If you're convinced the technological singularity is approx 30 years away, and that it's likely to be ~bad for humans (AI take over etc), then why are you trying for a baby with your wife?
I don't think this is the same kind of antinatalist point of "the world is bad, why bring more life into it". It seems like you seriously believe that something is different about this point in history, and to me it then seems a bit odd that you'd want to plunge someone new in at the deep end just when the robots take over!
How do you reconcile this?
PS: forgive me if I've misremembered about the baby part, think you said that a whole ago!
"And I notice it’s weird to be worried both that the future will be racked by labor shortages, and that we’ll suffer from technological unemployment and need to worry about universal basic income. You really have to choose one or the other."
Of course those two are inconsistent. The correct answer is that technological unemployment is a myth. There is little credible evidence for that outcome. Some who talk about it are (consciously or unconsciously) motivated by a desire to promote UBI for independent reasons.
I'm not particularly "worried" about labor shortages, but of these two outcomes, I consider it to be the one with more evidence, by far.
“there will probably be a technological singularity before 2100”
This really contradicts the science is slowing down paragraph. The one just before it.
I agree that we might probably be able to make babies smarter by 2100, but the increasingly large Amish and orthodox community might be against it. Of course it’s post singularity so who knows what they will believe.
Absent your singularity the population trends for the Amish and orthodox will hit the Malthusian barriers of their more primitive lifestyle. I don’t know much about the orthodox lifestyle, and what I know about the Amish comes from Witness and wiki, but the latter don’t seem to take charity, as their population grows and they buy up more land - I assume this is what they do - that land must become less productive. The US will be a fairly primitive society in that era. And mostly white again. Haha.
The US is, outside these groups, is now a population sink. Population sinks are interesting because all lineages die out to be replaced by new additions, who then die out over time. The descendants of Londoners in the 12th C are not modern Londoners but subsequent migrants (internal to the U.K. mostly until recently).
"Even this isn’t quite right, because a lot of Orthodox Jews do leave Orthodoxy, so along with those 100 million devout Orthodox there will probably be a few dozen million extra Reform Jews with a confused relationship to religion and lots of emotional baggage. It’ll be a great time for the rationalist community."
Go team! (Also, at this point we should be separating out the Ultra Orthodox from the Modern, because the latter can't keep up. It's a good question how many Jewish-sans-Streimel groups we'd have at that hypothetical point. To be honest, Conservative and Reform could have merged in 2015 and nobody would have noticed. The other Orthodox groups are more of a question mark.)
Isn't this a fully general case against caring about the future past a certain point? "Yes a bad thing will happen and it will have bad effects. But conditions will be different, it won't be THAT bad, etc." To be honest this reminds me of what people say about Global Warming. "A 2 degree increase by 2100" or whatever.
I agree that in the longer run sub-populations with higher fertility will keep most likely keep total fertility high enough.
However, if the Amish grow a lot more, I would also expect them to change.
Just like Warren Buffet can make money at 20% a year as long as he's small, he can't keep that up once he has an appreciable fraction of the total economy.
You just accept the population projections as fact, and this seems like a serious mistake. These projections have been wrong in the past, could continue to be wrong in the future, and one may make the case that they will predictably be wrong in the future. Our World In Data doesn't have a proof that their population projections are the most accurate projections possible given the information available in the present; they just have some model, the assumptions of which could be disputed.
Besides demographic shift, my real worry here is the causes of declining birth rates rather than its effects. Are we too socially and emotionally broken to start families, or just too poor and financially insecure? Too individualist? All options feel terrible, and I suspect these three are all true and interrelated.
Point is well taken about unreal years, but it makes me wonder - what is the farthest off real year? That is, the last year about which we can make meaningful predictions in most arenas? 2026? 2030?
If we make it though the singularity, however exactly that looks, i know it may have been half a joke but I could see a world where some form of religion becomes highly selected. The basic tenets would be: don’t do things that destroy the patterns that sustain humanity, like sexual reproduction being tied to fitness, having to struggle to develop, etc. I always call that Space Mormonism but in a science fiction future where there are still recognizable humans I see that as the most likely outcome.
Another point against the dysgenics issue is that to whatever extent the Flynn effect is QOL based, it will be more powerful in developing e.g. African countries which are also making more babies.
So even if the average IQ of the descendants of today's Americans will lower from dysgenics, it will be more than compensated for by 2nd/3rd gen sub-Saharan immigrants (already a smarter than average slice) going through the Flynn effect.
I wish the singularity argument had been the first argument. The "Don't worry it won't get bad until 2100" left me with my mouth hanging open and then when it eventually got followed up with "and it won't matter by 2100" I was able to close my mouth, but I wish I'd been able to know up front that Scott wasn't trying to predict anything about population trends but merely making a point that, like everything else, he thinks population trends are inherently unpredictable on a 80 year timescale.
My guess is that the climate change consequences will make many places unlivable enough to result in extreme strife and significant decline in the population due to wars and shortage of basic resources caused by wars, not by climate change. And not by 2100, but more like by 2040, the way current extreme weather events are happening. Basically take what is going on in Ukraine and add one or two orders of magnitude. It won't be just a shortage of heating gas or wheat in Europe, it will be famine and lack of technological basics all over the world. Population decline will be a consequence, but a welcome one, rather than a cause for worry. This is assuming no drastic changes like AGI takeover.
In terms of the Amish and the Orthodox. My understanding is the Amish can only exist because they occupy some of the worlds most fertile farmland. Their lifestyle/economy can’t scale. The same is true for the Orthodox community in Israel. The most common example is they aren’t subject to mandatory service in the IDF. An Israeli without and IDF isn’t an Israel for very long.
I expect the year 2050 to be a real year. I am so convinced of this that large fractions of my total wealth are in retirement savings, which will not be touched until after the 2050s.
Have you decided that saving for retirement is a waste of money? Do you tell small children that saving for retirement is a waste of money? Do you plan to take out a large mortgage, on the expectation that you will never have to pay it back?
Is there any prominent person you would be willing to take a bet with, at any odds, on the year 2050 being post singularity?
Re. "if innovation is destined to be only 10% of its current level in 2100, then a 30% population decline could lower that to 7%.":
Because of our high population, the bottleneck in tech progress isn't creativity and innovation, or even money, but attention. You can see this in government funding: Grant-funding agencies are limited less by budget than by the number of administrators who can oversee contracts. Projects that cost less than half a million dollars a year are often ignored, sometimes (in my personal experience) to the point of the administrator not bothering to read project reports, answer emails, or attend meetings.
You can also see this in the proportion of media attention given to elite colleges. The ratio of students in non-elite to elite colleges has increased by about a factor of 10 since 1950, yet the ratio of media references to graduates of elite vs. non-elite colleges seems to have grown. Go through any issue of WIRED magazine (well, any issue from the 1990s, when I still read it) & see how many articles you can find about research by people who didn't attend MIT, Stanford, or Carnegie Mellon. The proportion of Nobel prizes given to graduates of top-20-in-field colleges has also increased radically since about 1960; in physics, it went from something like 50% in the first half of the 20th century, to 100% after 1970 (last I checked). The ratio of venture capital given to graduates of elite colleges vs. people with no college education was small in the early 20th century; today, it might be infinite.
We have a glut of smart people, yet if anything there are fewer super-smart people at the top. There are AFAIK no contemporary equivalents to Einstein, Turing, von Neumann, or EO Wilson.
No matter how big the population grows, each person still has enough brain space for the same number of new ideas. The only way we have now of countering this effect is to continually fragment into more and more isolated specialties. But this renders everyone less and less capable of recognizing creativity and intelligence in the wild.
So I do not think a decrease in population will decrease effective innovation. It might even increase it.
I participated in the "predict 50 years in the future" contest. I focused on genetically engineering people, which is going to happen and going to majorly impact society, the economy, and politics.
It will likely be the case that there is a majority that is below average intelligence (compared to present average) and a small group of fortunate people who are extremely smart, happy and healthy in 2100. Massive gains can be achieved through genetic engineering, most importantly for the near future might be embryo selection.
Selecting embryos from a small pool can confer advantages, and selecting from extremely large pools can confer large advantages. With enough embryos to select from, it's possible to get massive gains. See Gwern's article on this (https://www.gwern.net/Embryo-selection).
There is good reason to think that this will be achieved in near future, and it will majorly impact society for the better. It's difficult to imagine what a society with a million 250 IQ people would look like. I do think that there will be a lot of interesting political and ethical issues. Will the practice be prohibited? How large will this population be? Will parents leave the country to have genius children? I discuss this in my article: https://parrhesia.substack.com/p/america-in-2072-a-society-stratified
I also think that effective altruists should focus very strongly on protecting this right and start devising rhetoric to defend it. Otherwise, the risk is that it is banned, and that lives lived are much worse. This is an injustice. Also, super genius babies could help prevent X-risks (a point made by James Miller).
Another scenario that can turn the population dynamic upside down will be the discovery and mass availability of rejuvenation therapies. I don't think bological immortality will cause overpopulation, but the population pyramids will start to look *really* weird
From a consequentialist perspective, it seems like we should take the non-existence of people seriously. If we could boost fertility globally and achieve 20 billion in 2100 instead of 10 billion, we are going to see a lot more human welfare. Failing to do this should be treated similar to a persistent disease that's killing literally billions of people. Low fertility must be regarded as one of the worst things in the world at present if we regard future possible people as having equal moral worth. If all that matters is the consequence, lowering fertility should be treated like a massive wave of tens of millions of infant deaths. This should be extremely concerning.
It’s comical to me than people really think depopulation is the big problem facing us. The world population is almost 2.5 times what it was when I was born. Many of the actually big problems facing humanity, like climate change, species extinction, ocean pollution, etc. are automatically improved when you have fewer humans causing them. I think the real fear is that the Ponzi economy is unsustainable unless there is constant population growth (= more people joining the scheme). But all economies eventually fail; just try to spend a sestertius now. So I can’t see that short term pain as some existential crisis.
"I guess it’s still true that if innovation is destined to be only 10% of its current level in 2100, then a 30% population decline could lower that to 7%."
But if it's educated / high-IQ people that have the lowest fertility (such that they're disproportionately responsible for the 30% decline) wouldn't that imply that innovation drops lower than 7%? In other words, the proportion of potential innovators would get smaller over time. This effect might even be stronger at the tail end of the IQ distribution: a small relative decrease in high-IQ individuals would mean a strong relative decrease in potential superstars.
Robin Hanson thinks falling fertility is <i>The</i> Big Problem of our times, but mostly in a total-utilitarian sense* rather than a looming disaster sense:
He has also dismissed writings about technological unemployment. I'm also not too concerned about that, both because of the evidence he's presented about recent history & the near future, and that if we do achieve that our civilization must be succeeding somewhat.
*After writing that I rechecked his post and saw that he thinks that lowering economic growth via lower population & knock-on effects would reduce our robustness to a variety of existential risks. So heightened probability of disasters, but not looming directly from lower populations.
"if we can’t genetic engineer superbabies with arbitrary IQs by 2100, we have failed so overwhelmingly as a civilization that we deserve whatever kind of terrible discourse our idiot grandchildren inflict on us"
I find annoying arguments throwing up our hands and saying we "deserve" bad outcome B if bad outcome A happens. We should try to be robust to bad outcomes. An example of Robin Hanson thinking about that for a low probability but severely negative outcome is here: https://www.overcomingbias.com/2008/07/refuge-markets.html
I should also note I'm much less convinced that the singularity will happen that quickly. If it happens in my lifetime, that will probably be thanks to life extension. 2100 is still far enough off I don't want to confidently speculate about it though.
Also: even supposing that a singularity fails to happen- that the inevitable unpredictable historical twist turns out to be that not much changes- aren't we actually going to need a stable human population at some point? I mean, we can't just keep growing geometrically century after century, right? Why, given a future without much technological progress, would it be better to put off that stabilization until a future generation? Presumably, they would face the same kind of negative consequences from falling population growth that we do today, only more people would be affected.
There seems to be this idea in the air that humanity can escape any future need for population stabilization with space colonization, but aside from just delaying the problem further, that seems to ignore just how profoundly terrible it would be to live in an extraterrestrial colony in the absence of radical technological change. Mars isn't the American West- perfectly suited to human life and covered in recently abandoned farms that smallpox-resistant colonists could pretend were untamed wilderness- it's a hellhole. Glorious sci-fi dreams and the pioneering spirit would only obscure the lived experience of that for so long, and you can't spend centuries dropping asteroids on the planet for terraforming when there are people already living there.
If some future generation is forced into space- or even into the uninhabitable parts of Earth- by overpopulation, that would strike me as a profound failure of humanity.
Scott, it seems the answer is ‘yes’ to some degree based off the comments, so my apologies if I’m misinterpreting you, but do you think we all have a very high chance of dying after AGI debuts? Innocently earnest question, I know it’s been asked a million times around these parts to different people and answered in million different ways, but I’m curious what your take-it-or-leave-it answer would be today at this very moment.
"I notice it’s weird to be worried both that the future will be racked by labor shortages, and that we’ll suffer from technological unemployment and need to worry about universal basic income. You really have to choose one or the other."
I can think of a couple of ways to reconcile these worries. For one thing, you can imagine someone's concern about the increased burden of support for the elderly to be primarily about direct physical support: nursing, caregiving, etc. Even granting a large population of unemployed people interested in working, it doesn't have to follow that enough of them will be interested in working in the narrow field of personal care.
Beyond that, even if we limit the discussion to worries about the increased burden of *financially* supporting the elderly, these worries need not be incompatible with worries about technological unemployment, either:
I assume you'll agree that fears of technological unemployment, by and large, cannot be about unemployment leading to resource scarcity in the aggregate, since by definition, firing human workers to replace them with more productive machines leads to more resources in the world, not fewer. Rather, then, these fears surely must be about the distribution of these resources: the increased productivity accruing to the owners of the machines will not flow to the displaced proletariat without some sort of redistribution (e.g. the universal basic income that you alluded to).
I further assume that you will, if not fully agree yourself, grant at least the reasonability of the proposition that the responsibility of supporting the elderly is not spread equally across society, but devolves primarily to those closest to them, mainly their children. (I'll further point out that if you choose to challenge this proposition while granting the first, *you'll* be the one in danger of inconsistency, since the problem of redistributing income across society to support the elderly is surely no different in kind from the problem of redistributing income across society to support everyone in need of support. If worries about technological unemployment are predicated on pessimism about society's ability to adjust its redistribution systems fast enough to support the unemployed in general, it's surely consistent to be likewise pessimistic about its ability to do so to support the unemployed elderly in particular.)
Thus, to me, it appears quite consistent to fear that in the future, many more jobs will be performed by machine, the rich will grow richer, but others will struggle to support themselves at all, let alone their aging parents, the responsibility for whose support they will now have fewer siblings to share.
I used to be near-totally apathetic about population decline, and in fact considered it a good thing compared to the alternative of growing until we hit our planet's carrying capacity. That said, regarding #6, I was recently startled by the release of the newest census results, which revealed that Canada (where I live) is becoming a country of olds with shocking rapidity.
I am not as sanguine as you regarding the ability of technology to close the productivity gap. Everybody thinks technology can replace somebody *else's* job, but rolls their eyes at talk of a robot replacing *theirs*. Given that the tech industry has completely failed in their quest to replace long-haul truck drivers, which I thought would be low-hanging fruit for modern AI, I'm now in "believe it when I see it" mode on this subject.
Ultimately, though, you're probably correct that making labor more valuable is likelier to be a good thing than a bad one.
Beyond the problem of low-fertility states with large entitlement programs (which won't be sustainable with fewer future taxpayers,) as I've written elsewhere: "While people with kids and people without kids can all be responsible, only people with kids have a special connection to the well-being of successive generations: the double helix. And with that special connection comes a stronger incentive to act in accordance with that special concern for the well-being of future generations, generations that will include one’s own children if you have them."
So what blog post should we start at if the idea of an AI technological singularity makes us laugh because even the height of well-kept technological infrastructure is max akin to the brain of a dying Huntington's patient who has just been shot in the head with a nail gun several times
The semi-things to "worry" about is that a) the way most retirement systems are funded is with a tax on workers. If they were funded by a VAT things would look different. b) the wage taxes that have paid into US retirement fund have not been enough and so tax revenues (wage or VA) need to increase unless benefits change.
I agree that 2100 is going to be very strange, to the point where trying to worry about fertility rates then is likely to be irrelevant. At the very least, we'll probably have one or more of the following:
1. Meaningful life extension, which could range from "people can feel like they're in their thirties until their fifties" to "we solved aging and also cancer".
2. Artificial wombs for people
3. Very good nanny robots
All three of those together could really change the math on population growth - and even if it doesn't, heavy life extension is likely to lead to offworld migration to get around people in power being reluctant to step aside (and now they won't be dying off).
Underpopulation arguments are always strange to read, as a moderately smart person who wants kids but can't afford them (nevermind likely needing Singularity-class tech tree improvements to make it biologically possible, thanks sterility). Some of that is due to "suboptimal" life choices - I could have stuck with school and joined the PMC, I could have stayed in soul-destroying but better remunerated white collar work - but there's only so much changeable on my income end vs. the stupidly high local cost of living. Forget adding a kid to that, if just paying "cheap" rent is already 1/3 of wages...
("why not just move?" Because I was born here, entire family is here, and it's all I've ever known. No friends available in more affordable areas to cushion what would be a massive disruption to my hidebound routine-heavy life. Very high costs make even objectively rational moves quite difficult.)
SF wouldn't be Childless Capital of America if the relative equality of income vs. cost of living were lower. It's often stereotyped as a playground for the childlessly wealthy drifting through life, but I can assure you, in the lower deciles most of us do want kids.The ones who go ahead and try anyway, and sacrifice basically their entire net worth to try at that SES...they're way braver than I am, and I wish them well, but man. Even without (let's be humble) elitist worries about dysgenics, that's far from ideal child-raising circumstances. Maxing out whatever potential *is* there, in such children, is surely less likely when flirting with poverty and family instability. I wish we could do better. If that means taking up the same standards as the underpopulation-worriers, because no one else will advocate natalist policies - well, so be it.
I also wonder about how much hinges on *relative* population levels, no matter what the *absolute* level is...not just in the native/immigrant dimension, but between country regions, and between countries. We see the same thing play out in "income inequality", "education inequality" (that degree graph with a mere * asterisk in place of a bar for black PhDs...yeah, that's A Thing.), "racial inequality", and so on. One doesn't have to be wokely progressive to perceive the dynamics at play, politically and otherwise.
Age distribution is a good example given in the post - it doesn't matter quite so much about the absolute number of young vs old, but whether we can maintain that pyramidal distribution, since otherwise our current form of elder welfare isn't sustainable. One can easily draw parallels to US politics (House vs Senate apportionment, school district financing, gerrymandering...). And international relations too - the US:China ratio settling towards 1:2 instead of the current 1:4 or so, that will certainly have ramifications. Obviously, it's harder to predict distributional shifts vs absolute shifts, but it's an intriguing thought experiment.
Lastly, just to complete the Rationalist Singularity trifecta (AI, technological productivity, and ____) - how should one update on cryonics or other forms of Death Is The Enemy? Making new kids matters somewhat less (in either direction) if we can hang onto already-extant people operating at reasonable efficiency for longer, or possibly even bunker a "backup generation" to guard against x-risk. (Argument From Fictional Evidence: worked great in the Fallout universe!) It's certainly still within self-interest utility-maximization to not die earlier if possible, but we're talking about vast system-level effects here.
Those of us who believe in homogenous nation states (pejoratively called “nationalism”) are not really invited to discuss the rationale for this position because “racist.”
It’s ok to want Britain to be for British people. And to have an idea of what a typical British person is on several dimensions of homogeneity.
Also, not only am I not worried about underpopulation, I simply don’t want to cover the entire planet with concrete and steel.
If population declines get bad enough, I'm sure some tyrant will simply compel or pay women to provide surrogacy services for the state. It might be a way for women to raise their social credit score, reduce their jailtime, get access to universities, etc. Of course, they baby they would gestate wouldn't be genetically theirs - the sperm and egg would come from more "promising" people. They could also recruit surrogates from outside their country - for example, among women fleeing a warzone or natural disaster.
Anecdotally, many great Jewish scientists are reported to be the grandchildren of famous rabbis. This could be a place to study whether (1) it's possible to reverse any IQ depletion effects by with policies that encourage smart people to have kids, (2) whether in practice those policies are justifiable within a secular morality, and (3) whether these "inversions" have always been happening, and with other effects are actually maintaining a dynamic equilibrium. (Generally people are no more religious than their parents, so the only way for religious Jews - who don't proselytize - to keep their numbers up is to have lots of kids.)
A lot of people quote Keynes to argue that in the long run we are all dead. That may be true, but the whole point of Keynes argument was that thinking that way ignores the issues that have to be grappled with before then, like the rest of our lives. His analogy was to a weather forecaster arguing that one need not batten the hatches in a storm because the storm will eventually pass. If you've ever been in a storm at sea or even just been out on choppy water, you know this is nonsense.
Sure a singularity, let's say Jesus Christ coming back and pulling off a vastly expanded miracle of the loaves and fishes, would solve a lot of problems. That still leaves us with the problem of managing things until that happens, and let's not forget the subsequent War of the Loaves and Fishes followed by Jesus throwing up his hands and trying his luck elsewhere in the galaxy.
> (there might be an exception for fields of science that couldn’t have existed in the 19th and early 20th century - see here for more)
This is already a fiendishly difficult question to quantify, so let me go ahead and make it downright intractable: what was the rate of discovery in 1880-1920 v. 1980-2020 specifically within fields that existed in the *18th* century? I think I'm inclined to believe that an awful lot of low-hanging fruit became available either all at once or in a quick feedback loop in the mid-19th century (for reasons largely but not exclusively due to materials science and global trade), and this heavily distorts treatments of the matter than only go back a hundred years or so.
I think in modelling the speed of scientific advancement it's worth throwing a 'friction' term in there. According to the low-hanging fruit idea, science is getting harder and more complicated, and more work needs to be done by more researchers to go the same distance. But as research is done faster by more people, it gets inherently less efficient.
One reason is that the cycle time for most science projects is several years, and throwing more scientists at the problem just means you're parallel threading, not necessarily speeding things up. Another is the difficulty of communication. With so much work done so quickly it's difficult to keep up with what's going. You need to spend a lot of time reading to make sure nobody's solved this problem before, and to find the hidden gem that could catalyze a breakthrough. And that's time not spent doing research. I think there are other contributors as well but those are the most common-sense.
The primary implication is that we may be overestimating the impact of the low-hanging fruit effect, and some of the science slowdown is due to 'research friction'. If the size of the researcher pool stops increasing as quickly, then this friction will stop increasing alongside it. If we can figure out better frameworks for science, maybe we can even decrease this friction term and speed science up in a way that seems hopeless with the low-hanging fruit argument.
On point 6, I think there is a real concern with what it will mean politically if automation/health mean that retirement ages don't go up with the overall demographics, and retirees become politically dominant. There have been plenty of gerontocratic societies in history, but the "democratic gerontocracy", where people of working age are simply outvoted, will be a new phenomenon in human history. There are already hints of this in countries that have extreme age gradients in voting patterns, such as the United Kingdom with its increasingly nostalgia-based Conservative government. Even if retirees act in rational self-interest, rather than voting for the "good old days", they have peculiar policy incentives, such as being unconcerned about wages and having little interest in education, but very concerned about healthcare hypersensitive to anything that erodes their savings, such as inflation. This could even be self-reinforcing: if the entire population of childbearing age lacks political clout, the government could end up making parenthood dangerously expensive. I wouldn't worry about "dysgenic" effects so much as the lifelong disadvantages caused by bad childhood circumstances, things like poor nutrition and lack of parental attention because parents are too busy with jobs. Even if society as a whole is well off, if most children grow up poor because of lack of economic support for parents, that means a significant degradation of human capital from one generation to the next.
A related danger is the return of the kind of rentier elite that existed before the Industrial Revolution, where most wealth is inherited (and passively accumulates purchasing power through being invested in property, index funds and so on) rather than gained through labour or entrepreneurship. Low fertility rates will accelerate the concentration of this inherited wealth into fewer and fewer individuals' hands. Richer countries will be able to import workers, but those new workers' assets and wages will be paltry compared to the amassed fortunes of their landlords and the owners of the businesses they work for, and they'd be liable to worse exploitation as a result. Immigration also plays into the political battle between old and young: if there is a mass of immigrant workers who cannot vote, that effectively amplifies the power of the mostly older citizens, even if the total population pyramid isn't so skewed.
That makes sense, at least for now. I gather the Amish have been moving across the Midwest in their pursuit of good farmland. They haven't bumped into agribusiness yet, at least not seriously. I know that the Amish are pretty good farmers, but those guys running agribusiness can be nasty.
> If we don’t die of something else first, there will probably be a technological singularity before 2100. The way things are looking now, it will probably involve AI somehow. If by some miracle that doesn’t happen, we’ll get one involving human genetic engineering for intelligence. I think there’s maybe a 5-10% chance we somehow manage to miss both of those entirely, but I’m not spending too many of my brain cycles worrying about this weird sliver of probability space.
I don't mean you any disrespect but you're so extremely, unfathomably wrong about this. The chance that NONE of the above happens is, in my casual estimation, 99.99999%. Honestly that's an arbitrary number of nines, I really just wanted to write 100%. Point being: not gonna happen.
That's not what I'm interested in, though. What I'm interested in is how I can BET on that. Those folks predicting "AI" by 2029 on Metaculus... let's just say if it was a real prediction market with money, I'd bet every single penny against that consensus. But it's not! So where's the real-money version of Metaculus that will let me separate futurists from their money?
I humbly suggest that Japan as it currently exists is not completely interchangeable with a Japan made of 1/3 ethnic Japanese people from today and 2/3 ethnic Japanese people 100 years from now.
Yet we see few calls for population control because "the youth will change our culture" (which they will). Ethnicity seems like a red herring if your goal is cultural preservation (which is what I assume drives concerns about 'demographic shifts').
I'll be honest. This blog usually flies over my head and this article is no exception. But be forewarned, the line "..a cloud of microscopic death robots that used to be our solar system has expanded as far as Sirius B.", will in fact be used in my first science fiction feature film. Release date TBD. (Before 2050 of course.)
The counterargument to the "dysgenic effect" is pretty simple: it should have been a constant throughout history, and yet here we are anyway. Clearly it's not an issue. Don't @ me.
This is a little out there, but I think some of the anxiety over depopulation stems from a deeper emotional-philosophical place. Population decline is the nation- or planet-level version of finding gray hairs in your beard. It’s an uncomfortable reminder of the reality of death and the impermanence of humanity.
Population decline prompts questions about what we are really doing here on this planet. Most ACX readers probably accept some version of the idea that we are amusing ourselves while we wait for the heat death of the universe, but that’s not something people are psychologically ready for (I think this has a lot to do with the dramatic increase in depression).
Narrative religions like Christianity and Islam were - and are - immensely powerful because they placed both individuals and humanity in a larger sequence of events that was building towards something. Even today believing strongly in certain religions is a strong indicator that people will want children. Some of that may be social pressure but religion also gives you a sense that you are part of something that will continue - it gives some purpose to having children.
It’s the same as the feeling you get when you’re in a movement that seems to be growing. But when the movement begins to decline and shrink, you feel a crisis of purpose. The threat of a declining world population or national population is giving people that same feeling: is humanity a sinking ship? And the answer is probably: yes, but don’t worry about it. Time passes and change is eternal.
How long will it take for our mesa optimizers to update so birth rates in the developed world recover?
Is western culture, which has been exported throughout the developed world, unfit for people? If so, is this unfitness intrinsic, or can it slowly be mutated away through either immigrants with different values or the above mesa optimizer update? What are the parallels to the urban penalty?
If it is intrinsic, can a different culture take its place as the world dominant culture, or will we be stuck with a cultural black hole that sucks in people and slowly kills their lineage?
> Workers will be able to expect high salaries and good working conditions.
This seem to me like moving your eyes off the ball, thinking about money instead of consumption. You can't eat high salaries, and what you can eat is more expensive if the work that was put into it was more expensive. You certainly can't eat hight salary that the government redistributed do that your old neighbour would eat too, or that you (a single child let's say) choose yourself to use for your parents medications.
Just a minor point, but my understanding is that at least in Nordic countries, fertility levels by educational level are now such that the highly educated are the most fertile group, and in 2020, the fertility of the highly educated actually increased while it decreased in other groups. See eg https://www.stat.fi/til/synt/2020/02/synt_2020_02_2021-12-03_tie_001_en.html for Finland.
Point 7, dysgenics: How about Africa? No one knows whether Africans' low average results on IQ tests are caused by genetic factors. But do we know for sure that they are not at all caused by genetic factors? If they are caused by genetic factors to any degree, altruists have a huge problem.
The proportion of Africans to non-Africans is supposed to alter dramatically in the next few decades. If Africa doesn't become self-sufficient in food and medicine in the next few decades, our children and grandchildren will have to be much more altruistic than we are: If nothing changes, a dwindling number of Westerners will have to provide for twice as many Africans.
Sure, things can change. But the null hypothesis, that things continue just as they are, should at least be considered as something that could happen.
About why science is slowing: Maybe I am biased by my field, but I do not think that the problem is low hanging fruits.I am a BSc, and I invent all the time intersting publishable things. I know they are publishable, because I almost always find that they were published in the last few years. The problem is that the constant flow of low hanging fruits is simply not wide enough to feed that many scientists. Some scientific work is incremental, and trying to give it ten times as much workers would just mean that five will be very busy thinking basically the same thoughts, three would find some intellectual desserts they can specialise in, and the other two will try to do something that is impossible now and will be very easy in five years, because at least if they succeed it would have some value.
I agree with most of your post above, but I think there is something additional and interesting to be said with regards to point number 6 specifically, which is relatively easy to miss if you have an American-centric perspective only.
And the point is that meaningful changes in the ratio of working adults to retirees matter not just economically, but also politically. Older people simply have different concerns and goals they want to get out of the political process, and many of them are short-term and fundamentally incompatible with the type of medium- or long-term thinking that is required to solve systemic issues that hold countries back.
I'm not sure how relevant this is for the US specifically, but as an example of what I mean, in many European countries (and especially Eastern Europe) pensions for retirees are primarily public, in that they are provided directly by the government to most people over the standard retirement age. My personal experience living in these countries (which is also confirmed by most surveys) is that for these people, the issue they care most about when voting is ensuring that those pensions stay as high as possible, and everything else is secondary to that. Macron's recent underperformance in the French presidential election, as another example, has also been attributed to his promise to reform the pension system (through raising the retirement age), which angered many older people who would rely on it as a primary source of income in their latter years.
Scientific consensus is clear that old people vote at far higher rates than younger people do, and that they are less concerned about issues such as climate change (which is understandable, since they'll be dead long before this would really materialize). In these European countries I mentioned above, many of them are truly single-issue voters who incentivize governments to spend a lot of their budget on pensions, leaving far less for everything else. Systematic improvements to the educational system, environmental concerns, anti-corruption fights (especially in Eastern Europe) etc all end up falling by the wayside, even though they are incredibly important for the long-term development of a country, because appeal for them comes almost solely from young people, who are already less likely to vote and are becoming a smaller percentage of the population every year.
I basically endorse all this except I am slightly less "optimistic" about getting a singularity soon. How about a bet where I pay you one bitcoin now, and you pay me two bitcoins in 2037 if there isn't any artificial superintelligence yet?
Re 8. I am not convinced by the premise that increasing the number of researchers we should expect to see a proportional increase in the rate of scientific discovery. Even assuming an infinite supply of low-hanging fruits, I would naturally expect diminishing returns.
In science it is not at all uncommon to find out that at any given moment there are a bunch of other groups that are working on the same thing as you. Scientific ideas are discovered when it's the time to discover them, when enough background has been accumulated that you can have a certain idea - and you are not the only one that will have that idea at roughly that same time. So I expect that just increasing the number of researchers is not going to give you a proportional increase in the rate of scientific discovery, just increasing the number of people working on the same thing with only a marginal speedup in discoveries.
If my model is correct instead, there shouldn't be any significant reduction in scientific progress with not-increasing (or diminishing) population numbers.
OK, we're all going to be paperclipped in 2053, but I'm surprised that even here in the craziest most Parfitpilled corners of the internet Scott doesn't feel the need to include as one of his eight fake worries about underpopulation the argument that, all else equal, it's better for there to be more people than fewer.
If the Singularity is called off and the world asymptotes to a steady-state with 10 billion people that persists for a million years... isn't that world much worse than a world in which 20 billion people get to live for a million years? Maybe not twice as good, and there's lots of arguments against this position - you might consider and discard the basic idea that being alive is good but not even to acknowledge it seems weird to me.
I was surprised by the Icelandic IQ decline research, so I've read through it, and it only indicates a projection of a decline of the mean IQ score in Iceland, they don't actually measure it. A big difference IMO.
I'm trying to write a sci dystopia and I'm having trouble imagining how things can go as wrong as I want them to go for story purposes, at least without the society completely collapsing due to war or something.
'When fertility is negatively correlated with a socially valued trait, it is said to be dysgenic (the opposite of eugenic). So while fertility within populations may be slightly dysgenic, fertility across populations is strongly dysgenic. Average cognitive ability in the world is going down.'
Some technologies have a minimum population requirement. Galileo Galilei only had access to great lenses to experiment with, because lots of people bought them to make glass windows. In the preceding centuries cities were smaller, markets more fragmented and cost of transportation demanded a higher price. In general, technological progress is accumulation of steps of work in single artefacts and market size is among the conditions of their existence.
A smaller city would not sustain a ship crew sized masonry to keep its portrait galleries up to date. Rome at its peak did.
A likely candidate for something prohibited by smaller population is use of rare elements. Will we continue to filter tons of granite for some ppm of erbium if we have less miners in the future and their wages rise? Or will we find "easier" ways and just return to copper?
Hmmm. So I don’t think these are necessarily “straw men” that you knock down, but I don’t think you really spend much time considering the arguments in favor of “aging demographics are absolutely awful” that I, personally, find to be most compelling.
I.e., 1) the reality of unwanted childlessness among women and the attendant moral and humanitarian consequences of this; 2.) that this has literally never happened before in human history and we have no idea wtf happens when our population structure inverts; 3.) you reference but don’t spend much time at all engaging with the likely economic effects of having almost no young people and so many old people — we have no idea whether the civilization we have built in the West is even sustainable under those terms. This isn’t just about “fewer people” it’s about demographic composition.
There’s also an underlying moral framework I have that you seem to lack — which is that two people who are living and existing and happy is better than one. And that having less human beings who could be living and existing and being happy, when you could have more, is inherently bad.
Also “2100 is so far away why worry” seems completely ridiculous to me. That’s seems to me to be an argument against literally everything, and also undercuts your (near) certainty that technological change will render demographic change meaningless.
"I understand why people don’t want to talk about the issue this way, because if you say demographic shift is a problem, people will call you a racist conspiracy theorist. I don’t think it’s racist to care about ethnic demographic shift - I think Japan as it currently exists is not completely interchangeable with a Japan made of 1/3 ethnic Japanese people and 2/3 ethnic Kenyans."
I think it is, if not racist, then at least extremely pointless to worry about demographic shift. What is the utility in having e.g. a US that is 60% white vs one that is 30% white? I struggle to think of non-xenophobic reasons for preferring the former.
I wrote something similar a while ago from a mostly different perspective: traits that cause people to want to have children in an environment that contains birth control seem to be heritable and so we should expect them to be selected for. That's mostly Agreeableness in Big 5 terms but probably it's more complicated than that. I also thought the technological landscape changing would render the issue moot, anyways.
Scientists in future will be what farmers used to be.
Imagine a world where we completely recycle everything we produce and consume. In such a world you couldn't simply invent a new product and start selling it. You would need to provide the whole closed loop process. How it will interact with the environment, how the product will be recycles etc. Studies after studies will be required before the regulators will allow it and then the post-marketing surveillance and more studies to confirm that recycling is being done efficiently.
Being a scientist is a very boring job. I don't know why people think romantically about it. It is an important job, for sure. Just like being a farmer is important to ensure we don't starve.
I never thought I would imagine a comic book series about the Amish and the Orthodox fighting for control of the world (and pacifism doesn't rule out some sort of struggle!)
Has anyone ever noticed that economic growth correlates exactly with population growth? That economic growth can be artificially stimulated by any measure that increases population, including the immigrants for whom western economies will soon be competing?
Think of a bacterial population in a petri dish. Their population growth curve is precisely like that of the world's human population, and reproduction is automatic with only one factor controlling it: the amount of nutrient left in the agar. Our curve is the same shape, but it's not (yet) natural resources or food that controls its slope, it is economic factors in charge. And as far as I can tell, there is a bi-directional dependence. Increase wealth (eg discover North Sea oil) and the population curve is stimulated. Conversely, increase the population, say with a flood of immigrants, and economic activity rises. And if grow or die are our only choices, I can't see how there will be any kind of future stable state.
There is a lot of speculation that China's official population statistics are inflated and its population has been falling for decades. The overcount may be on the scale of a Japan (~120M) which would put China's population at 1.28B. There seems to be tacit admission to this by the CCP based on this being reported in the SCMP.
> I don’t think it’s racist to care about ethnic demographic shift - I think Japan as it currently exists is not completely interchangeable with a Japan made of 1/3 ethnic Japanese people and 2/3 ethnic Kenyans.
Even conceding this much is overlooking an important point (one that racists would very much like you to overlook). The Japan of 2020 is not interchangeable with the hypothetical Japan of 2120 where 66% of its population has Kenyan ancestry. But! Neither is it interchangeable with the hypothetical Japan of 2120 where the population is 100% ethnic Japanese!
Both of these Japans are hypothetical and exist far in the future, after a hundred years of technological and economic change. Ethnicity is probably the *least* interesting change that will happen in 100 years. What's a more significant difference between America of today and America of 1920 - the fact that we have atomic bombs or the fact that we have more black people?
Racists want to frame it this way because it invites you to compare the Japan of today and the Kenya of today and ask which one you prefer. But what the question is *actually* asking is if you prefer the Japan that accepts Kenyan immigrants for 100 years or the Japan that doesn't, and when you look at it that way you might notice that none of the other migration waves that were supposed to destroy our country actually did that. Immigrants assimilate, cultural conflicts of the day get paved over by the unstoppable forces of economics, and a hundred years later, the only record of the Great Kenyan Migration is that the stereotypical generic anime protagonist now has dark skin.
"After all, consider the century 1820 - 1920. It gave us the steamship, the railroad, the automobile, the factory, mass production, electricity, refrigeration, radio, the airplane, etc, etc, etc, with a population only about 10 - 20% as high as today. The effective innovating population - the number of educated people living in countries on the technological frontier - was probably an even smaller proportion. About half of these innovations came from Britain, a country with about 0.3% the current world population."
Excellent point. Today, it's much, much easier for smart people across the world to put their intelligence to use developing new science or technology thanks to better university systems, cheap international travel, and liberal immigration laws in most innovative countries. During the 1820 - 1920 timeframe, how many Chinese geniuses were stuck on family farms for their entire lives because there was simply to route to higher education?
Even if the overall population trend is dysgenic, we are getting better at leveraging the talents of the smartest people.
On #7, shouldn't we expect a long-run offset from increasing access to education? In the US there may not be as much space for gains in that regard. Access is pretty broad from what I can tell - OECD numbers have the US around 92% high school completion and 50% for post-secondary degrees. (https://nces.ed.gov/programs/coe/indicator/cac/intl-ed-attainment).
Other OECD countries like Turkey (13%), Austria (34%), Italy (20%), Mexico (19%), and Germany (31%) presumably have room to keep growing their rates of educational attainment as well. Even if dysgenic effects are really happening at a rate of .03 IQ points per decade, it seems like any slowing effect that might have on technological advancement would be more than offset by the increasing numbers of educated people in the mix. I'd rather have 2 researchers working on a problem than one researcher who has an IQ .03 points higher.
Fairly material change in that it projects population will now peak in 2086 rather than post 2100 (both lower growth rates and more death thanks to covid etc).
Also, there is quite a lot of noise that China has been misrepresenting its population and it's actually 180 million less than stated, and most importantly in the key segment of the fertile part of the population (with a theory that local authorities inflated their schoolchildren numbers in order to get more funding from the centre). e.g. https://www.project-syndicate.org/commentary/china-2020-census-inflates-population-figures-downplays-demographic-challenge-by-yi-fuxian-2021-08 So if true, China's population is both a lower starting point and will drop more quickly too.
All in all, this makes me update with less confidence of these projections in general. They seem to be a bit like the projections of solar price points in how they are always wrong in a single direction; and the potentially different China picture would be another major update in the lower direction. Given that much of the 'it will be OK' is coming from projections that Africa will still have huge growth rates, and how quickly other countries have seen their growth rates collapse, how much faith should we have that that will actually be the case?
There's an interesting Amish custom which is practiced in every district I've known or heard of (This is something like three districts in rural Indiana from friends and acquaintances as a teen, and another half dozen from talking to people who moved here and there.) It's called Rumspringa (which roughly translates to 'jumping around.')
See, the various restrictions that the Amish church imposes on its members only apply to members of the church and their children (where 'childhood' is held to end at 16). But Amish youth aren't expected to join the church until they're ready and willing; the whole process of joining is a bit of an ordeal to ensure commitment. You have to sit in an un-weather-controlled booth (typically in 90+ degree weather) in isolation all day for a week or more studying the scripture and memorizing key doctrinal statements by heart. You have to be discipled by a member in good standing with repeated meetings over a month or more. Your application has to be approved by your father, your mentor, and the local bishop (exact set of approvers varies based on district.)
And so far so good. If you were a clever fellow designing a church system to use human biases to keep them in the church, you might well make entering the church hard. That gives people a sunk cost to joining, which should be pretty effective at making people reluctant to leave. Combine that with various other Amish practices:
* Officially Shunning those who stray. No local Amish is allowed to speak with them or deal with them.
* Using a primary language other than English, so you're marked with an accent to the rest of the world.
* Only mandating honest dealings with other Amish, so your favorable community immediately turns hostile if you leave.
* (at least in the local district where I grew up) Kidnapping anyone who questions the church too much and taking them to a secular mental institution for brief involuntary commitment and evaluation.
And you see a system REALLY set up to retain people. But if I were designing this system and trying to get as good of Amish retention as possible, I'd probably strongly encourage kids to join the church at 16. They're now technically adults, but they're still under the roof of their parents, dependent on Amish friends and family for support, etc. It should generally be possible to apply more pressure to get them in to the church, and you've already made it so hard to exit.
The Amish don't do this. At 16 you're an adult, and you aren't expected to join the church till you're ready (usually 20-25.) That intervening 4-10 year period is called Rumspringa. In it the youth are EXPECTED to try out the world. The laws of the church don't apply, and noone acts like they should in any way. Amish kids of this age in the area I grew up usually worked at local RV factories during this time period, which is hot, high-intensity, 12 hour days for 4 days a week, and pay very well for low-skill labor. Lots of overtime, lots of non-Amish drop out because they don't have 10 years of getting accustomed to long hard days.
So, what do Amish kids do with $50k/year, working 4 days/week, with no expenses and no church? They live it up! Bars, nightclubs, sports cars, new big trucks (more popular than sports cars as a status symbol in Red America), hot girls/guys, showy clothes, etc. They spend years indulging in the 'vices' that they were forbidden as kids, and will be forbidden if they ever come back to the church.
And then they come back. Not all of them for sure, but most. I'm not entirely sure WHY; as someone who grew up English and only had some Amish friends I never felt like I fully understood why so many come back. Those who leave without ever joining the church are only subject to the usual social pressures, not the awful institutional shunning that applies to leaving church members; it's not THAT hard to leave. Some come back because they fall in love with another Amish who's more devoted than them. Some because they want to be part of the Amish community.
It's kind of puzzling to me. I wouldn't expect 'encourage your impressionable youth to indulge in all the pleasures of the flesh' to be an effective retention strategy for an ascetic cult, but it seems to work.
On the point about "Age Pyramid Concerns Are Real, But Not Compatible With Technological Unemployment Concerns"- actually, no, those can both be problems at the same time, because not all workers are interchangeable carbon units that can be shuffled between industries at zero cost. The people who are rendered unemployable by shifts in AI technology are not necessarily going to have the skill-set or temperament to look after an increasing population of the elderly (especially at the higher end of the healthcare industry), and the people who are not being born as a consequence of collapsing fertility can't necessarily be replaced by robots unless you have full-blown AGI (which will create it's own problems, not least of which is that robots don't provide consumption.)
Also, I hate to be a cracked record on this point BUT SERIOUSLY SCOTT YOU NEED TO GO READ PETER ZEIHAN. The points above actually grossly underestimate the seriousness of fertility collapse because (A) currently prevailing international patterns of trade relating to import/consumption and export/production are dependent on demographic structures which are now going away, which would lead to mass recession even if the welfare state wasn't being bankrupted, (B) fertility collapse has been occurring for decades in many countries that do not have the wealth and expertise needed for mass automation, such as eastern europe and China, and (C) political anxieties triggered by A and B in combination with US military withdrawal from overseas are causing breakdowns in global security that trigger wars and further disrupt essential supply chains. Russia invading Ukraine (which is likely to cause a global famine) is simply the first big example.
(I would also say that dysgenic fertility is probably more serious than it looks, partly because we don't have a good handle on the additional impact of deleterious mutations but also because the effects are most concentrated in the younger age groups on which we depend for labour and innovation, and there's been a couple of other direct genetic studies on the subject from Ohio and the UK Biobank data. But anyway, yeah, not the most pressing immediate crisis.)
Something that can offset the decline is the possibility that technologies that will extend human lifespan a lot will be developed before the end of the century.
20 years ago, if a scientist would have said that aging is a disease and it could be possible to cure it, it would have meant the end of his career. Today, it is almost the mainstream view of the scientific community.
I believe it is possible that the first human who will live 200 years is already alive. If it’s true, having less children will not be so problematic for the countries suffering from a low fertility rate.
Another terrifying possibility is that technologies that allow the conception and growth of babies without the need of women are developed – and then used by totalitarian states like China to artificially create millions of citizens, raised and educated by the state. Parents need not apply.
Some people are already contemplating this. The technological possibility doesn’t seem that far off :
I think underpop as proxy for demographic fears is correct. But also, what about the David Goldman thesis, that younger countries, provided they're rich enough to afford good tech, are militarily dominant, and ageing ones definitely arent?
Does an paper like "Nanosecond protonic programmable resistors for analog deep learning" (https://www.science.org/doi/10.1126/science.abp8064) change anyone's priors on the timing of the AGI apocalypse? The MIT group are claiming about a factor of 1e4 to1e6 improvement in some types machine learning.
If you think there is a 10% chance we don't have a singularity - would you play Russian roulette with those odds? Given how bad a long running civilizational decline would be, this seems worth worrying about.
And even if you think we're going to be able to genetically engineer superbabies with astronomical IQ, will they compose a substantial fraction of the population? Will they actually occupy positions of political or cultural power, rather than being shut out by the mob? Are you so sure governments won't make genetically engineering humans illegal?
One point to note, that I don't think is addressed here, is that these projections of population growth all assume "business as usual" with our fossil-fuelled civilization. They assume no nuclear war, no severe pandemics, and no breakdown of our increasingly fragile and interconnected global economic system as the impacts of climate change intensify (best case on current projections is that there will be massive migration by 2050 caused by famines, droughts and flooding - and we are not on track for a best case scenario in terms of emissions profiles, quite the opposite). Assuming that we will stay on the rails for another 50 - or even 100 - years is dubious.
If AGI happens within 15-20 years, which seems eminently possible, then any concerns about innovation flies out the window and a bigger issue about humanity's survival will take center-stage.
Ironically, a slower innovation pace due to aging + fewer "low-hanging fruit" could paradoxically be good for humanity as it elongates every innovation cycle and perhaps even makes AGI impossible.
Whatever the downsides of that world is surely better than extinction.
Respectfully, this is not quite correct. Within a couple generations of the actual immigrants, ethnic identity is quite diluted, in large part because there's a significant degree of intermarriage between immigrant groups, and in part because you start to lose the original language spoken at home.
1. Demographic shifts. The people who are unspeakable in this context are right to be worried about this. *Specifically* they are right to be worried on a racial/ethnic sense because when it comes to "overpopulation" or "I'm not having kids about the climate" or "There's nothing worse than the undisturbed sleep of a white man under the patriarchy" , it isn't sub saharan africans or redneck American conservatives that think these things, it's Nice, Affluent, White, Liberals. Everybody in every culture implicitly cares about the preservation and continuation of their culture, even as it adapts and changes. Similarly, if you're a fan of Japanese culture in any respect - you *should* want them to find a way to remain fundamentally Japanese. Presently, the only groups who are allowed to speak these worries aloud are white affluent liberals afraid that new tech workers are going to push out the Hispanic color of San Francisco. The reason to be worried is not that you fundamentally do not want immigrants or particularly hate them, the reason to be worried is that your ethnic group and culture seems *uniquely* obligated to not care about these things. It's ok to care a little bit.
2. Similarly, almost all of the comments about Britain in the 1800s apply equally to the United States today. If you're not worried about overpopulation because there will be X many researchers due to population growth in other demographics, you need to be pretty darn convinced that they will adopt a similar or better focus on research than the cultures currently at the frontier, or be inculcated into those cultures - given 1, why are you convinced that that will be the case?
3. Great powers (and conflict between them) are effectively the only places where we see these explosions of technological, moral, social, innovation. There's exactly one place looking prominent to become a great power, partially due to its large population. Allowing the CCP to take the reigns of humanity's future is equivalent to creating a suffering-maximizing AI, so we should be wanting higher birthrates in basically every country not CCP-aligned by default (even though China is going to have its own demographic period problem).
"I guess it’s still true that if innovation is destined to be only 10% of its current level in 2100, then a 30% population decline could lower that to 7%. I find it hard to worry about such a small difference, but maybe that’s a flaw in me and not in the territory."
I think this difference is worth worrying about. Not because of certain catastrophe, but because of the difference each percentage point in the possible futures we may inhabit.
My premises are that innovation helps the poor and rich, innovation saves lives, and innovation has creates compound returns over time for poor, rich, and lives saved/well-lived. And that growth creates optimism and forestalls human inflicted social upheaval and nastiness.
Imagine a small spur in innovation caused by useful but not super-strong AI, human are still doing a lot of the innovating over the first 50 years, the low-hanging fruit gets plucked, innovation starts to slow again, and there is still 3% less innovation than there otherwise would be in 2100. That's still terrible from the counterfactual perspective. 2100 might be around 50 years behind where it could be. That's not catastrophic. But 1970 at the global level was far, far, far more awful than it is today.
In that scenario it matters materially.
In the strong AI scenarios... Damned if I know.
But I am pretty sure that we should build now as we mean to continue and that includes families and trying to bring the positive value of happy, healthy people into the world. And so, the real catastrophe aren't the 8 pseudo-concerns, but that life is good. Pass it on.
Scott points out that Nigeria is the country to watch during the next 50-100 years. True, Nigeria is the most populous country in Africa, and is poised for a massive population increase; but the other African countries are also worth noticing. Here is a UN demographic forecast for 2100: The twenty most populous countries in the world. 11 are on the African continent:
India 1,516
China 1,021
Nigeria 794
USA 447
DR of Congo 379
Pakistan 352
Indonesia 306
Tanzania 304
Ethiopia 250
Uganda 214
Egypt 199
Niger 192
Brazil 190
Bangladesh 174
Angola 172
Iraq 156
Mexico 151
Kenya 142
Sudan 139
Source: UN World Population Prospects: The 2017 Revision, Key Findings and Advance
Tables. Working Paper No. ESA/P/WP/248.
Side note: In 2015, the 20 most populous countries contained 70 percent of total world population. Average population size in 2015 (194 countries) was 33 million. But median population size was only 7 million. Thus most countries on Earth have had, and will continue to have, small populations. If we can further assume that average incomes will continue to rise in the Global South, the above 20 are likely to be the most powerful states in the 22nd century.
Final reflection: God have mercy on the populations in the many small and ageing states, in today’s world as well as in tomorrow’s.
I haven't been contributing much to the comments since the move from SSC, and I hate to jump in with probably-ignorant criticism, but: the Zero Population Growth people had charts too. I saw them when they came to my high school like nine times in the eighties and nineties. And until yesterday I was told by everyone that the United States was unique among wealthy nations in that our birth rates remained so high; everyone seems to have been caught by surprise when it turns out they're low.
I guess my question is: how and why did everyone get it so wrong for so long, and what has changed to make everyone so certain that the current take is accurate? Maybe we're just bad at measuring population numbers and graphing growth projections. I remember the ZPG people being particularly concerned about explosive growth in India, China and the United States in the 21st Century. Now, whoops, everyone sees those populations declining and the population explosion is coming from sub-Saharan Africa. What happened? Why are these numbers better than the numbers thrown around thirty years ago? What makes these projections useful in a way that the last generation's were not?
Tim Dyson at London School of Economics has written an excellent book on the social, psychological and economic consequences of the ongoing global demographic transition (Population and Development, 2010). Well worth an ACX review. Some arguments from the chapter on likely psychological and social consequences:
• The vertical bonds between parents and children grow stronger
• The «planned life course» becomes a more widespread way of thinking of one’s life
• The life course of women becomes more similar to the life course of men
• Death becomes something remote & distant in most peoples’ lives
• Parents invest more in the lives of each child & time spent in education increases everywhere.
Why are fertility rates falling everywhere? It is useful to employ the trichotomy between distal (“far from”), intermediate, and proximate ("close to”) causal factors assumed to influence the number of children born per woman, and (more specifically) to explain the worldwide trend toward lower fertility.
Distal factor:
• Global mortality decline
Intermediate factors:
• Increased predictability of individual life courses - cognitive shift to a «planned» life course
• Shift from agriculture to industry-and-services economy
• Urbanization
• Stronger states, and corresponding effective legal guarantees against interpersonal violence
• Social structure opens up to allow for skill-based upward mobility
• Mandatory education & higher percentage of young people in higher education
• Women gain access to wage labor and to career paths
• Child labor made illegal and effectively enforced
• Introduction of broad-based formal social security systems
Proximate factors:
• Delayed birth of first child & delayed marriages
• Effective contraception made available: In particular, contraception that can be administered by the woman alone, and independent of coitus
Overall Scott outlines much relevant information but somehow lost in all the discussion is the biology of reproduction and the genetic evolution of Homo Sapiens
The world population increased from 1 billion in 1800 to 7.9 billion today and is headed to 11 Billion according to the UN Median Forecast.
At some point Homo Sapien population will come into equilibrium with the earths resources.
Such an interesting article. I have been looking for commentary on problems similar to those discussed above. I‘m glad I stumbled across your Substack!
Maybe I'm a bit too mundane in my thinking but it seems to me the biggest concern, certainly in the short term, is how the welfare state can maintain things with the coming demographic shift. The modern welfare state only works because the young (and in some cases the yet born) are paying for the old's benefits. That adds up right now because there are currently a lot more young than old. When that pool shrinks where is the money going to come from? Just trimming benefits or pushing the retirement age back a few years isn't going to be enough. Most developed countries are running significant deficits as it is. I don't see how this wouldn't cause a massive shift in society and who is responsible for what.
Well, everyone that want to know what happens when there’s dramatic demographic shifts just need to look at Israel in the coming decades. I’m going to present some data that I’m getting from a 2017 blog post made by the Treasure Teen, he’s an Israeli that worked for a few years in the treasure ministry. I once approached him about translating those blog posts and got his blessing, but never ended up doing it. Is anyone interested in this? Might translate them if so.
Anyway, to the data he presents:
In 2015 180,000 babies were born in Israel. 100,000 of them were non hasidic jews (55%), 40,000 hasidic jews (23%) and 40,000 arab israelis (23%). According to Israel Central Bureau of Statistics by 2040, in the age group 0 to 15 non hasidic jews will make only 45%.
My own aside: from what I can tell his using data provided by the Israel Central Bureau of Statistics, the actual official agency in Israel for statistics, so rest assured his not some crank. To give you some background about current israeli demographic, in 2022 74.7% percent were jewish, 21.2% percent were arab israelis and 5% were “others”, mostly immigrants from the former USSR that aren’t technically jewish, the non jewish family of USSR immigrants and such like.
Here’s a survey of the self definition of 20 years old and up jews in 2020:
43.1% secular
21.1% traditionalist/not so religious
12.3% traditionalist-religous
11.3% religious
10.1% hasidic
Honestly I don’t know what’s the different between traditionalist and religious, but the important thing is that in 2020 self definition survey the ratio is 9 non hasidic jews to 1 hasidic jews. In the 2015 birth survey the ration is 10 non hasidic jews to 4 hasidic jews. Obviously not everyone that is born to a hasidic family stays hasidic, but to maintain current jewish demographic MOST of the hasidic babies already born will need to turn non-hasidic. I don’t have good data on how many hasidic people stop being hasidic, but I have a feeling it’s not that much.
The problem of course have to do with different rates of tax paying, and grants from the government. According to a 2011 survey, non hasidic jews paid 2,955 shekels a month through various taxes, while hasidic jews paid 705 a month. As for using grant money from the government, non hasidic jews used 1,964 shekels, and hasidic jews used 3,256 shekels a month. But the real money is in services that the government provides to children, mainly health care and education. It’s harder to get exact data on that, but if anyone is interested I’ll try to fish something out.
One less piece of data: in 2014 birth rate of hasidic jews were about 7, religious women about 4, and the rest of the jewish population hovers between 3 and 2, secular women having the lowest birth rate at about 2. I’m getting this data by looking at a graph that was presented to the knesset by it’s research facility, that’s why the numbers are so pretty.
This is getting long and I’m kind of in a rush, but ask away if you want more data and I’ll try to find it for you. If anyone is interested I might translate some economic writing I like about the subject.
A doubt about any sort of ultra-religious community somehow breeding their way into becoming the majority: I have the impression both the Amish and Ultra-Orthodox are somewhat parasitic on the rest of society. If they grow by millions and millions as projected I suspect they'll undergo major schisms, changes, et cetera. Growing a society is never that simple.
I've read long ago an article about a model that describes dynamics of means of production, consumption, and population. Basically, humans create artifacts which in turn increase carrying capacity of their environment which in turn increase population which again creates artifacts to increase carrying capacity . However each element of the system also use (consume) other elements. E.g. humans have to expend their time on creating and maintaining artificial environment and that slows down their reproduction. It's an extremely interesting framework to discuss population dynamics and economic growth.
Scott, I believe I can put your worry of long-term intelligence decline at rest.
You write:
“It’s a dysgenic argument where we assume at any given time the people with higher degrees have on average higher genetic intelligence levels. If they’re reproducing less, the genetic intelligence level of the population will decrease. There is some debate in the scientific community about whether this is happening, but as far as I can tell the people who claim it isn’t have no good refutation for the common sense argument that it has to be. The people who claim that it is make more sense, and have measured the effect in Iceland, an isolated population that it’s easy to measure genetic effects in.»
…The studies you refer, make the point that people with low education sire more children than people with high education, for the birth cohorts under study (I wish I could put the last part of that sentence in italics, but apparently that is not possible in the comments).
Here is the thing: The birth cohorts in these (and most other studies that have investigated the recent relationship between education and fertility ) capture the educational/social gradient in fertility for some very particular birth cohorts; namely the birth cohorts that were in the fertile part of their life cycle in the middle of the demographic transition.
…The point is that the demographic transition is a hierarchical diffusion process. Fertility first falls among high-status urban groups (early adopters). The last to adopt low fertility are low-status rural groups (late adopters). The pattern is the same everywhere it has been studied. (Digression: This educational gradient in fertility for countries that have not yet reached the end of the demographic transition, is the father of the saying “The rich get richer, while the poor get children”.)
The point you overlook, is that this is a transition effect. Once the transition is brought to an end, and all social groups have adopted low fertility, there is no longer any educational gradient in favor of those with low education.
…The Scandinavian countries are examples. There, we now (admittedly quite recently) see that low-educated males in particular turn out childless to a much larger extent than high-educated males. Males with high education also benefit more, fertility-wise, from serial monogamy (which is the functional equivalent to the practice of rich/successful men having more wives than poor/unsuccessful men, in countries practicing legal polygamy).
The US and Iceland are both a bit “behind the curve” when it comes to getting to the end of the demographic transition, compared to most European countries. Also, there is an interaction effect between high status and white skin in the US, which adds to the furor. But relax, Scott: Once you get to the end of the transition, high education will start to pay a fertility dividend again also in the US.
In conclusion, to channel the Critical Drinker’s standard quip in his Youtube movie reviews: “Nah, it’ll be fine”.
More than agree with all your arguments showing underpopulation is not an x-risk by any means. Not to rehash your points, but let me add a thought experiment. Let's go kind of extreme, and imagine world population shrinks all the way to 1B. Pretty drastic considering we're now nearing 8B, or at least well over 7B.
What would happen then? Obviously it's a huge change in numbers, so many things would change qualitatively along the way. We'd be in for a surprising world with lots of descendents of Nigerians, Indians and Amish, and with a lopsided demographic pyramid. What's wrong with that? Let's explore further:
Would this bring material or economic collapse? I can't see any reason why it would - since the Industrial Revolution we've learned to be hugely productive, and there's no particular reason why we would stop being so. It just doesn't take that many people's hard work anymore to provide for every 1000 humans.
Would the human spirit stagnate? Again, no reason for that - cultural renewal has never stopped in human history, why would it then? When this world had 1B people creativity was at its highest - not just industry, but also in music and the arts.
Would the downward trend be unstoppable? Yet again, no reason for that. Concerns over underpopulation are barely a few years old, and (some) people are already taking them seriously. Cultures change unpredictably in a space of decades, and I don't think there's any reasonable scenario where we get down to 1B in less than 150-200 years. Plenty of time for various groups of humans to decide to put a cultural and economic premium on breeding if they feel like it's time to stop decreasing.
Obviously I'm being as big-picture as it gets here. Would a change of this magnitude bring along unpleasant changes and instability along the way? Probably too, but then again, unpleasant changes and instability are a constant of human history, and I don't see any overriding reason why those would be worse than the ones we would just get by continuing to do the human thing with a growing population.
You know what would happen though? The sheer impact of human activity on planet Earth would be much more manageable... and this might as well be our best insurance towards long term survival.
Disagree on point 6. Not all employment is equal. It seems plausible at least that there could be a labor shortage in elder care and technological unemployment in other fields at the same time. An elder care labor shortage would happen if there are more old people, young people aren't eager to join elder care, and elder care is hard to automate. All these seem likely to me. At the same time other fields that young people are eager to join may be automated more rapidly, leading to a difficulty in finding jobs that they are willing to do.
I get pretty frustrated when people point at that Iceland study as convincing evidence. I won't deny that they do the work in the stats, but they omit all the relevant sociology and historical variables!
Let's first start with the basic flaw - using years of educational attainment as a proxy for intelligence. Ceterus paribus, that might hold. But jeepers, are other things not equal when it comes to education. I bet you'd find a strong signal in years-of-education vs birth longitude in Americans, but you'd be nuts to think it was causal.
What do you notice? Oh - the effect sizes go to basically zero for the 20year category for men (e.g doctors). But there is still a 10x stronger effect for women. So we have genetic markers that identify women that are much more likely to reach 20-years of schooling, but they have zero impact on men. So I'm supposed to believe we have a gene cluster that makes your smarter, but it stops working when men apply to med-school. Ok.
Woah - most women don't finish more that 10 years of schooling total until WWII! And look at that scale on the left - those are individual counts not 000's. So we have (eyeballing) maybe 10 individuals in the 20-years-education category in the 1910's. All the best studies have their strongest effects in the n=10 bucket.
Hmm. Iceland didn't even have a university until 1911. So how did they even 10 women in Iceland have 20 years of education in the 1910s? What else was going on in pre-war Iceland?
So we have Iceland, the poorest country in Europe, which didn't normally educate women past 10th grade. It was a remote outpost of the Norwegian and then the Danish crowns, with a rich class structure inherited from nobles, land-owners, clergy, merchants, and on down through tenant farmers and thralls. And the researchers have found a genetic marker for the 10 women in 1915 with 20 years of school. This is news? I expect every one of them was an aristocrat, Danish, or kin.
Onto the "dysgenic" decline. Why do these numbers change?
Iceland was a Danish colony, and Iceland broke ties during WWII. Some of the the Danes left, and the ones who stayed mixed in. The allies arrive with a garrison of 30 thousand people against a local population of only 125 thousand total. OMG - 10 American GIs for every 5 young local women. Talk about dysgenic -- more like unhygienic.
And now everyone goes to school, not just the children of Danish administration. And a bunch of US/UK/Canadian genes get mixed it. The Danish genes have declined, sailing home to Denmark inside living Danes. This is not a mystery that demands explanation. This is yet another echo of the the 20c stories of decolonization and global war.
I'm not suggesting that my just-so story refutes the theory categorically, but I blanch when ever someone references this study as some kind of slam dunk. It makes you sound like Daniel Kahneman before the waves of the replication crisis reached his toes.
The technological singularity concerns are real, and to me what is most likely to cause it is the conflict between the US and China/ China's allies. As long as there is serious international competition for power, we simply can't slow down. You mentioned technological change slowing down, but I worry that it's not slowing down fast enough. This AI stuff is real, and it's coming for us all if we don't actually organize and regulate AI (and biological engineering, chemical weapons, etc). We can't be libertarian on AI unless we're suicidal. And it won't matter if we officially regulate AI if powerhouse countries with massive spy agencies/ research departments (like, say, the US and China) are trying to use whatever new toy they can to beat the other side (even if that new toy, is, say microscopic death robots that mindlessly wipe us out then expand across the galaxy).
So I think the question all the clever people on here should be asking is simply "how the hell do you get the US and China on the same page before the technological singularity?" Because a nuclear war is the tip of the iceberg when it comes to potential things that could go REALLY wrong here. We know how strong the tribal competitive instinct is in humans. We know how dangerous potential technologies are. And still, we hurdle towards the brink. We need something better than the United Nations, and we need it ASAP. That's the only reasonable path forward I see. Agreed, 2050 doesn't even feel like a real year. And still no one seems to be tackling these issues seriously. It's like we've all just given up. Glad at least you're writing a few posts on it before it's over Scott-
“I guess it’s still true that if innovation is destined to be only 10% of its current level in 2100, then a 30% population decline could lower that to 7%.”
You can’t do the math like that. Speed of progress is not so much about the absolute number of people. It’s about expectations.
When you anticipate a big population increase the pressure is high to build foundational tech. Investors give you money. Banks give you loans. All in anticipation that there is surging demand.
I think you miss the danger of the sclerotization of society. New people aren’t just any people - they’re young people. More open to new ideas. More prone to go into novel fields or set out to test novel hypothesis. Not only that but there is some decrease in openness to new ideas an increasingly octogenarian environment (I recall some study found decreased patents in universities with older professors or some such). The human population won’t be falling until 2100 in large part because life expectancy will be increasing. Japans population grown by 10,000,000s since 1960, but the number of 0-14 yo children has nearly halved (28 million to 15 million)! The dangerous demographic changes concealed by topline numbers isn’t that people will be less intelligent or more diverse - the danger is a world without young people is a world without early adapters of new ideas.
I have also 2 more spiritual arguments that I’m making for growing the population :
The 1st argument is that human consciousness is, as far as we know, the highest form of consciousness (or one of the highest forms of consciousness) on the planet, probably in the entire solar system, maybe in the entire galaxy, or even in the universe.
So every time we make a baby, we basically transform solar energy into a very high form of consciousness. We transform an inert part of the universe into a part of the universe that can observe and understand itself.
It’s a miracle.
On a more prosaic level, most of us are glad to be alive, and glad that our parents made us, therefore we can presume that most of our children will be glad they we had them, as will be the children of their children, etc. for millions of generations.
Every couple that doesn’t have kids is effectively preventing probably millions of people from existing in the future, across many generations.
My 2nd argument is on a more personal level: every one of us, without exception, is at the end of an *uninterrupted* chain of at least 3.7 billion years of life evolution (and maybe more, since it’s possible life was brought on Earth from elsewhere https://en.wikipedia.org/wiki/Panspermia ).
For 3.7 billion years (that’s 3 700 000 000 years), every one of our ancestors managed to survive and reproduce, without any exception.
Not having kids means we would be the very first in this chain to break it.
I’m arguing that we should have respect, and even be in awe, for this long chain, and think carefully if we really want to be the first one in almost 4 billion years to break it.
- one omission from the Amish Appendix: the Mormons. They don't have the fertility rate of the Amish (and it is going down), but it is high and they start from a much higher point.
'Primitives' like Amish, Mormons, orthodox Jews, Nigerians, Afghanis multiply while 'advanced' effective altruist rationalists wither.
- nothing in the physical world keeps doubling indefinitely. Not Covid cases, not the Amish. So the Amish population will eventually peter out before 2250, too.
- fertility rate among the Puritans in the 16th and 17th century was Amish-high at 7+. It came down.
1) Let's just go ahead and say the world isn't going to end, because if it is why discuss any of this.
2) I agree that "genetically engineered babies" would be a great thing to have. What measures would increase that possibility? I think the best methods would be:
A) Encourage fertility amongst those likely to produce the offspring that would invent such things.
B) Not do anything that would allow the societies likely to produce such outcomes to seriously degrade/collapse.
It seems to me that aging societies being flooded with low IQ immigrants are less likely to produce such an outcome than fertile societies without low IQ immigrants. Since genetic engineering will fix the potential immigrants low IQ problem, we should encourage such an outcome, even if present low IQ potential immigrants have to remain in their original counties.
What's the IQ of the West if all those Nigerians immigrate to the west? 97.5 seems very optimistic.
3) I think you misunderstand the worry with demographic pyramids. We have not been able to make medical care more efficient or labor saving for a very long time now. All those pensioners aren't going to need mass produced wheat or widgets. They are going to want the same overly expensive economy consuming medical services they want today. What happens when that is most of the economy? Is there anything left for human flourishing and innovation?
4) It's unclear to me what we are supposed to be more concerned about then long term effects? Like am I supposed to expend my mental energy on the latest Current Thing? When I think about more immediate problems we face, not a one of them would be alleviated by dysgenic trends (whether births or immigration).
Nearly every opinion I have on how to make a society better is shared by pro-fertile people, and nearly everything I think would make it worse is shared by anti-fertility people.
P.S. Having kids is good. Not having kids makes our life worse and makes you an objectively worse person.
Regarding dysgenics, isn’t the whole notion kind of presumptuous? Especially in the context of a large economically unequal society?
I picture a nobleman in 1600s England bemoaning that the lower classes breeding too much will cause average intelligence to decrease, ignoring the fact that 99% of the population does not have the chance to prove themselves “intelligent” by any officially accepted standards and that his own station at birth provides opportunities for higher education that the vast majority lacks. A farmhand might have been a professor (and might posses the genetic ability to have been so which he will pass on) given some different opportunities in life, better nutrition in infancy, a realistic path to literacy and higher education, etc.
Of course modern America isn’t as stark an example as 1600s Europe but we definitely do not live in a society where raw gifts of intelligence or talent at birth guarantee a rise to higher educational attainment and it feels naive and eugenics-y to think that we do.
Maybe I’m misrepresenting the argument and I’m willing to learn more, but as far as I see it Idiocracy is not a reliable way to think about the future of the world.
I have heard before the idea that it would be terrible to make a baby who might live in a troubled world, but I don't understand it. There's some assumption behind it that goes against some assumption of mine that's blocking me. It might relate to whether existence is good in itself (me) or neutral till you add pleasure or pain (?). Or something about how much we trust our ability to predict the future -- like in a Ray Bradbury story in which the world was going to end so two parents poisoned their children the night before so they wouldn't have to live through it, and then of course the world didn't end and everything was OK except their kids were dead. I wouldn't bet 10c on any particular outcome at this point.
If I thought AI was going to take over the world in 30 years and kill everybody, I'd still be glad my kids exist. They'd get 30 years of life before then (I see Scott has the same answer there). And maybe one of them could do something about it.
"I understand why people don’t want to talk about the issue this way, because if you say demographic shift is a problem, people will call you a racist conspiracy theorist.
I find this instinct a bit weird as well, since it seems to be a matter of scope and only to be applied selectivly. In many cases we assume a desire for exclusivity and raising boundries to be a sign of love or care. For example when a person gets married we accept that this is a sign of love for that person, not proof for hate of all other people. Exclusivity in regards to family and wider community is also often interpreted as a sign of care for ones community or family, not animosity towards all other families and communites. But with regards to nations the benefit of doubt seems to reverse.
And just to be clear: there is plenty of evidence (like hateful rethoric) to make the judgement that at least some of the desire for boundries on nation-states is based on hate. But still, how confident should we be in mind-reading this to be the main motive? Would suck if we were wrong about something this important, since if there are good motives on both sides we ought to take quite a different approach than if it´s a struggle between good and evil.
Present methods will never generate AGI and there is a lot of focus on machine learning because it actually generates results that people can use, at least to some extent; the other approaches are not yielding results at the kinds of speed people want because this is an extraordinarily difficult problem.
Machine learning is very reminiscent of Clever Hans.
If you are just messing around with these programs, it's very easy to get impressed.
But the more specific you make your demands - for example, with these image creation programs, the way I would when I was commissioning an actual artist, where I have something specific in mind - the more obvious it is that it is very much Clever Hans.
It can spit out a lot of pretty stuff - but it can't generate what I am asking for.
Moreover, it is very obvious when I actually have something specific in mind that it isn't actually even trying to generate what I'm asking for half the time, but is instead putting out something pretty. And I can tell this because I am trying to put this stuff to practical use and therefore have specific unchanging intentions.
This is not the sort of thing people notice with testing, because their test cases are not specific enough, and often aren't actually probing for problems. I see people do many of the same tests when experimenting with the program, and then see actual users trying to use these programs and it is very obvious what is going on when people are trying to put these programs to practical use is not actually matching the test cases very well.
If your goal is some broad request, sometimes it will give you a good result - ask, say, for a magical orb swirling with the magic of spring, or for a trail through a magical forest, and it will give you some very pretty things that are 100% usable. And depending on the AI these might be pretty good.
But if you have something very specific in mind visually, these programs cannot do it. They will often come up with visually interesting things - and I see people being happy with things that are produced all the time - but when I am trying to generate one specific thing that I actually already have in mind, it can't do it.
And this is the Clever Hans. These programs don't need to be able to understand English to generate images that will pass muster with most people.\
It's very obvious to me the more I use them that this is what is really going on, and it isn't really solvable because the approach is fundamentally flawed.
Worse, you can't just throw them at mobs of people for rating because people will upvote things that are visually pleasing regardless of how well they fit the prompt.
They will also completely fail at many requests and I think this is too easy to disregard as the model being underpowered and not that machine learning has some fundamental limitations and points of attraction that the models end up getting sucked towards because they aren't actually intelligent.
It's true of all of these approaches, because these approaches aren't designed to generate intelligent output.
And the scaling is terrible. I was listening in on a talk by the guy who is behind MidJourney, and it was very interesting, as he pointed out that at the present rate of growth (which he doesn't expect will happen for obvious reasons, but still), MidJourney would be using every cloud computer in five months.
And this is a fairly bad AI in a macroscopic sense.
This approach is both insanely expensive and not going the way people think it is.
Also, speaking as someone with knowledge of biomedical engineering and genetics:
We're a long ways off from understanding the genetic basis of human intelligence. Intelligence is an extremely polygenic trait, and the more genes are involved in a trait, the harder it is to actually make a super-whatever (genius, in this case).
Moreover, without very unethical human genetic engineering, doing experiments on this is extremely hard which makes it even harder to make progress.
While I expect us to have made SOME progress by 2100 thanks to large genetic databases, I would not expect the majority of children to be hyperintelligent by then even in the developed world, and in fact, I don't expect we'll be generating super-geniuses by then.
This is an extraordinarily difficult problem to solve. Computers have made it possible that we may eventually work it out, but 50 years (which would put the supergeniuses at adulthood by 2100) is pushing it, and even in 100 years I think there's better than 50% odds we won't be able to generate people of IQ 150+ consistently (though I would expect us to be able to boost intelligence at least somewhat by then).
The biggest threats aren't human genetic engineering or AIs, it's biowarfare and Russian and Chinese aggression, along with attacks on democracy.
6. Age Pyramid Concerns Are Real, But Not Compatible With Technological Unemployment Concerns
Likewise, I think population decline concerns are not compatible with global warming concerns. If each human creates a carbon footprint, then the subtraction of every human moves us closer to stopping global warming. If you really believe the latter is going to destroy the planet, then you should be happy about human population decline or at least neutral about it.
If you don't view human overpopulation and overconsumption as the #1 problem for the entire planet, I think it is because you cannot imagine giving up your own overconsumption -- hence, denial. Worry about underpopulation is simply insanity.
The U.S.A. had a spike in births (after a long general decline for over a decade) in 2021 that appears to be lead by college educated women working remotely.
Slightly Against Underpopulation Worries
Genuine question here, no snark.
If you're convinced the technological singularity is approx 30 years away, and that it's likely to be ~bad for humans (AI take over etc), then why are you trying for a baby with your wife?
I don't think this is the same kind of antinatalist point of "the world is bad, why bring more life into it". It seems like you seriously believe that something is different about this point in history, and to me it then seems a bit odd that you'd want to plunge someone new in at the deep end just when the robots take over!
How do you reconcile this?
PS: forgive me if I've misremembered about the baby part, think you said that a whole ago!
The first 5 minutes of idiocracy sums it up
"In general, less educated people reproduce less than uneducated people (although this picks up slightly at the doctorate level)."
First "less" should be gone. (This was a confusing one!)
"And I notice it’s weird to be worried both that the future will be racked by labor shortages, and that we’ll suffer from technological unemployment and need to worry about universal basic income. You really have to choose one or the other."
Of course those two are inconsistent. The correct answer is that technological unemployment is a myth. There is little credible evidence for that outcome. Some who talk about it are (consciously or unconsciously) motivated by a desire to promote UBI for independent reasons.
I'm not particularly "worried" about labor shortages, but of these two outcomes, I consider it to be the one with more evidence, by far.
How is the technological singularity not just rationalist millenarianism?
And finite natural resources do not merit consideration? Hand up who wants to live on Mars?
People with more education might have fewer kids, but does that actually mean that people with more heritable IQ have fewer kids?
(Ie can we naively multiply correlations here?)
“there will probably be a technological singularity before 2100”
This really contradicts the science is slowing down paragraph. The one just before it.
I agree that we might probably be able to make babies smarter by 2100, but the increasingly large Amish and orthodox community might be against it. Of course it’s post singularity so who knows what they will believe.
Absent your singularity the population trends for the Amish and orthodox will hit the Malthusian barriers of their more primitive lifestyle. I don’t know much about the orthodox lifestyle, and what I know about the Amish comes from Witness and wiki, but the latter don’t seem to take charity, as their population grows and they buy up more land - I assume this is what they do - that land must become less productive. The US will be a fairly primitive society in that era. And mostly white again. Haha.
The US is, outside these groups, is now a population sink. Population sinks are interesting because all lineages die out to be replaced by new additions, who then die out over time. The descendants of Londoners in the 12th C are not modern Londoners but subsequent migrants (internal to the U.K. mostly until recently).
"Even this isn’t quite right, because a lot of Orthodox Jews do leave Orthodoxy, so along with those 100 million devout Orthodox there will probably be a few dozen million extra Reform Jews with a confused relationship to religion and lots of emotional baggage. It’ll be a great time for the rationalist community."
Go team! (Also, at this point we should be separating out the Ultra Orthodox from the Modern, because the latter can't keep up. It's a good question how many Jewish-sans-Streimel groups we'd have at that hypothetical point. To be honest, Conservative and Reform could have merged in 2015 and nobody would have noticed. The other Orthodox groups are more of a question mark.)
Can you steelmann the case for and against AI Singularity. I personally think it’s a big issue but have a hard time making others agree with me
Isn't this a fully general case against caring about the future past a certain point? "Yes a bad thing will happen and it will have bad effects. But conditions will be different, it won't be THAT bad, etc." To be honest this reminds me of what people say about Global Warming. "A 2 degree increase by 2100" or whatever.
Or is that your position?
I agree that in the longer run sub-populations with higher fertility will keep most likely keep total fertility high enough.
However, if the Amish grow a lot more, I would also expect them to change.
Just like Warren Buffet can make money at 20% a year as long as he's small, he can't keep that up once he has an appreciable fraction of the total economy.
We should be trying to understand why it’s going down at all, because it might get worse. I think it’s because of poor diet and chemicals.
You just accept the population projections as fact, and this seems like a serious mistake. These projections have been wrong in the past, could continue to be wrong in the future, and one may make the case that they will predictably be wrong in the future. Our World In Data doesn't have a proof that their population projections are the most accurate projections possible given the information available in the present; they just have some model, the assumptions of which could be disputed.
That said, I agree about point 9.
Wordcels make this argument when they're horny and shy to admit it.
Besides demographic shift, my real worry here is the causes of declining birth rates rather than its effects. Are we too socially and emotionally broken to start families, or just too poor and financially insecure? Too individualist? All options feel terrible, and I suspect these three are all true and interrelated.
Point is well taken about unreal years, but it makes me wonder - what is the farthest off real year? That is, the last year about which we can make meaningful predictions in most arenas? 2026? 2030?
If we make it though the singularity, however exactly that looks, i know it may have been half a joke but I could see a world where some form of religion becomes highly selected. The basic tenets would be: don’t do things that destroy the patterns that sustain humanity, like sexual reproduction being tied to fitness, having to struggle to develop, etc. I always call that Space Mormonism but in a science fiction future where there are still recognizable humans I see that as the most likely outcome.
Another point against the dysgenics issue is that to whatever extent the Flynn effect is QOL based, it will be more powerful in developing e.g. African countries which are also making more babies.
So even if the average IQ of the descendants of today's Americans will lower from dysgenics, it will be more than compensated for by 2nd/3rd gen sub-Saharan immigrants (already a smarter than average slice) going through the Flynn effect.
I wish the singularity argument had been the first argument. The "Don't worry it won't get bad until 2100" left me with my mouth hanging open and then when it eventually got followed up with "and it won't matter by 2100" I was able to close my mouth, but I wish I'd been able to know up front that Scott wasn't trying to predict anything about population trends but merely making a point that, like everything else, he thinks population trends are inherently unpredictable on a 80 year timescale.
My guess is that the climate change consequences will make many places unlivable enough to result in extreme strife and significant decline in the population due to wars and shortage of basic resources caused by wars, not by climate change. And not by 2100, but more like by 2040, the way current extreme weather events are happening. Basically take what is going on in Ukraine and add one or two orders of magnitude. It won't be just a shortage of heating gas or wheat in Europe, it will be famine and lack of technological basics all over the world. Population decline will be a consequence, but a welcome one, rather than a cause for worry. This is assuming no drastic changes like AGI takeover.
In terms of the Amish and the Orthodox. My understanding is the Amish can only exist because they occupy some of the worlds most fertile farmland. Their lifestyle/economy can’t scale. The same is true for the Orthodox community in Israel. The most common example is they aren’t subject to mandatory service in the IDF. An Israeli without and IDF isn’t an Israel for very long.
I expect the year 2050 to be a real year. I am so convinced of this that large fractions of my total wealth are in retirement savings, which will not be touched until after the 2050s.
Have you decided that saving for retirement is a waste of money? Do you tell small children that saving for retirement is a waste of money? Do you plan to take out a large mortgage, on the expectation that you will never have to pay it back?
Is there any prominent person you would be willing to take a bet with, at any odds, on the year 2050 being post singularity?
So idiocracy is real?
Re. "if innovation is destined to be only 10% of its current level in 2100, then a 30% population decline could lower that to 7%.":
Because of our high population, the bottleneck in tech progress isn't creativity and innovation, or even money, but attention. You can see this in government funding: Grant-funding agencies are limited less by budget than by the number of administrators who can oversee contracts. Projects that cost less than half a million dollars a year are often ignored, sometimes (in my personal experience) to the point of the administrator not bothering to read project reports, answer emails, or attend meetings.
You can also see this in the proportion of media attention given to elite colleges. The ratio of students in non-elite to elite colleges has increased by about a factor of 10 since 1950, yet the ratio of media references to graduates of elite vs. non-elite colleges seems to have grown. Go through any issue of WIRED magazine (well, any issue from the 1990s, when I still read it) & see how many articles you can find about research by people who didn't attend MIT, Stanford, or Carnegie Mellon. The proportion of Nobel prizes given to graduates of top-20-in-field colleges has also increased radically since about 1960; in physics, it went from something like 50% in the first half of the 20th century, to 100% after 1970 (last I checked). The ratio of venture capital given to graduates of elite colleges vs. people with no college education was small in the early 20th century; today, it might be infinite.
We have a glut of smart people, yet if anything there are fewer super-smart people at the top. There are AFAIK no contemporary equivalents to Einstein, Turing, von Neumann, or EO Wilson.
No matter how big the population grows, each person still has enough brain space for the same number of new ideas. The only way we have now of countering this effect is to continually fragment into more and more isolated specialties. But this renders everyone less and less capable of recognizing creativity and intelligence in the wild.
So I do not think a decrease in population will decrease effective innovation. It might even increase it.
I participated in the "predict 50 years in the future" contest. I focused on genetically engineering people, which is going to happen and going to majorly impact society, the economy, and politics.
It will likely be the case that there is a majority that is below average intelligence (compared to present average) and a small group of fortunate people who are extremely smart, happy and healthy in 2100. Massive gains can be achieved through genetic engineering, most importantly for the near future might be embryo selection.
Selecting embryos from a small pool can confer advantages, and selecting from extremely large pools can confer large advantages. With enough embryos to select from, it's possible to get massive gains. See Gwern's article on this (https://www.gwern.net/Embryo-selection).
There is good reason to think that this will be achieved in near future, and it will majorly impact society for the better. It's difficult to imagine what a society with a million 250 IQ people would look like. I do think that there will be a lot of interesting political and ethical issues. Will the practice be prohibited? How large will this population be? Will parents leave the country to have genius children? I discuss this in my article: https://parrhesia.substack.com/p/america-in-2072-a-society-stratified
I also think that effective altruists should focus very strongly on protecting this right and start devising rhetoric to defend it. Otherwise, the risk is that it is banned, and that lives lived are much worse. This is an injustice. Also, super genius babies could help prevent X-risks (a point made by James Miller).
Another scenario that can turn the population dynamic upside down will be the discovery and mass availability of rejuvenation therapies. I don't think bological immortality will cause overpopulation, but the population pyramids will start to look *really* weird
From a consequentialist perspective, it seems like we should take the non-existence of people seriously. If we could boost fertility globally and achieve 20 billion in 2100 instead of 10 billion, we are going to see a lot more human welfare. Failing to do this should be treated similar to a persistent disease that's killing literally billions of people. Low fertility must be regarded as one of the worst things in the world at present if we regard future possible people as having equal moral worth. If all that matters is the consequence, lowering fertility should be treated like a massive wave of tens of millions of infant deaths. This should be extremely concerning.
edit: total consequentialist perspective
It’s comical to me than people really think depopulation is the big problem facing us. The world population is almost 2.5 times what it was when I was born. Many of the actually big problems facing humanity, like climate change, species extinction, ocean pollution, etc. are automatically improved when you have fewer humans causing them. I think the real fear is that the Ponzi economy is unsustainable unless there is constant population growth (= more people joining the scheme). But all economies eventually fail; just try to spend a sestertius now. So I can’t see that short term pain as some existential crisis.
"I guess it’s still true that if innovation is destined to be only 10% of its current level in 2100, then a 30% population decline could lower that to 7%."
But if it's educated / high-IQ people that have the lowest fertility (such that they're disproportionately responsible for the 30% decline) wouldn't that imply that innovation drops lower than 7%? In other words, the proportion of potential innovators would get smaller over time. This effect might even be stronger at the tail end of the IQ distribution: a small relative decrease in high-IQ individuals would mean a strong relative decrease in potential superstars.
Robin Hanson thinks falling fertility is <i>The</i> Big Problem of our times, but mostly in a total-utilitarian sense* rather than a looming disaster sense:
https://www.overcomingbias.com/2010/11/fertility-the-big-problem.html
He has also dismissed writings about technological unemployment. I'm also not too concerned about that, both because of the evidence he's presented about recent history & the near future, and that if we do achieve that our civilization must be succeeding somewhat.
*After writing that I rechecked his post and saw that he thinks that lowering economic growth via lower population & knock-on effects would reduce our robustness to a variety of existential risks. So heightened probability of disasters, but not looming directly from lower populations.
"if we can’t genetic engineer superbabies with arbitrary IQs by 2100, we have failed so overwhelmingly as a civilization that we deserve whatever kind of terrible discourse our idiot grandchildren inflict on us"
I find annoying arguments throwing up our hands and saying we "deserve" bad outcome B if bad outcome A happens. We should try to be robust to bad outcomes. An example of Robin Hanson thinking about that for a low probability but severely negative outcome is here: https://www.overcomingbias.com/2008/07/refuge-markets.html
I should also note I'm much less convinced that the singularity will happen that quickly. If it happens in my lifetime, that will probably be thanks to life extension. 2100 is still far enough off I don't want to confidently speculate about it though.
Also: even supposing that a singularity fails to happen- that the inevitable unpredictable historical twist turns out to be that not much changes- aren't we actually going to need a stable human population at some point? I mean, we can't just keep growing geometrically century after century, right? Why, given a future without much technological progress, would it be better to put off that stabilization until a future generation? Presumably, they would face the same kind of negative consequences from falling population growth that we do today, only more people would be affected.
There seems to be this idea in the air that humanity can escape any future need for population stabilization with space colonization, but aside from just delaying the problem further, that seems to ignore just how profoundly terrible it would be to live in an extraterrestrial colony in the absence of radical technological change. Mars isn't the American West- perfectly suited to human life and covered in recently abandoned farms that smallpox-resistant colonists could pretend were untamed wilderness- it's a hellhole. Glorious sci-fi dreams and the pioneering spirit would only obscure the lived experience of that for so long, and you can't spend centuries dropping asteroids on the planet for terraforming when there are people already living there.
If some future generation is forced into space- or even into the uninhabitable parts of Earth- by overpopulation, that would strike me as a profound failure of humanity.
Scott, it seems the answer is ‘yes’ to some degree based off the comments, so my apologies if I’m misinterpreting you, but do you think we all have a very high chance of dying after AGI debuts? Innocently earnest question, I know it’s been asked a million times around these parts to different people and answered in million different ways, but I’m curious what your take-it-or-leave-it answer would be today at this very moment.
"I notice it’s weird to be worried both that the future will be racked by labor shortages, and that we’ll suffer from technological unemployment and need to worry about universal basic income. You really have to choose one or the other."
I can think of a couple of ways to reconcile these worries. For one thing, you can imagine someone's concern about the increased burden of support for the elderly to be primarily about direct physical support: nursing, caregiving, etc. Even granting a large population of unemployed people interested in working, it doesn't have to follow that enough of them will be interested in working in the narrow field of personal care.
Beyond that, even if we limit the discussion to worries about the increased burden of *financially* supporting the elderly, these worries need not be incompatible with worries about technological unemployment, either:
I assume you'll agree that fears of technological unemployment, by and large, cannot be about unemployment leading to resource scarcity in the aggregate, since by definition, firing human workers to replace them with more productive machines leads to more resources in the world, not fewer. Rather, then, these fears surely must be about the distribution of these resources: the increased productivity accruing to the owners of the machines will not flow to the displaced proletariat without some sort of redistribution (e.g. the universal basic income that you alluded to).
I further assume that you will, if not fully agree yourself, grant at least the reasonability of the proposition that the responsibility of supporting the elderly is not spread equally across society, but devolves primarily to those closest to them, mainly their children. (I'll further point out that if you choose to challenge this proposition while granting the first, *you'll* be the one in danger of inconsistency, since the problem of redistributing income across society to support the elderly is surely no different in kind from the problem of redistributing income across society to support everyone in need of support. If worries about technological unemployment are predicated on pessimism about society's ability to adjust its redistribution systems fast enough to support the unemployed in general, it's surely consistent to be likewise pessimistic about its ability to do so to support the unemployed elderly in particular.)
Thus, to me, it appears quite consistent to fear that in the future, many more jobs will be performed by machine, the rich will grow richer, but others will struggle to support themselves at all, let alone their aging parents, the responsibility for whose support they will now have fewer siblings to share.
I used to be near-totally apathetic about population decline, and in fact considered it a good thing compared to the alternative of growing until we hit our planet's carrying capacity. That said, regarding #6, I was recently startled by the release of the newest census results, which revealed that Canada (where I live) is becoming a country of olds with shocking rapidity.
I am not as sanguine as you regarding the ability of technology to close the productivity gap. Everybody thinks technology can replace somebody *else's* job, but rolls their eyes at talk of a robot replacing *theirs*. Given that the tech industry has completely failed in their quest to replace long-haul truck drivers, which I thought would be low-hanging fruit for modern AI, I'm now in "believe it when I see it" mode on this subject.
Ultimately, though, you're probably correct that making labor more valuable is likelier to be a good thing than a bad one.
Beyond the problem of low-fertility states with large entitlement programs (which won't be sustainable with fewer future taxpayers,) as I've written elsewhere: "While people with kids and people without kids can all be responsible, only people with kids have a special connection to the well-being of successive generations: the double helix. And with that special connection comes a stronger incentive to act in accordance with that special concern for the well-being of future generations, generations that will include one’s own children if you have them."
https://paultaylor.substack.com/p/skin-in-the-game-part-2
So what blog post should we start at if the idea of an AI technological singularity makes us laugh because even the height of well-kept technological infrastructure is max akin to the brain of a dying Huntington's patient who has just been shot in the head with a nail gun several times
The semi-things to "worry" about is that a) the way most retirement systems are funded is with a tax on workers. If they were funded by a VAT things would look different. b) the wage taxes that have paid into US retirement fund have not been enough and so tax revenues (wage or VA) need to increase unless benefits change.
Metaculus: “10% embryo selection for IQ: when?”
https://www.metaculus.com/questions/9785/10-embryo-selection-for-iq-when/
It seems like we shouldn’t worry about declines in IQ when technology can start driving the reverse trend
Overall Amish birth rates apparently have gone down over time, although subsets of them are still high: https://medium.com/migration-issues/how-long-until-were-all-amish-268e3d0de87
I agree that 2100 is going to be very strange, to the point where trying to worry about fertility rates then is likely to be irrelevant. At the very least, we'll probably have one or more of the following:
1. Meaningful life extension, which could range from "people can feel like they're in their thirties until their fifties" to "we solved aging and also cancer".
2. Artificial wombs for people
3. Very good nanny robots
All three of those together could really change the math on population growth - and even if it doesn't, heavy life extension is likely to lead to offworld migration to get around people in power being reluctant to step aside (and now they won't be dying off).
Underpopulation arguments are always strange to read, as a moderately smart person who wants kids but can't afford them (nevermind likely needing Singularity-class tech tree improvements to make it biologically possible, thanks sterility). Some of that is due to "suboptimal" life choices - I could have stuck with school and joined the PMC, I could have stayed in soul-destroying but better remunerated white collar work - but there's only so much changeable on my income end vs. the stupidly high local cost of living. Forget adding a kid to that, if just paying "cheap" rent is already 1/3 of wages...
("why not just move?" Because I was born here, entire family is here, and it's all I've ever known. No friends available in more affordable areas to cushion what would be a massive disruption to my hidebound routine-heavy life. Very high costs make even objectively rational moves quite difficult.)
SF wouldn't be Childless Capital of America if the relative equality of income vs. cost of living were lower. It's often stereotyped as a playground for the childlessly wealthy drifting through life, but I can assure you, in the lower deciles most of us do want kids.The ones who go ahead and try anyway, and sacrifice basically their entire net worth to try at that SES...they're way braver than I am, and I wish them well, but man. Even without (let's be humble) elitist worries about dysgenics, that's far from ideal child-raising circumstances. Maxing out whatever potential *is* there, in such children, is surely less likely when flirting with poverty and family instability. I wish we could do better. If that means taking up the same standards as the underpopulation-worriers, because no one else will advocate natalist policies - well, so be it.
I also wonder about how much hinges on *relative* population levels, no matter what the *absolute* level is...not just in the native/immigrant dimension, but between country regions, and between countries. We see the same thing play out in "income inequality", "education inequality" (that degree graph with a mere * asterisk in place of a bar for black PhDs...yeah, that's A Thing.), "racial inequality", and so on. One doesn't have to be wokely progressive to perceive the dynamics at play, politically and otherwise.
Age distribution is a good example given in the post - it doesn't matter quite so much about the absolute number of young vs old, but whether we can maintain that pyramidal distribution, since otherwise our current form of elder welfare isn't sustainable. One can easily draw parallels to US politics (House vs Senate apportionment, school district financing, gerrymandering...). And international relations too - the US:China ratio settling towards 1:2 instead of the current 1:4 or so, that will certainly have ramifications. Obviously, it's harder to predict distributional shifts vs absolute shifts, but it's an intriguing thought experiment.
Lastly, just to complete the Rationalist Singularity trifecta (AI, technological productivity, and ____) - how should one update on cryonics or other forms of Death Is The Enemy? Making new kids matters somewhat less (in either direction) if we can hang onto already-extant people operating at reasonable efficiency for longer, or possibly even bunker a "backup generation" to guard against x-risk. (Argument From Fictional Evidence: worked great in the Fallout universe!) It's certainly still within self-interest utility-maximization to not die earlier if possible, but we're talking about vast system-level effects here.
Good points about ethnicity/demographic change.
Those of us who believe in homogenous nation states (pejoratively called “nationalism”) are not really invited to discuss the rationale for this position because “racist.”
It’s ok to want Britain to be for British people. And to have an idea of what a typical British person is on several dimensions of homogeneity.
Also, not only am I not worried about underpopulation, I simply don’t want to cover the entire planet with concrete and steel.
If population declines get bad enough, I'm sure some tyrant will simply compel or pay women to provide surrogacy services for the state. It might be a way for women to raise their social credit score, reduce their jailtime, get access to universities, etc. Of course, they baby they would gestate wouldn't be genetically theirs - the sperm and egg would come from more "promising" people. They could also recruit surrogates from outside their country - for example, among women fleeing a warzone or natural disaster.
Anecdotally, many great Jewish scientists are reported to be the grandchildren of famous rabbis. This could be a place to study whether (1) it's possible to reverse any IQ depletion effects by with policies that encourage smart people to have kids, (2) whether in practice those policies are justifiable within a secular morality, and (3) whether these "inversions" have always been happening, and with other effects are actually maintaining a dynamic equilibrium. (Generally people are no more religious than their parents, so the only way for religious Jews - who don't proselytize - to keep their numbers up is to have lots of kids.)
A lot of people quote Keynes to argue that in the long run we are all dead. That may be true, but the whole point of Keynes argument was that thinking that way ignores the issues that have to be grappled with before then, like the rest of our lives. His analogy was to a weather forecaster arguing that one need not batten the hatches in a storm because the storm will eventually pass. If you've ever been in a storm at sea or even just been out on choppy water, you know this is nonsense.
Sure a singularity, let's say Jesus Christ coming back and pulling off a vastly expanded miracle of the loaves and fishes, would solve a lot of problems. That still leaves us with the problem of managing things until that happens, and let's not forget the subsequent War of the Loaves and Fishes followed by Jesus throwing up his hands and trying his luck elsewhere in the galaxy.
> (there might be an exception for fields of science that couldn’t have existed in the 19th and early 20th century - see here for more)
This is already a fiendishly difficult question to quantify, so let me go ahead and make it downright intractable: what was the rate of discovery in 1880-1920 v. 1980-2020 specifically within fields that existed in the *18th* century? I think I'm inclined to believe that an awful lot of low-hanging fruit became available either all at once or in a quick feedback loop in the mid-19th century (for reasons largely but not exclusively due to materials science and global trade), and this heavily distorts treatments of the matter than only go back a hundred years or so.
I think in modelling the speed of scientific advancement it's worth throwing a 'friction' term in there. According to the low-hanging fruit idea, science is getting harder and more complicated, and more work needs to be done by more researchers to go the same distance. But as research is done faster by more people, it gets inherently less efficient.
One reason is that the cycle time for most science projects is several years, and throwing more scientists at the problem just means you're parallel threading, not necessarily speeding things up. Another is the difficulty of communication. With so much work done so quickly it's difficult to keep up with what's going. You need to spend a lot of time reading to make sure nobody's solved this problem before, and to find the hidden gem that could catalyze a breakthrough. And that's time not spent doing research. I think there are other contributors as well but those are the most common-sense.
The primary implication is that we may be overestimating the impact of the low-hanging fruit effect, and some of the science slowdown is due to 'research friction'. If the size of the researcher pool stops increasing as quickly, then this friction will stop increasing alongside it. If we can figure out better frameworks for science, maybe we can even decrease this friction term and speed science up in a way that seems hopeless with the low-hanging fruit argument.
On point 6, I think there is a real concern with what it will mean politically if automation/health mean that retirement ages don't go up with the overall demographics, and retirees become politically dominant. There have been plenty of gerontocratic societies in history, but the "democratic gerontocracy", where people of working age are simply outvoted, will be a new phenomenon in human history. There are already hints of this in countries that have extreme age gradients in voting patterns, such as the United Kingdom with its increasingly nostalgia-based Conservative government. Even if retirees act in rational self-interest, rather than voting for the "good old days", they have peculiar policy incentives, such as being unconcerned about wages and having little interest in education, but very concerned about healthcare hypersensitive to anything that erodes their savings, such as inflation. This could even be self-reinforcing: if the entire population of childbearing age lacks political clout, the government could end up making parenthood dangerously expensive. I wouldn't worry about "dysgenic" effects so much as the lifelong disadvantages caused by bad childhood circumstances, things like poor nutrition and lack of parental attention because parents are too busy with jobs. Even if society as a whole is well off, if most children grow up poor because of lack of economic support for parents, that means a significant degradation of human capital from one generation to the next.
A related danger is the return of the kind of rentier elite that existed before the Industrial Revolution, where most wealth is inherited (and passively accumulates purchasing power through being invested in property, index funds and so on) rather than gained through labour or entrepreneurship. Low fertility rates will accelerate the concentration of this inherited wealth into fewer and fewer individuals' hands. Richer countries will be able to import workers, but those new workers' assets and wages will be paltry compared to the amassed fortunes of their landlords and the owners of the businesses they work for, and they'd be liable to worse exploitation as a result. Immigration also plays into the political battle between old and young: if there is a mass of immigrant workers who cannot vote, that effectively amplifies the power of the mostly older citizens, even if the total population pyramid isn't so skewed.
I am presenting the following two items without further comment. I think they are both credible.
I collected them because the tickled my priors which are:
1.Malthusians are wrong and like the Bourbons in 18th Century France, they have neither learned anything nor forgotten anything. &
2. The conventional wisdom is always wrong.
That makes sense, at least for now. I gather the Amish have been moving across the Midwest in their pursuit of good farmland. They haven't bumped into agribusiness yet, at least not seriously. I know that the Amish are pretty good farmers, but those guys running agribusiness can be nasty.
> If we don’t die of something else first, there will probably be a technological singularity before 2100. The way things are looking now, it will probably involve AI somehow. If by some miracle that doesn’t happen, we’ll get one involving human genetic engineering for intelligence. I think there’s maybe a 5-10% chance we somehow manage to miss both of those entirely, but I’m not spending too many of my brain cycles worrying about this weird sliver of probability space.
I don't mean you any disrespect but you're so extremely, unfathomably wrong about this. The chance that NONE of the above happens is, in my casual estimation, 99.99999%. Honestly that's an arbitrary number of nines, I really just wanted to write 100%. Point being: not gonna happen.
That's not what I'm interested in, though. What I'm interested in is how I can BET on that. Those folks predicting "AI" by 2029 on Metaculus... let's just say if it was a real prediction market with money, I'd bet every single penny against that consensus. But it's not! So where's the real-money version of Metaculus that will let me separate futurists from their money?
Sources on the population growth rate? India is at replacement rate this year according to the WHO.
I humbly suggest that Japan as it currently exists is not completely interchangeable with a Japan made of 1/3 ethnic Japanese people from today and 2/3 ethnic Japanese people 100 years from now.
Yet we see few calls for population control because "the youth will change our culture" (which they will). Ethnicity seems like a red herring if your goal is cultural preservation (which is what I assume drives concerns about 'demographic shifts').
I'll be honest. This blog usually flies over my head and this article is no exception. But be forewarned, the line "..a cloud of microscopic death robots that used to be our solar system has expanded as far as Sirius B.", will in fact be used in my first science fiction feature film. Release date TBD. (Before 2050 of course.)
The counterargument to the "dysgenic effect" is pretty simple: it should have been a constant throughout history, and yet here we are anyway. Clearly it's not an issue. Don't @ me.
I think you're slightly missing the point and most people that talk about the subject specifically talk about 6, but mask
Okay, my first thought regarding a more Amish and Orthodox Jewish U.S.A. is it will be a boon for hat makers.
Second thought is to re-read News from Nowhere and watch Fiddler on the Roof for a clue of what that will look like.
Third thought is: That sounds AWESOME!
While I’m more fond of the postwar mid-20th century as a destination, these seem like good trends to me (and my Stetson will fit in alright)!
This is a little out there, but I think some of the anxiety over depopulation stems from a deeper emotional-philosophical place. Population decline is the nation- or planet-level version of finding gray hairs in your beard. It’s an uncomfortable reminder of the reality of death and the impermanence of humanity.
Population decline prompts questions about what we are really doing here on this planet. Most ACX readers probably accept some version of the idea that we are amusing ourselves while we wait for the heat death of the universe, but that’s not something people are psychologically ready for (I think this has a lot to do with the dramatic increase in depression).
Narrative religions like Christianity and Islam were - and are - immensely powerful because they placed both individuals and humanity in a larger sequence of events that was building towards something. Even today believing strongly in certain religions is a strong indicator that people will want children. Some of that may be social pressure but religion also gives you a sense that you are part of something that will continue - it gives some purpose to having children.
It’s the same as the feeling you get when you’re in a movement that seems to be growing. But when the movement begins to decline and shrink, you feel a crisis of purpose. The threat of a declining world population or national population is giving people that same feeling: is humanity a sinking ship? And the answer is probably: yes, but don’t worry about it. Time passes and change is eternal.
How long will it take for our mesa optimizers to update so birth rates in the developed world recover?
Is western culture, which has been exported throughout the developed world, unfit for people? If so, is this unfitness intrinsic, or can it slowly be mutated away through either immigrants with different values or the above mesa optimizer update? What are the parallels to the urban penalty?
If it is intrinsic, can a different culture take its place as the world dominant culture, or will we be stuck with a cultural black hole that sucks in people and slowly kills their lineage?
> Workers will be able to expect high salaries and good working conditions.
This seem to me like moving your eyes off the ball, thinking about money instead of consumption. You can't eat high salaries, and what you can eat is more expensive if the work that was put into it was more expensive. You certainly can't eat hight salary that the government redistributed do that your old neighbour would eat too, or that you (a single child let's say) choose yourself to use for your parents medications.
Just a minor point, but my understanding is that at least in Nordic countries, fertility levels by educational level are now such that the highly educated are the most fertile group, and in 2020, the fertility of the highly educated actually increased while it decreased in other groups. See eg https://www.stat.fi/til/synt/2020/02/synt_2020_02_2021-12-03_tie_001_en.html for Finland.
Point 7, dysgenics: How about Africa? No one knows whether Africans' low average results on IQ tests are caused by genetic factors. But do we know for sure that they are not at all caused by genetic factors? If they are caused by genetic factors to any degree, altruists have a huge problem.
The proportion of Africans to non-Africans is supposed to alter dramatically in the next few decades. If Africa doesn't become self-sufficient in food and medicine in the next few decades, our children and grandchildren will have to be much more altruistic than we are: If nothing changes, a dwindling number of Westerners will have to provide for twice as many Africans.
Sure, things can change. But the null hypothesis, that things continue just as they are, should at least be considered as something that could happen.
About why science is slowing: Maybe I am biased by my field, but I do not think that the problem is low hanging fruits.I am a BSc, and I invent all the time intersting publishable things. I know they are publishable, because I almost always find that they were published in the last few years. The problem is that the constant flow of low hanging fruits is simply not wide enough to feed that many scientists. Some scientific work is incremental, and trying to give it ten times as much workers would just mean that five will be very busy thinking basically the same thoughts, three would find some intellectual desserts they can specialise in, and the other two will try to do something that is impossible now and will be very easy in five years, because at least if they succeed it would have some value.
I agree with most of your post above, but I think there is something additional and interesting to be said with regards to point number 6 specifically, which is relatively easy to miss if you have an American-centric perspective only.
And the point is that meaningful changes in the ratio of working adults to retirees matter not just economically, but also politically. Older people simply have different concerns and goals they want to get out of the political process, and many of them are short-term and fundamentally incompatible with the type of medium- or long-term thinking that is required to solve systemic issues that hold countries back.
I'm not sure how relevant this is for the US specifically, but as an example of what I mean, in many European countries (and especially Eastern Europe) pensions for retirees are primarily public, in that they are provided directly by the government to most people over the standard retirement age. My personal experience living in these countries (which is also confirmed by most surveys) is that for these people, the issue they care most about when voting is ensuring that those pensions stay as high as possible, and everything else is secondary to that. Macron's recent underperformance in the French presidential election, as another example, has also been attributed to his promise to reform the pension system (through raising the retirement age), which angered many older people who would rely on it as a primary source of income in their latter years.
Scientific consensus is clear that old people vote at far higher rates than younger people do, and that they are less concerned about issues such as climate change (which is understandable, since they'll be dead long before this would really materialize). In these European countries I mentioned above, many of them are truly single-issue voters who incentivize governments to spend a lot of their budget on pensions, leaving far less for everything else. Systematic improvements to the educational system, environmental concerns, anti-corruption fights (especially in Eastern Europe) etc all end up falling by the wayside, even though they are incredibly important for the long-term development of a country, because appeal for them comes almost solely from young people, who are already less likely to vote and are becoming a smaller percentage of the population every year.
The UK population projection here looks rather outdated (I think it's from 2015) and doesn't take into account recent falls in fertility rates. See https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/populationprojections/bulletins/nationalpopulationprojections/2020basedinterim for something more current. This projects that *over the next 10 years* deaths will exceed births. So in fact the population growth, even over that period, is not mostly immigration but entirely immigration.
Even that assumes that the large decline in UK TFR over the last 10 years quickly levels off, see:
https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/populationprojections/methodologies/nationalpopulationprojectionsfertilityassumptions2020basedinterim#:~:text=The%20fertility%20assumptions%20are%20set,%2C%20will%20be%20mid%2D2045.
I basically endorse all this except I am slightly less "optimistic" about getting a singularity soon. How about a bet where I pay you one bitcoin now, and you pay me two bitcoins in 2037 if there isn't any artificial superintelligence yet?
Re 8. I am not convinced by the premise that increasing the number of researchers we should expect to see a proportional increase in the rate of scientific discovery. Even assuming an infinite supply of low-hanging fruits, I would naturally expect diminishing returns.
In science it is not at all uncommon to find out that at any given moment there are a bunch of other groups that are working on the same thing as you. Scientific ideas are discovered when it's the time to discover them, when enough background has been accumulated that you can have a certain idea - and you are not the only one that will have that idea at roughly that same time. So I expect that just increasing the number of researchers is not going to give you a proportional increase in the rate of scientific discovery, just increasing the number of people working on the same thing with only a marginal speedup in discoveries.
If my model is correct instead, there shouldn't be any significant reduction in scientific progress with not-increasing (or diminishing) population numbers.
OK, we're all going to be paperclipped in 2053, but I'm surprised that even here in the craziest most Parfitpilled corners of the internet Scott doesn't feel the need to include as one of his eight fake worries about underpopulation the argument that, all else equal, it's better for there to be more people than fewer.
If the Singularity is called off and the world asymptotes to a steady-state with 10 billion people that persists for a million years... isn't that world much worse than a world in which 20 billion people get to live for a million years? Maybe not twice as good, and there's lots of arguments against this position - you might consider and discard the basic idea that being alive is good but not even to acknowledge it seems weird to me.
I was surprised by the Icelandic IQ decline research, so I've read through it, and it only indicates a projection of a decline of the mean IQ score in Iceland, they don't actually measure it. A big difference IMO.
yes, its always struck me as a strange concern.
I'm trying to write a sci dystopia and I'm having trouble imagining how things can go as wrong as I want them to go for story purposes, at least without the society completely collapsing due to war or something.
Re: dysgenics, here's Noah Carl's latest analysis:
'When fertility is negatively correlated with a socially valued trait, it is said to be dysgenic (the opposite of eugenic). So while fertility within populations may be slightly dysgenic, fertility across populations is strongly dysgenic. Average cognitive ability in the world is going down.'
https://noahcarl.substack.com/p/iq-and-fertility
Some technologies have a minimum population requirement. Galileo Galilei only had access to great lenses to experiment with, because lots of people bought them to make glass windows. In the preceding centuries cities were smaller, markets more fragmented and cost of transportation demanded a higher price. In general, technological progress is accumulation of steps of work in single artefacts and market size is among the conditions of their existence.
Take a tour through the history of Roman sculpture and watch the amount of detail decrease as their population declined. https://en.wikipedia.org/wiki/Roman_sculpture
A smaller city would not sustain a ship crew sized masonry to keep its portrait galleries up to date. Rome at its peak did.
A likely candidate for something prohibited by smaller population is use of rare elements. Will we continue to filter tons of granite for some ppm of erbium if we have less miners in the future and their wages rise? Or will we find "easier" ways and just return to copper?
Hmmm. So I don’t think these are necessarily “straw men” that you knock down, but I don’t think you really spend much time considering the arguments in favor of “aging demographics are absolutely awful” that I, personally, find to be most compelling.
I.e., 1) the reality of unwanted childlessness among women and the attendant moral and humanitarian consequences of this; 2.) that this has literally never happened before in human history and we have no idea wtf happens when our population structure inverts; 3.) you reference but don’t spend much time at all engaging with the likely economic effects of having almost no young people and so many old people — we have no idea whether the civilization we have built in the West is even sustainable under those terms. This isn’t just about “fewer people” it’s about demographic composition.
There’s also an underlying moral framework I have that you seem to lack — which is that two people who are living and existing and happy is better than one. And that having less human beings who could be living and existing and being happy, when you could have more, is inherently bad.
Also “2100 is so far away why worry” seems completely ridiculous to me. That’s seems to me to be an argument against literally everything, and also undercuts your (near) certainty that technological change will render demographic change meaningless.
"I understand why people don’t want to talk about the issue this way, because if you say demographic shift is a problem, people will call you a racist conspiracy theorist. I don’t think it’s racist to care about ethnic demographic shift - I think Japan as it currently exists is not completely interchangeable with a Japan made of 1/3 ethnic Japanese people and 2/3 ethnic Kenyans."
I think it is, if not racist, then at least extremely pointless to worry about demographic shift. What is the utility in having e.g. a US that is 60% white vs one that is 30% white? I struggle to think of non-xenophobic reasons for preferring the former.
I wrote something similar a while ago from a mostly different perspective: traits that cause people to want to have children in an environment that contains birth control seem to be heritable and so we should expect them to be selected for. That's mostly Agreeableness in Big 5 terms but probably it's more complicated than that. I also thought the technological landscape changing would render the issue moot, anyways.
https://hopefullyintersting.blogspot.com/2018/05/falling-fertility-rates-shouldnt-be.html
Scientists in future will be what farmers used to be.
Imagine a world where we completely recycle everything we produce and consume. In such a world you couldn't simply invent a new product and start selling it. You would need to provide the whole closed loop process. How it will interact with the environment, how the product will be recycles etc. Studies after studies will be required before the regulators will allow it and then the post-marketing surveillance and more studies to confirm that recycling is being done efficiently.
Being a scientist is a very boring job. I don't know why people think romantically about it. It is an important job, for sure. Just like being a farmer is important to ensure we don't starve.
I never thought I would imagine a comic book series about the Amish and the Orthodox fighting for control of the world (and pacifism doesn't rule out some sort of struggle!)
Has anyone ever noticed that economic growth correlates exactly with population growth? That economic growth can be artificially stimulated by any measure that increases population, including the immigrants for whom western economies will soon be competing?
Think of a bacterial population in a petri dish. Their population growth curve is precisely like that of the world's human population, and reproduction is automatic with only one factor controlling it: the amount of nutrient left in the agar. Our curve is the same shape, but it's not (yet) natural resources or food that controls its slope, it is economic factors in charge. And as far as I can tell, there is a bi-directional dependence. Increase wealth (eg discover North Sea oil) and the population curve is stimulated. Conversely, increase the population, say with a flood of immigrants, and economic activity rises. And if grow or die are our only choices, I can't see how there will be any kind of future stable state.
There is a lot of speculation that China's official population statistics are inflated and its population has been falling for decades. The overcount may be on the scale of a Japan (~120M) which would put China's population at 1.28B. There seems to be tacit admission to this by the CCP based on this being reported in the SCMP.
> I don’t think it’s racist to care about ethnic demographic shift - I think Japan as it currently exists is not completely interchangeable with a Japan made of 1/3 ethnic Japanese people and 2/3 ethnic Kenyans.
Even conceding this much is overlooking an important point (one that racists would very much like you to overlook). The Japan of 2020 is not interchangeable with the hypothetical Japan of 2120 where 66% of its population has Kenyan ancestry. But! Neither is it interchangeable with the hypothetical Japan of 2120 where the population is 100% ethnic Japanese!
Both of these Japans are hypothetical and exist far in the future, after a hundred years of technological and economic change. Ethnicity is probably the *least* interesting change that will happen in 100 years. What's a more significant difference between America of today and America of 1920 - the fact that we have atomic bombs or the fact that we have more black people?
Racists want to frame it this way because it invites you to compare the Japan of today and the Kenya of today and ask which one you prefer. But what the question is *actually* asking is if you prefer the Japan that accepts Kenyan immigrants for 100 years or the Japan that doesn't, and when you look at it that way you might notice that none of the other migration waves that were supposed to destroy our country actually did that. Immigrants assimilate, cultural conflicts of the day get paved over by the unstoppable forces of economics, and a hundred years later, the only record of the Great Kenyan Migration is that the stereotypical generic anime protagonist now has dark skin.
Here's the other side to the Amish taking over the world thing: https://www.theonion.com/amish-give-up-1819563928.
"After all, consider the century 1820 - 1920. It gave us the steamship, the railroad, the automobile, the factory, mass production, electricity, refrigeration, radio, the airplane, etc, etc, etc, with a population only about 10 - 20% as high as today. The effective innovating population - the number of educated people living in countries on the technological frontier - was probably an even smaller proportion. About half of these innovations came from Britain, a country with about 0.3% the current world population."
Excellent point. Today, it's much, much easier for smart people across the world to put their intelligence to use developing new science or technology thanks to better university systems, cheap international travel, and liberal immigration laws in most innovative countries. During the 1820 - 1920 timeframe, how many Chinese geniuses were stuck on family farms for their entire lives because there was simply to route to higher education?
Even if the overall population trend is dysgenic, we are getting better at leveraging the talents of the smartest people.
And how about the 'Idiocracy thesis'? That seems to be a much more worrisome scenario than depopulation, and a more likely theme of Musk et al, no?
On #7, shouldn't we expect a long-run offset from increasing access to education? In the US there may not be as much space for gains in that regard. Access is pretty broad from what I can tell - OECD numbers have the US around 92% high school completion and 50% for post-secondary degrees. (https://nces.ed.gov/programs/coe/indicator/cac/intl-ed-attainment).
But other countries still have a lot of room for growth. China went from 9% to 15% college attainment over the last decade. (https://www.bofit.fi/en/monitoring/weekly/2021/vw202120_3/).
Other OECD countries like Turkey (13%), Austria (34%), Italy (20%), Mexico (19%), and Germany (31%) presumably have room to keep growing their rates of educational attainment as well. Even if dysgenic effects are really happening at a rate of .03 IQ points per decade, it seems like any slowing effect that might have on technological advancement would be more than offset by the increasing numbers of educated people in the mix. I'd rather have 2 researchers working on a problem than one researcher who has an IQ .03 points higher.
Scott, your data is out of date - it's based on the 2019 census info and there was a recent 2022 update https://ourworldindata.org/world-population-update-2022
Fairly material change in that it projects population will now peak in 2086 rather than post 2100 (both lower growth rates and more death thanks to covid etc).
Also, there is quite a lot of noise that China has been misrepresenting its population and it's actually 180 million less than stated, and most importantly in the key segment of the fertile part of the population (with a theory that local authorities inflated their schoolchildren numbers in order to get more funding from the centre). e.g. https://www.project-syndicate.org/commentary/china-2020-census-inflates-population-figures-downplays-demographic-challenge-by-yi-fuxian-2021-08 So if true, China's population is both a lower starting point and will drop more quickly too.
All in all, this makes me update with less confidence of these projections in general. They seem to be a bit like the projections of solar price points in how they are always wrong in a single direction; and the potentially different China picture would be another major update in the lower direction. Given that much of the 'it will be OK' is coming from projections that Africa will still have huge growth rates, and how quickly other countries have seen their growth rates collapse, how much faith should we have that that will actually be the case?
There's an interesting Amish custom which is practiced in every district I've known or heard of (This is something like three districts in rural Indiana from friends and acquaintances as a teen, and another half dozen from talking to people who moved here and there.) It's called Rumspringa (which roughly translates to 'jumping around.')
See, the various restrictions that the Amish church imposes on its members only apply to members of the church and their children (where 'childhood' is held to end at 16). But Amish youth aren't expected to join the church until they're ready and willing; the whole process of joining is a bit of an ordeal to ensure commitment. You have to sit in an un-weather-controlled booth (typically in 90+ degree weather) in isolation all day for a week or more studying the scripture and memorizing key doctrinal statements by heart. You have to be discipled by a member in good standing with repeated meetings over a month or more. Your application has to be approved by your father, your mentor, and the local bishop (exact set of approvers varies based on district.)
And so far so good. If you were a clever fellow designing a church system to use human biases to keep them in the church, you might well make entering the church hard. That gives people a sunk cost to joining, which should be pretty effective at making people reluctant to leave. Combine that with various other Amish practices:
* Officially Shunning those who stray. No local Amish is allowed to speak with them or deal with them.
* Using a primary language other than English, so you're marked with an accent to the rest of the world.
* Only mandating honest dealings with other Amish, so your favorable community immediately turns hostile if you leave.
* (at least in the local district where I grew up) Kidnapping anyone who questions the church too much and taking them to a secular mental institution for brief involuntary commitment and evaluation.
And you see a system REALLY set up to retain people. But if I were designing this system and trying to get as good of Amish retention as possible, I'd probably strongly encourage kids to join the church at 16. They're now technically adults, but they're still under the roof of their parents, dependent on Amish friends and family for support, etc. It should generally be possible to apply more pressure to get them in to the church, and you've already made it so hard to exit.
The Amish don't do this. At 16 you're an adult, and you aren't expected to join the church till you're ready (usually 20-25.) That intervening 4-10 year period is called Rumspringa. In it the youth are EXPECTED to try out the world. The laws of the church don't apply, and noone acts like they should in any way. Amish kids of this age in the area I grew up usually worked at local RV factories during this time period, which is hot, high-intensity, 12 hour days for 4 days a week, and pay very well for low-skill labor. Lots of overtime, lots of non-Amish drop out because they don't have 10 years of getting accustomed to long hard days.
So, what do Amish kids do with $50k/year, working 4 days/week, with no expenses and no church? They live it up! Bars, nightclubs, sports cars, new big trucks (more popular than sports cars as a status symbol in Red America), hot girls/guys, showy clothes, etc. They spend years indulging in the 'vices' that they were forbidden as kids, and will be forbidden if they ever come back to the church.
And then they come back. Not all of them for sure, but most. I'm not entirely sure WHY; as someone who grew up English and only had some Amish friends I never felt like I fully understood why so many come back. Those who leave without ever joining the church are only subject to the usual social pressures, not the awful institutional shunning that applies to leaving church members; it's not THAT hard to leave. Some come back because they fall in love with another Amish who's more devoted than them. Some because they want to be part of the Amish community.
It's kind of puzzling to me. I wouldn't expect 'encourage your impressionable youth to indulge in all the pleasures of the flesh' to be an effective retention strategy for an ascetic cult, but it seems to work.
I'm curious what your personal approach to investing/retirement planning is, given your "2050 might not be a real/comprehensible year" view.
The defection rate for ultra Orthodox Jews is quite low - it’s only high among modern Orthodox Jews
From what I understand for ultra orthodox it’s even lower than the Amish defection rate
So if there are tens of million of orthodox defectors (which i think is too high an estimate) there will prob be just as many amish ones
"I think Japan as it currently exists is not completely interchangeable with a Japan made of 1/3 ethnic Japanese people and 2/3 ethnic Kenyans. "
The really important questions are: Can they do アニメ? What's their view on ロリエロ漫画?
Also *someone* has to make Final Fantasy XXXVIII.
On the point about "Age Pyramid Concerns Are Real, But Not Compatible With Technological Unemployment Concerns"- actually, no, those can both be problems at the same time, because not all workers are interchangeable carbon units that can be shuffled between industries at zero cost. The people who are rendered unemployable by shifts in AI technology are not necessarily going to have the skill-set or temperament to look after an increasing population of the elderly (especially at the higher end of the healthcare industry), and the people who are not being born as a consequence of collapsing fertility can't necessarily be replaced by robots unless you have full-blown AGI (which will create it's own problems, not least of which is that robots don't provide consumption.)
Also, I hate to be a cracked record on this point BUT SERIOUSLY SCOTT YOU NEED TO GO READ PETER ZEIHAN. The points above actually grossly underestimate the seriousness of fertility collapse because (A) currently prevailing international patterns of trade relating to import/consumption and export/production are dependent on demographic structures which are now going away, which would lead to mass recession even if the welfare state wasn't being bankrupted, (B) fertility collapse has been occurring for decades in many countries that do not have the wealth and expertise needed for mass automation, such as eastern europe and China, and (C) political anxieties triggered by A and B in combination with US military withdrawal from overseas are causing breakdowns in global security that trigger wars and further disrupt essential supply chains. Russia invading Ukraine (which is likely to cause a global famine) is simply the first big example.
Pretty good five-minute run-down here.
https://www.youtube.com/watch?v=rkuhWA9GdCo
(I would also say that dysgenic fertility is probably more serious than it looks, partly because we don't have a good handle on the additional impact of deleterious mutations but also because the effects are most concentrated in the younger age groups on which we depend for labour and innovation, and there's been a couple of other direct genetic studies on the subject from Ohio and the UK Biobank data. But anyway, yeah, not the most pressing immediate crisis.)
Something that can offset the decline is the possibility that technologies that will extend human lifespan a lot will be developed before the end of the century.
20 years ago, if a scientist would have said that aging is a disease and it could be possible to cure it, it would have meant the end of his career. Today, it is almost the mainstream view of the scientific community.
I believe it is possible that the first human who will live 200 years is already alive. If it’s true, having less children will not be so problematic for the countries suffering from a low fertility rate.
Another terrifying possibility is that technologies that allow the conception and growth of babies without the need of women are developed – and then used by totalitarian states like China to artificially create millions of citizens, raised and educated by the state. Parents need not apply.
Some people are already contemplating this. The technological possibility doesn’t seem that far off :
https://www.scmp.com/news/china/science/article/3165325/chinese-scientists-create-ai-nanny-look-after-babies-artificial
I think underpop as proxy for demographic fears is correct. But also, what about the David Goldman thesis, that younger countries, provided they're rich enough to afford good tech, are militarily dominant, and ageing ones definitely arent?
Does an paper like "Nanosecond protonic programmable resistors for analog deep learning" (https://www.science.org/doi/10.1126/science.abp8064) change anyone's priors on the timing of the AGI apocalypse? The MIT group are claiming about a factor of 1e4 to1e6 improvement in some types machine learning.
If you think there is a 10% chance we don't have a singularity - would you play Russian roulette with those odds? Given how bad a long running civilizational decline would be, this seems worth worrying about.
And even if you think we're going to be able to genetically engineer superbabies with astronomical IQ, will they compose a substantial fraction of the population? Will they actually occupy positions of political or cultural power, rather than being shut out by the mob? Are you so sure governments won't make genetically engineering humans illegal?
Really interesting read - thank you!
One point to note, that I don't think is addressed here, is that these projections of population growth all assume "business as usual" with our fossil-fuelled civilization. They assume no nuclear war, no severe pandemics, and no breakdown of our increasingly fragile and interconnected global economic system as the impacts of climate change intensify (best case on current projections is that there will be massive migration by 2050 caused by famines, droughts and flooding - and we are not on track for a best case scenario in terms of emissions profiles, quite the opposite). Assuming that we will stay on the rails for another 50 - or even 100 - years is dubious.
If AGI happens within 15-20 years, which seems eminently possible, then any concerns about innovation flies out the window and a bigger issue about humanity's survival will take center-stage.
Ironically, a slower innovation pace due to aging + fewer "low-hanging fruit" could paradoxically be good for humanity as it elongates every innovation cycle and perhaps even makes AGI impossible.
Whatever the downsides of that world is surely better than extinction.
> worrying about underpopulation on Mars
I mean ol' Musky is doing his best to combat that too.
Respectfully, this is not quite correct. Within a couple generations of the actual immigrants, ethnic identity is quite diluted, in large part because there's a significant degree of intermarriage between immigrant groups, and in part because you start to lose the original language spoken at home.
I think there are a few issues here:
1. Demographic shifts. The people who are unspeakable in this context are right to be worried about this. *Specifically* they are right to be worried on a racial/ethnic sense because when it comes to "overpopulation" or "I'm not having kids about the climate" or "There's nothing worse than the undisturbed sleep of a white man under the patriarchy" , it isn't sub saharan africans or redneck American conservatives that think these things, it's Nice, Affluent, White, Liberals. Everybody in every culture implicitly cares about the preservation and continuation of their culture, even as it adapts and changes. Similarly, if you're a fan of Japanese culture in any respect - you *should* want them to find a way to remain fundamentally Japanese. Presently, the only groups who are allowed to speak these worries aloud are white affluent liberals afraid that new tech workers are going to push out the Hispanic color of San Francisco. The reason to be worried is not that you fundamentally do not want immigrants or particularly hate them, the reason to be worried is that your ethnic group and culture seems *uniquely* obligated to not care about these things. It's ok to care a little bit.
2. Similarly, almost all of the comments about Britain in the 1800s apply equally to the United States today. If you're not worried about overpopulation because there will be X many researchers due to population growth in other demographics, you need to be pretty darn convinced that they will adopt a similar or better focus on research than the cultures currently at the frontier, or be inculcated into those cultures - given 1, why are you convinced that that will be the case?
3. Great powers (and conflict between them) are effectively the only places where we see these explosions of technological, moral, social, innovation. There's exactly one place looking prominent to become a great power, partially due to its large population. Allowing the CCP to take the reigns of humanity's future is equivalent to creating a suffering-maximizing AI, so we should be wanting higher birthrates in basically every country not CCP-aligned by default (even though China is going to have its own demographic period problem).
"I guess it’s still true that if innovation is destined to be only 10% of its current level in 2100, then a 30% population decline could lower that to 7%. I find it hard to worry about such a small difference, but maybe that’s a flaw in me and not in the territory."
I think this difference is worth worrying about. Not because of certain catastrophe, but because of the difference each percentage point in the possible futures we may inhabit.
My premises are that innovation helps the poor and rich, innovation saves lives, and innovation has creates compound returns over time for poor, rich, and lives saved/well-lived. And that growth creates optimism and forestalls human inflicted social upheaval and nastiness.
Imagine a small spur in innovation caused by useful but not super-strong AI, human are still doing a lot of the innovating over the first 50 years, the low-hanging fruit gets plucked, innovation starts to slow again, and there is still 3% less innovation than there otherwise would be in 2100. That's still terrible from the counterfactual perspective. 2100 might be around 50 years behind where it could be. That's not catastrophic. But 1970 at the global level was far, far, far more awful than it is today.
In that scenario it matters materially.
In the strong AI scenarios... Damned if I know.
But I am pretty sure that we should build now as we mean to continue and that includes families and trying to bring the positive value of happy, healthy people into the world. And so, the real catastrophe aren't the 8 pseudo-concerns, but that life is good. Pass it on.
Scott points out that Nigeria is the country to watch during the next 50-100 years. True, Nigeria is the most populous country in Africa, and is poised for a massive population increase; but the other African countries are also worth noticing. Here is a UN demographic forecast for 2100: The twenty most populous countries in the world. 11 are on the African continent:
India 1,516
China 1,021
Nigeria 794
USA 447
DR of Congo 379
Pakistan 352
Indonesia 306
Tanzania 304
Ethiopia 250
Uganda 214
Egypt 199
Niger 192
Brazil 190
Bangladesh 174
Angola 172
Iraq 156
Mexico 151
Kenya 142
Sudan 139
Source: UN World Population Prospects: The 2017 Revision, Key Findings and Advance
Tables. Working Paper No. ESA/P/WP/248.
Side note: In 2015, the 20 most populous countries contained 70 percent of total world population. Average population size in 2015 (194 countries) was 33 million. But median population size was only 7 million. Thus most countries on Earth have had, and will continue to have, small populations. If we can further assume that average incomes will continue to rise in the Global South, the above 20 are likely to be the most powerful states in the 22nd century.
Final reflection: God have mercy on the populations in the many small and ageing states, in today’s world as well as in tomorrow’s.
I haven't been contributing much to the comments since the move from SSC, and I hate to jump in with probably-ignorant criticism, but: the Zero Population Growth people had charts too. I saw them when they came to my high school like nine times in the eighties and nineties. And until yesterday I was told by everyone that the United States was unique among wealthy nations in that our birth rates remained so high; everyone seems to have been caught by surprise when it turns out they're low.
I guess my question is: how and why did everyone get it so wrong for so long, and what has changed to make everyone so certain that the current take is accurate? Maybe we're just bad at measuring population numbers and graphing growth projections. I remember the ZPG people being particularly concerned about explosive growth in India, China and the United States in the 21st Century. Now, whoops, everyone sees those populations declining and the population explosion is coming from sub-Saharan Africa. What happened? Why are these numbers better than the numbers thrown around thirty years ago? What makes these projections useful in a way that the last generation's were not?
Tim Dyson at London School of Economics has written an excellent book on the social, psychological and economic consequences of the ongoing global demographic transition (Population and Development, 2010). Well worth an ACX review. Some arguments from the chapter on likely psychological and social consequences:
• The vertical bonds between parents and children grow stronger
• The «planned life course» becomes a more widespread way of thinking of one’s life
• The life course of women becomes more similar to the life course of men
• Death becomes something remote & distant in most peoples’ lives
• Parents invest more in the lives of each child & time spent in education increases everywhere.
Regarding the appendix:
Some subgroups always have a higher growth rate than the general population.
As the subgroup increases in size to match the general population, the growth rate also seems to fall to match the general population.
Why are fertility rates falling everywhere? It is useful to employ the trichotomy between distal (“far from”), intermediate, and proximate ("close to”) causal factors assumed to influence the number of children born per woman, and (more specifically) to explain the worldwide trend toward lower fertility.
Distal factor:
• Global mortality decline
Intermediate factors:
• Increased predictability of individual life courses - cognitive shift to a «planned» life course
• Shift from agriculture to industry-and-services economy
• Urbanization
• Stronger states, and corresponding effective legal guarantees against interpersonal violence
• Social structure opens up to allow for skill-based upward mobility
• Mandatory education & higher percentage of young people in higher education
• Women gain access to wage labor and to career paths
• Child labor made illegal and effectively enforced
• Introduction of broad-based formal social security systems
Proximate factors:
• Delayed birth of first child & delayed marriages
• Effective contraception made available: In particular, contraception that can be administered by the woman alone, and independent of coitus
• Widespread access to risk-free abortion
Overall Scott outlines much relevant information but somehow lost in all the discussion is the biology of reproduction and the genetic evolution of Homo Sapiens
The world population increased from 1 billion in 1800 to 7.9 billion today and is headed to 11 Billion according to the UN Median Forecast.
At some point Homo Sapien population will come into equilibrium with the earths resources.
Although Homo Sapiens emerged ~300,000 years ago most human males currently alive genetically date to perhaps a common ancestor Adam ~80,000 years ago. (added note: This date was apparently pushed further back in 2013) https://www.newscientist.com/article/dn23240-the-father-of-all-men-is-340000-years-old/)
One shouldn't confuse genetics and the culture of specific small populations.
Consider the Denisovans https://en.wikipedia.org/wiki/Denisovan or Neanderthal's
https://en.wikipedia.org/wiki/Neanderthal
https://www.bbc.com/future/article/20211008-what-if-other-human-species-hadnt-died-out
Such an interesting article. I have been looking for commentary on problems similar to those discussed above. I‘m glad I stumbled across your Substack!
Maybe I'm a bit too mundane in my thinking but it seems to me the biggest concern, certainly in the short term, is how the welfare state can maintain things with the coming demographic shift. The modern welfare state only works because the young (and in some cases the yet born) are paying for the old's benefits. That adds up right now because there are currently a lot more young than old. When that pool shrinks where is the money going to come from? Just trimming benefits or pushing the retirement age back a few years isn't going to be enough. Most developed countries are running significant deficits as it is. I don't see how this wouldn't cause a massive shift in society and who is responsible for what.
Well, everyone that want to know what happens when there’s dramatic demographic shifts just need to look at Israel in the coming decades. I’m going to present some data that I’m getting from a 2017 blog post made by the Treasure Teen, he’s an Israeli that worked for a few years in the treasure ministry. I once approached him about translating those blog posts and got his blessing, but never ended up doing it. Is anyone interested in this? Might translate them if so.
Anyway, to the data he presents:
In 2015 180,000 babies were born in Israel. 100,000 of them were non hasidic jews (55%), 40,000 hasidic jews (23%) and 40,000 arab israelis (23%). According to Israel Central Bureau of Statistics by 2040, in the age group 0 to 15 non hasidic jews will make only 45%.
My own aside: from what I can tell his using data provided by the Israel Central Bureau of Statistics, the actual official agency in Israel for statistics, so rest assured his not some crank. To give you some background about current israeli demographic, in 2022 74.7% percent were jewish, 21.2% percent were arab israelis and 5% were “others”, mostly immigrants from the former USSR that aren’t technically jewish, the non jewish family of USSR immigrants and such like.
Here’s a survey of the self definition of 20 years old and up jews in 2020:
43.1% secular
21.1% traditionalist/not so religious
12.3% traditionalist-religous
11.3% religious
10.1% hasidic
Honestly I don’t know what’s the different between traditionalist and religious, but the important thing is that in 2020 self definition survey the ratio is 9 non hasidic jews to 1 hasidic jews. In the 2015 birth survey the ration is 10 non hasidic jews to 4 hasidic jews. Obviously not everyone that is born to a hasidic family stays hasidic, but to maintain current jewish demographic MOST of the hasidic babies already born will need to turn non-hasidic. I don’t have good data on how many hasidic people stop being hasidic, but I have a feeling it’s not that much.
The problem of course have to do with different rates of tax paying, and grants from the government. According to a 2011 survey, non hasidic jews paid 2,955 shekels a month through various taxes, while hasidic jews paid 705 a month. As for using grant money from the government, non hasidic jews used 1,964 shekels, and hasidic jews used 3,256 shekels a month. But the real money is in services that the government provides to children, mainly health care and education. It’s harder to get exact data on that, but if anyone is interested I’ll try to fish something out.
One less piece of data: in 2014 birth rate of hasidic jews were about 7, religious women about 4, and the rest of the jewish population hovers between 3 and 2, secular women having the lowest birth rate at about 2. I’m getting this data by looking at a graph that was presented to the knesset by it’s research facility, that’s why the numbers are so pretty.
This is getting long and I’m kind of in a rush, but ask away if you want more data and I’ll try to find it for you. If anyone is interested I might translate some economic writing I like about the subject.
A doubt about any sort of ultra-religious community somehow breeding their way into becoming the majority: I have the impression both the Amish and Ultra-Orthodox are somewhat parasitic on the rest of society. If they grow by millions and millions as projected I suspect they'll undergo major schisms, changes, et cetera. Growing a society is never that simple.
I've read long ago an article about a model that describes dynamics of means of production, consumption, and population. Basically, humans create artifacts which in turn increase carrying capacity of their environment which in turn increase population which again creates artifacts to increase carrying capacity . However each element of the system also use (consume) other elements. E.g. humans have to expend their time on creating and maintaining artificial environment and that slows down their reproduction. It's an extremely interesting framework to discuss population dynamics and economic growth.
Scott, I believe I can put your worry of long-term intelligence decline at rest.
You write:
“It’s a dysgenic argument where we assume at any given time the people with higher degrees have on average higher genetic intelligence levels. If they’re reproducing less, the genetic intelligence level of the population will decrease. There is some debate in the scientific community about whether this is happening, but as far as I can tell the people who claim it isn’t have no good refutation for the common sense argument that it has to be. The people who claim that it is make more sense, and have measured the effect in Iceland, an isolated population that it’s easy to measure genetic effects in.»
…The studies you refer, make the point that people with low education sire more children than people with high education, for the birth cohorts under study (I wish I could put the last part of that sentence in italics, but apparently that is not possible in the comments).
Here is the thing: The birth cohorts in these (and most other studies that have investigated the recent relationship between education and fertility ) capture the educational/social gradient in fertility for some very particular birth cohorts; namely the birth cohorts that were in the fertile part of their life cycle in the middle of the demographic transition.
…The point is that the demographic transition is a hierarchical diffusion process. Fertility first falls among high-status urban groups (early adopters). The last to adopt low fertility are low-status rural groups (late adopters). The pattern is the same everywhere it has been studied. (Digression: This educational gradient in fertility for countries that have not yet reached the end of the demographic transition, is the father of the saying “The rich get richer, while the poor get children”.)
The point you overlook, is that this is a transition effect. Once the transition is brought to an end, and all social groups have adopted low fertility, there is no longer any educational gradient in favor of those with low education.
…The Scandinavian countries are examples. There, we now (admittedly quite recently) see that low-educated males in particular turn out childless to a much larger extent than high-educated males. Males with high education also benefit more, fertility-wise, from serial monogamy (which is the functional equivalent to the practice of rich/successful men having more wives than poor/unsuccessful men, in countries practicing legal polygamy).
The US and Iceland are both a bit “behind the curve” when it comes to getting to the end of the demographic transition, compared to most European countries. Also, there is an interaction effect between high status and white skin in the US, which adds to the furor. But relax, Scott: Once you get to the end of the transition, high education will start to pay a fertility dividend again also in the US.
In conclusion, to channel the Critical Drinker’s standard quip in his Youtube movie reviews: “Nah, it’ll be fine”.
More than agree with all your arguments showing underpopulation is not an x-risk by any means. Not to rehash your points, but let me add a thought experiment. Let's go kind of extreme, and imagine world population shrinks all the way to 1B. Pretty drastic considering we're now nearing 8B, or at least well over 7B.
What would happen then? Obviously it's a huge change in numbers, so many things would change qualitatively along the way. We'd be in for a surprising world with lots of descendents of Nigerians, Indians and Amish, and with a lopsided demographic pyramid. What's wrong with that? Let's explore further:
Would this bring material or economic collapse? I can't see any reason why it would - since the Industrial Revolution we've learned to be hugely productive, and there's no particular reason why we would stop being so. It just doesn't take that many people's hard work anymore to provide for every 1000 humans.
Would the human spirit stagnate? Again, no reason for that - cultural renewal has never stopped in human history, why would it then? When this world had 1B people creativity was at its highest - not just industry, but also in music and the arts.
Would the downward trend be unstoppable? Yet again, no reason for that. Concerns over underpopulation are barely a few years old, and (some) people are already taking them seriously. Cultures change unpredictably in a space of decades, and I don't think there's any reasonable scenario where we get down to 1B in less than 150-200 years. Plenty of time for various groups of humans to decide to put a cultural and economic premium on breeding if they feel like it's time to stop decreasing.
Obviously I'm being as big-picture as it gets here. Would a change of this magnitude bring along unpleasant changes and instability along the way? Probably too, but then again, unpleasant changes and instability are a constant of human history, and I don't see any overriding reason why those would be worse than the ones we would just get by continuing to do the human thing with a growing population.
You know what would happen though? The sheer impact of human activity on planet Earth would be much more manageable... and this might as well be our best insurance towards long term survival.
Disagree on point 6. Not all employment is equal. It seems plausible at least that there could be a labor shortage in elder care and technological unemployment in other fields at the same time. An elder care labor shortage would happen if there are more old people, young people aren't eager to join elder care, and elder care is hard to automate. All these seem likely to me. At the same time other fields that young people are eager to join may be automated more rapidly, leading to a difficulty in finding jobs that they are willing to do.
I get pretty frustrated when people point at that Iceland study as convincing evidence. I won't deny that they do the work in the stats, but they omit all the relevant sociology and historical variables!
Let's first start with the basic flaw - using years of educational attainment as a proxy for intelligence. Ceterus paribus, that might hold. But jeepers, are other things not equal when it comes to education. I bet you'd find a strong signal in years-of-education vs birth longitude in Americans, but you'd be nuts to think it was causal.
Let's start with the meat - table s2 - https://www.pnas.org/doi/10.1073/pnas.1612113114#st02
What do you notice? Oh - the effect sizes go to basically zero for the 20year category for men (e.g doctors). But there is still a 10x stronger effect for women. So we have genetic markers that identify women that are much more likely to reach 20-years of schooling, but they have zero impact on men. So I'm supposed to believe we have a gene cluster that makes your smarter, but it stops working when men apply to med-school. Ok.
How else might we explain this? Hmm. Lets look at what's going on with women then - educational attainment over the 20c. https://www.pnas.org/doi/10.1073/pnas.1612113114#sfig06
Woah - most women don't finish more that 10 years of schooling total until WWII! And look at that scale on the left - those are individual counts not 000's. So we have (eyeballing) maybe 10 individuals in the 20-years-education category in the 1910's. All the best studies have their strongest effects in the n=10 bucket.
Let's open wiki - https://en.wikipedia.org/wiki/Education_in_Iceland#History
Hmm. Iceland didn't even have a university until 1911. So how did they even 10 women in Iceland have 20 years of education in the 1910s? What else was going on in pre-war Iceland?
https://en.wikipedia.org/wiki/History_of_Iceland#Danish_Iceland
So we have Iceland, the poorest country in Europe, which didn't normally educate women past 10th grade. It was a remote outpost of the Norwegian and then the Danish crowns, with a rich class structure inherited from nobles, land-owners, clergy, merchants, and on down through tenant farmers and thralls. And the researchers have found a genetic marker for the 10 women in 1915 with 20 years of school. This is news? I expect every one of them was an aristocrat, Danish, or kin.
Onto the "dysgenic" decline. Why do these numbers change?
https://en.wikipedia.org/wiki/Allied_occupation_of_Iceland#Outcome
Iceland was a Danish colony, and Iceland broke ties during WWII. Some of the the Danes left, and the ones who stayed mixed in. The allies arrive with a garrison of 30 thousand people against a local population of only 125 thousand total. OMG - 10 American GIs for every 5 young local women. Talk about dysgenic -- more like unhygienic.
And now everyone goes to school, not just the children of Danish administration. And a bunch of US/UK/Canadian genes get mixed it. The Danish genes have declined, sailing home to Denmark inside living Danes. This is not a mystery that demands explanation. This is yet another echo of the the 20c stories of decolonization and global war.
I'm not suggesting that my just-so story refutes the theory categorically, but I blanch when ever someone references this study as some kind of slam dunk. It makes you sound like Daniel Kahneman before the waves of the replication crisis reached his toes.
The technological singularity concerns are real, and to me what is most likely to cause it is the conflict between the US and China/ China's allies. As long as there is serious international competition for power, we simply can't slow down. You mentioned technological change slowing down, but I worry that it's not slowing down fast enough. This AI stuff is real, and it's coming for us all if we don't actually organize and regulate AI (and biological engineering, chemical weapons, etc). We can't be libertarian on AI unless we're suicidal. And it won't matter if we officially regulate AI if powerhouse countries with massive spy agencies/ research departments (like, say, the US and China) are trying to use whatever new toy they can to beat the other side (even if that new toy, is, say microscopic death robots that mindlessly wipe us out then expand across the galaxy).
So I think the question all the clever people on here should be asking is simply "how the hell do you get the US and China on the same page before the technological singularity?" Because a nuclear war is the tip of the iceberg when it comes to potential things that could go REALLY wrong here. We know how strong the tribal competitive instinct is in humans. We know how dangerous potential technologies are. And still, we hurdle towards the brink. We need something better than the United Nations, and we need it ASAP. That's the only reasonable path forward I see. Agreed, 2050 doesn't even feel like a real year. And still no one seems to be tackling these issues seriously. It's like we've all just given up. Glad at least you're writing a few posts on it before it's over Scott-
“I guess it’s still true that if innovation is destined to be only 10% of its current level in 2100, then a 30% population decline could lower that to 7%.”
You can’t do the math like that. Speed of progress is not so much about the absolute number of people. It’s about expectations.
When you anticipate a big population increase the pressure is high to build foundational tech. Investors give you money. Banks give you loans. All in anticipation that there is surging demand.
I think you miss the danger of the sclerotization of society. New people aren’t just any people - they’re young people. More open to new ideas. More prone to go into novel fields or set out to test novel hypothesis. Not only that but there is some decrease in openness to new ideas an increasingly octogenarian environment (I recall some study found decreased patents in universities with older professors or some such). The human population won’t be falling until 2100 in large part because life expectancy will be increasing. Japans population grown by 10,000,000s since 1960, but the number of 0-14 yo children has nearly halved (28 million to 15 million)! The dangerous demographic changes concealed by topline numbers isn’t that people will be less intelligent or more diverse - the danger is a world without young people is a world without early adapters of new ideas.
I have also 2 more spiritual arguments that I’m making for growing the population :
The 1st argument is that human consciousness is, as far as we know, the highest form of consciousness (or one of the highest forms of consciousness) on the planet, probably in the entire solar system, maybe in the entire galaxy, or even in the universe.
So every time we make a baby, we basically transform solar energy into a very high form of consciousness. We transform an inert part of the universe into a part of the universe that can observe and understand itself.
It’s a miracle.
On a more prosaic level, most of us are glad to be alive, and glad that our parents made us, therefore we can presume that most of our children will be glad they we had them, as will be the children of their children, etc. for millions of generations.
Every couple that doesn’t have kids is effectively preventing probably millions of people from existing in the future, across many generations.
My 2nd argument is on a more personal level: every one of us, without exception, is at the end of an *uninterrupted* chain of at least 3.7 billion years of life evolution (and maybe more, since it’s possible life was brought on Earth from elsewhere https://en.wikipedia.org/wiki/Panspermia ).
For 3.7 billion years (that’s 3 700 000 000 years), every one of our ancestors managed to survive and reproduce, without any exception.
Not having kids means we would be the very first in this chain to break it.
I’m arguing that we should have respect, and even be in awe, for this long chain, and think carefully if we really want to be the first one in almost 4 billion years to break it.
A few points:
- one omission from the Amish Appendix: the Mormons. They don't have the fertility rate of the Amish (and it is going down), but it is high and they start from a much higher point.
For an illustration of extrapolating to 2100: https://twitter.com/akhivae/status/1448392869698494472
'Primitives' like Amish, Mormons, orthodox Jews, Nigerians, Afghanis multiply while 'advanced' effective altruist rationalists wither.
- nothing in the physical world keeps doubling indefinitely. Not Covid cases, not the Amish. So the Amish population will eventually peter out before 2250, too.
- fertility rate among the Puritans in the 16th and 17th century was Amish-high at 7+. It came down.
1) Let's just go ahead and say the world isn't going to end, because if it is why discuss any of this.
2) I agree that "genetically engineered babies" would be a great thing to have. What measures would increase that possibility? I think the best methods would be:
A) Encourage fertility amongst those likely to produce the offspring that would invent such things.
B) Not do anything that would allow the societies likely to produce such outcomes to seriously degrade/collapse.
It seems to me that aging societies being flooded with low IQ immigrants are less likely to produce such an outcome than fertile societies without low IQ immigrants. Since genetic engineering will fix the potential immigrants low IQ problem, we should encourage such an outcome, even if present low IQ potential immigrants have to remain in their original counties.
What's the IQ of the West if all those Nigerians immigrate to the west? 97.5 seems very optimistic.
3) I think you misunderstand the worry with demographic pyramids. We have not been able to make medical care more efficient or labor saving for a very long time now. All those pensioners aren't going to need mass produced wheat or widgets. They are going to want the same overly expensive economy consuming medical services they want today. What happens when that is most of the economy? Is there anything left for human flourishing and innovation?
4) It's unclear to me what we are supposed to be more concerned about then long term effects? Like am I supposed to expend my mental energy on the latest Current Thing? When I think about more immediate problems we face, not a one of them would be alleviated by dysgenic trends (whether births or immigration).
Nearly every opinion I have on how to make a society better is shared by pro-fertile people, and nearly everything I think would make it worse is shared by anti-fertility people.
P.S. Having kids is good. Not having kids makes our life worse and makes you an objectively worse person.
Regarding dysgenics, isn’t the whole notion kind of presumptuous? Especially in the context of a large economically unequal society?
I picture a nobleman in 1600s England bemoaning that the lower classes breeding too much will cause average intelligence to decrease, ignoring the fact that 99% of the population does not have the chance to prove themselves “intelligent” by any officially accepted standards and that his own station at birth provides opportunities for higher education that the vast majority lacks. A farmhand might have been a professor (and might posses the genetic ability to have been so which he will pass on) given some different opportunities in life, better nutrition in infancy, a realistic path to literacy and higher education, etc.
Of course modern America isn’t as stark an example as 1600s Europe but we definitely do not live in a society where raw gifts of intelligence or talent at birth guarantee a rise to higher educational attainment and it feels naive and eugenics-y to think that we do.
Maybe I’m misrepresenting the argument and I’m willing to learn more, but as far as I see it Idiocracy is not a reliable way to think about the future of the world.
I have heard before the idea that it would be terrible to make a baby who might live in a troubled world, but I don't understand it. There's some assumption behind it that goes against some assumption of mine that's blocking me. It might relate to whether existence is good in itself (me) or neutral till you add pleasure or pain (?). Or something about how much we trust our ability to predict the future -- like in a Ray Bradbury story in which the world was going to end so two parents poisoned their children the night before so they wouldn't have to live through it, and then of course the world didn't end and everything was OK except their kids were dead. I wouldn't bet 10c on any particular outcome at this point.
If I thought AI was going to take over the world in 30 years and kill everybody, I'd still be glad my kids exist. They'd get 30 years of life before then (I see Scott has the same answer there). And maybe one of them could do something about it.
"I understand why people don’t want to talk about the issue this way, because if you say demographic shift is a problem, people will call you a racist conspiracy theorist.
I find this instinct a bit weird as well, since it seems to be a matter of scope and only to be applied selectivly. In many cases we assume a desire for exclusivity and raising boundries to be a sign of love or care. For example when a person gets married we accept that this is a sign of love for that person, not proof for hate of all other people. Exclusivity in regards to family and wider community is also often interpreted as a sign of care for ones community or family, not animosity towards all other families and communites. But with regards to nations the benefit of doubt seems to reverse.
And just to be clear: there is plenty of evidence (like hateful rethoric) to make the judgement that at least some of the desire for boundries on nation-states is based on hate. But still, how confident should we be in mind-reading this to be the main motive? Would suck if we were wrong about something this important, since if there are good motives on both sides we ought to take quite a different approach than if it´s a struggle between good and evil.
You're far too worried about AI.
Present methods will never generate AGI and there is a lot of focus on machine learning because it actually generates results that people can use, at least to some extent; the other approaches are not yielding results at the kinds of speed people want because this is an extraordinarily difficult problem.
Machine learning is very reminiscent of Clever Hans.
If you are just messing around with these programs, it's very easy to get impressed.
But the more specific you make your demands - for example, with these image creation programs, the way I would when I was commissioning an actual artist, where I have something specific in mind - the more obvious it is that it is very much Clever Hans.
It can spit out a lot of pretty stuff - but it can't generate what I am asking for.
Moreover, it is very obvious when I actually have something specific in mind that it isn't actually even trying to generate what I'm asking for half the time, but is instead putting out something pretty. And I can tell this because I am trying to put this stuff to practical use and therefore have specific unchanging intentions.
This is not the sort of thing people notice with testing, because their test cases are not specific enough, and often aren't actually probing for problems. I see people do many of the same tests when experimenting with the program, and then see actual users trying to use these programs and it is very obvious what is going on when people are trying to put these programs to practical use is not actually matching the test cases very well.
If your goal is some broad request, sometimes it will give you a good result - ask, say, for a magical orb swirling with the magic of spring, or for a trail through a magical forest, and it will give you some very pretty things that are 100% usable. And depending on the AI these might be pretty good.
But if you have something very specific in mind visually, these programs cannot do it. They will often come up with visually interesting things - and I see people being happy with things that are produced all the time - but when I am trying to generate one specific thing that I actually already have in mind, it can't do it.
And this is the Clever Hans. These programs don't need to be able to understand English to generate images that will pass muster with most people.\
It's very obvious to me the more I use them that this is what is really going on, and it isn't really solvable because the approach is fundamentally flawed.
Worse, you can't just throw them at mobs of people for rating because people will upvote things that are visually pleasing regardless of how well they fit the prompt.
They will also completely fail at many requests and I think this is too easy to disregard as the model being underpowered and not that machine learning has some fundamental limitations and points of attraction that the models end up getting sucked towards because they aren't actually intelligent.
It's true of all of these approaches, because these approaches aren't designed to generate intelligent output.
And the scaling is terrible. I was listening in on a talk by the guy who is behind MidJourney, and it was very interesting, as he pointed out that at the present rate of growth (which he doesn't expect will happen for obvious reasons, but still), MidJourney would be using every cloud computer in five months.
And this is a fairly bad AI in a macroscopic sense.
This approach is both insanely expensive and not going the way people think it is.
Also, speaking as someone with knowledge of biomedical engineering and genetics:
We're a long ways off from understanding the genetic basis of human intelligence. Intelligence is an extremely polygenic trait, and the more genes are involved in a trait, the harder it is to actually make a super-whatever (genius, in this case).
Moreover, without very unethical human genetic engineering, doing experiments on this is extremely hard which makes it even harder to make progress.
While I expect us to have made SOME progress by 2100 thanks to large genetic databases, I would not expect the majority of children to be hyperintelligent by then even in the developed world, and in fact, I don't expect we'll be generating super-geniuses by then.
This is an extraordinarily difficult problem to solve. Computers have made it possible that we may eventually work it out, but 50 years (which would put the supergeniuses at adulthood by 2100) is pushing it, and even in 100 years I think there's better than 50% odds we won't be able to generate people of IQ 150+ consistently (though I would expect us to be able to boost intelligence at least somewhat by then).
The biggest threats aren't human genetic engineering or AIs, it's biowarfare and Russian and Chinese aggression, along with attacks on democracy.
6. Age Pyramid Concerns Are Real, But Not Compatible With Technological Unemployment Concerns
Likewise, I think population decline concerns are not compatible with global warming concerns. If each human creates a carbon footprint, then the subtraction of every human moves us closer to stopping global warming. If you really believe the latter is going to destroy the planet, then you should be happy about human population decline or at least neutral about it.
If you don't view human overpopulation and overconsumption as the #1 problem for the entire planet, I think it is because you cannot imagine giving up your own overconsumption -- hence, denial. Worry about underpopulation is simply insanity.
Rodes.pub/OneBillion
* The optimum human population for the planet is one billion
* Population decline will begin around 2050 with a peak around ten billion
* Living conditions will deteriorate in this century causing an even faster decline in fertility
* A population of one billion will be reached around the year 2300
yeah so uk is just importing low iq ppl is not gonna make life better. quantity ≠ quality
The U.S.A. had a spike in births (after a long general decline for over a decade) in 2021 that appears to be lead by college educated women working remotely.
https://ifstudies.org/blog/is-remote-work-behind-the-spike-in-us-birth-rates
I