babies now have much, much less risk of Type II diabetes so the food industry can continue to turn out highly processed, filled with sugar, junk food for fun and profit since the consumers are now able to eat as much slop as they want with much less fear of consequences.
Your “98% risk reduction” switches the topic from embryo selection using PRS to hypothetical large-scale gene editing. Even under that premise, it can’t be 98% of total T2D risk—non-genetic drivers remain, heritability is <1, and PRS captures only part of the genetic share. At best, that number might refer to a slice of the polygenic component—not total lifetime risk.
Back to the actual technology: the blog’s linear extrapolation—“$3,250 buys a 12-point reduction, so a cure costs ~$27k”—assumes constant returns and that the remaining risk is purchasable. But it doesn’t work that way.
Maybe this analogy helps: a seatbelt might cut fatality risk by 50% in a crash. But you can’t wear eight seatbelts to get to near-zero risk. There’s irreducible residual risk from the crash itself.
The linear extrapolation overstates the benefit and risks distorting how we think about cost-effectiveness and where this technology is actually useful.
And maybe you think this calculation isn't literal. It's illustrative. I'd buy that except the author brings up the prospect of not having amputations and insulin injections for only $27,000. That's rhetorically pretty loaded. Either the number is a rough utility estimate and we stay in abstraction, or it’s a claim about tangible medical outcomes—at which point it needs much more caution and nuance. You can’t have it both ways, and that's the point of my initial critique of the framing.
>And then there’s AI. The average age at diagnosis for Type II diabetes is 45 years. Will there still be people growing gradually older and getting Type II diabetes and taking insulin injections in 2070? If not, what are we even doing here?
Very much agreed.
I wish the people in this endeavor well, but, barring a truly Antarctic class AI winter, I can't see them making much difference (from a broad policy view).
Here's my question: out of the four kids in the header image, which would you keep?
Personally I've spent some time staring at it and probably the third? It's got the lowest chance of diabetes, Alzheimer's, and hypertension. But on the other hand it's also got the lowest IQ. How should I be thinking about trading off IQ for a slightly reduced probability of hypertension? Should I just take the smarter kid and be careful what I feed them?
If this is a typical result of embryo selection then this small menu of barely-different options doesn't seem worth the effort or expense.
There is something about this entire discussion that makes me nauseous. I feel like I’ve woken up from a coma 15 years in the future. It’s just so surreal to see this topic garner such serious, meticulous discussion because we’ve actually reached the point where it can’t be avoided. My unwillingness to even want to face the issue also just makes me feel old.
Unhappily, unless we face it now, we'll have nothing to say when it is commonplace. I'm quietly amused, in a grim way, by the silence on stem cell therapies.
You remember, if you're old enough, the dazzling new advance that was going to do away with all hereditary diseases? Being blocked by those stupid old sky-fairy believers and Luddites? We were only five/fifteen years away from no more [insert ill of your choice]!
And then they got the permissions they wanted, after hard fighting and resistance, and the embryonic version was sidelined without fanfare (because that was all about research, in fact,. and not miracle cures) and the emphasis was now on adult stem cell research. Which did *not* require destruction of embryos, which was what us stupid old sky-fairy believers were happy with in the first place. And then attention moved on to "hey maybe we don't need embryos at all, we can convert other cells into stem cells".
Yes, stem cell therapy is working. But it's not the dazzling miracle cure for every ill that the promoters sold it as, and I wonder if this embryonic selection is going to go the same way: give it twenty years or so, and it'll be the preserve of for-profit companies holding out the lure of miracle cures:
https://www.dvcstem.com/protocol
>suppose that I had a natural, unselected child, and that child became high school valedictorian and got into Harvard. Would my first reaction be “Oh no! This slightly raises her risk of anorexia!”? If not, why should this be our reaction to artificially increasing IQ?
...because her genetic risk of anorexia was already locked in at birth, whether or not she went to Harvard, and there's nothing you can do about it now, and therefore no reason to suddenly react as if something new has happened? So you may as well celebrate the good thing?
Whereas, if you are deciding whether or not to hit the 'Harvard + Anorexia' button *before* they are born, then you absolutely have to consider both the upside and the downside.
I think a more analogous story would have captured the point that you are making a decision. Something like: 'If you hear that a child got into Harvard, do you put slightly more effort into checking their eating habits and asking them how they feel about their body, to screen for early warning signs of anorexia?'
But the answer to that is... yes, sure, you should do that if the connection is predictive enough for it to be a rational precaution.
(Note, I take the point of the paragraph, which I think is just 'many people would happily take this tradeoff.' I just think that asking what your emotional reaction to someone getting into Harvard would be is a poor way to frame the intuition, because celebrating a good thing in a world where bad things exist is very different from deciding which tradeoffs to make)
If everyone starts using this technology, it will have unforseen consequences for sure. For example, there is a correlation between schizophrenia in the family and creativity. Maybe if in future, we filter out many of the genes associated with schizophrenia, we will also filter out many of the geniuses.
It’s not just certain ethnicities, believe me. It’s now normal in the US upper middle class to implant only female embryos, and of course (white, native-born) women of the same class talk very freely about how they would abort a male embryo/foetus if they got pregnant the natural way. (Epistemic warning: anecdotal observation.)
I think this article overlooks a few important points.
First, there is a concerning genetic correlation between IQ and Autism, which I raised in my article (https://theinfinitesimal.substack.com/p/science-fictions-are-outpacing-science) but which this post avoids mentioning in favor of anorexia. Given the general concern and anxiety people currently have about ASD risk, especially in the high-edu / high-income clientele that Herasight is catering to, this really should be front and center. If I had to guess, Scott ignores ASD in favor of anorexia because proponents of PGT argue that IQ is only genetically correlated with Aspberger's and not clinical ASD and therefore poses less of a risk. To the best of our current knowledge this is not true, see for example Figure 3 in Grove et al. (https://pmc.ncbi.nlm.nih.gov/articles/PMC6454898/) showing a statistically significant association between the IQ PGS and childhood autism (hCHA). If you're selling parents a product that might increase the risk of childhood autism, I think that needs to be on the warning label.
Second, all of the results for antagonistic pleiotropy are based on population-level estimates which will be confounded by assortative mating (in particular the fact that people with high education / high income / high IQ / high health mate assortatively will make "good" traits look correlated; and the fact that people with mental health problems / low income / low health mate assortatively will make "bad" traits look correlated). What is relevant for embryos is the within-family genetic correlation and this is simply too noisy from the current data (e.g. Tan et al. family GWAS https://www.medrxiv.org/content/10.1101/2024.10.01.24314703v1). There also plenty of traits that do not have good GWAS data or good biobank measurements to evaluate this question at all. So this continues to be a leap of faith.
On the topic of population genetic correlations, the post considers the market benefit of IQ versus the cost of embryo selection. But this analysis uses phenotypic correlations between IQ and income, which is even more confounded, and will include any paths where SES -> IQ and SES -> Income. Interestingly, in Tan et al. the population level genetic correlation with individual income is actually much higher for the Educational Attainment GWAS than for the IQ GWAS. If the goal is purely return on investment and we're relying on population r_g, then selection should be on EA rather than IQ. EA also has generally higher genetic correlation with various health outcomes, too. I'm surprised this wasn't mentioned.
Finally, there are a few references to embryo selection curing or eliminating certain conditions but it is important to understand that embryo selection works very differently for polygenic traits than monogenic traits. When you screen out a rare Mendelian disease, you are removing carriers of the mutation and therefore removing that risk from the offspring and arguably "curing" the disease. But when you screen out for common PGS, you are subtly shifting the underlying liability from just above some clinical disease "threshold" to just below the threshold -- this is where the majority of the "risk" reduction comes from. The easiest way to understand this is with the example of obesity, which is typically defined as BMI>40. Embryo selection can appear to reduce obesity risk by an impressive 50%, but what it actually does is reduce BMI from a mean of 41 to 40 (in obese families). YMMV on whether a BMI reduction of 1pt is worth the cost/benefit but it is clearly not a *cure* for Obesity except for in a very pedantic way that happens to make the statistics look good. This is not a nitpick, because many of the traits that are being screened on are actually just arbitrary dichotomizations of an underlying continuum (Alzheimer's: underlying cognitive / decline tests; Schizophrenia: defined by a sum of symptoms; T2D : Glucose/hba1c levels; Hypertension: blood pressure; Gout : uric acid; VTE : clotting factors; Glaucoma : eye pressure / nerve damage; Osteoporosis: bone mineral density scores). Even for cancer, the most likely mechanism is an extension of the age of onset or rate of growth rather than a "cure" to the disease. In my opinion, it is misleading to present large relative risk reductions on threshold traits when the reality is a small quantitative reduction on the underlying continuous trait.
Fin (apologies for the long comment as always)
I was curious about pleiotropy! I wondered if the correlations shown were population or within-family; concerning to hear they are population. So I take your comment to mean we still don’t have good estimates of pleiotropy from within-family data?
For people already doing IVF, this seems like a no-brainer. But you also suggest people might electively do IVF in order to do this, and it's much less clear to me that is a net win. There are a lot of selection effects that occur in natural reproduction that aren't replicated by IVF (for example ICSI replacing sperm competition), and many potential insults to the embryo that are IVF specific (e.g. freeze-thaw). If you have a small number of embryos to choose from you could easily come out net behind.
This is really compelling—and raises some quiet discomfort for me. It feels like we’re drifting toward this subtle optimization trap, where even our kids’ traits start to feel like variables we’re supposed to tweak.
I get the appeal—if you can reduce suffering or give your child a small edge, why wouldn’t you? But I also wonder what we lose when everything becomes selectable. Some of the deepest parts of being human come from the unchosen: the weird mix of strengths and struggles that shape who we become.