There was a post/plug a while back about a high school analytics summer program, maybe a predictive analytics program or fellowship. I think it was at an institute associated with Berkley. I can’t find it. Does the hive mind remember? Thanks in advance!
I was surprised in my thread below "Can an AI grow bored?" at the objections by those who consider AI x-risk to be high to future AI experiencing anything like emotions. I would have expected the opposite. I want to ask a slightly different question: Do you believe AI-x risk to be high AND do you/don't you believe advanced AIs will have qualia?
People don't have a good sense of consciousness, even to the extent that anyone does (which isn't much of an extent). They're strongly anti-dualist, but then also often think that only neurons can produce consciousness somehow. But even though I acknoledge this kind of thing, I still struggle to get myself to believe that AI will be conscious.
At the very least, I think it's reasonable to believe that if AIs do become conscious, they will do so in a way entirely alien to our own kind of consciousness. Unless they emerge through human brain emulation, their cognitive processes would be so different to ours that I don't think there's any reason to believe they will have things comparable to human emotions.
Things like emotions seem to be a product of how our brains are designed and the nature of our non-conscious psychology. We experience the mental state of a particular emotion seemingly because we evolved to have this state to increase fitness e.g. we experience the mental state of anger to motivate actions that are aggressive, threatening etc. to reduce threats to safety, resources, social esteem. It's almost certainly the case that we *feel* anger because we have the mental state of being angry (otherwise it would be some grand coincidence that our mental states and conscious experiences are so closely aligned).
If a cognitive system lacks the same architecture as the human brain, I don't see any reason why the conscious experiences it would have would closely match our own at all. And unless it was specifically designed to mimic the human brain, I don't see why it would share that architecture. If that's the only way to get an advanced AI, then we can be fairly certain AI is not going to happen any time remotely soon.
It also raises the issue of the function of consciousness and epiphenomenalism. It seems like all an AI would need to work are it's physical hardware and associated physical phenomena - voltages on a silicone chip, basically.
We don't really have trouble explaining why computers work - the movement of electrons and so on explains it all. So if an AI was having conscious experiences of boredom, it's hard to conceptualize this in physical terms. It seems like a particular movement of electrons on a chip would be needed to give rise to conscious boredom (if that's actually possible) - but in order for the experience itself to actually DO anything, the experience needs to afect the behavior of electrons (or whatever physical thing is responsible for computation.
On the one hand, if the experience isn't doing anything itself, then it's irrelevant and all we're really interested in is a certain pattern of computation that puts the AI in the 'mental state' of 'boredom' (i.e. such a condition that results in the AI behaving in a way that we would call it 'bored' of a certain task, such as avoiding certain actions despite the weight that a model would assign to those things), so we haven't really acheived anything by bringing in the issue of conscious experience.
On the other hand, if we are going to say that these experiences are themselves going to impact the physical things happeneing on the hardware, then this is very problematic, because its unclear how this could possibly work. It implies that we could (at least in principle) observe a system and its physical behavior would diverge from that we would predict based on physical laws (unless you're some sort of panpsychic who believes consciousness in some basic form is already responsible for physical laws). This of course isn't actually unique to AI systems and cannot really be conceptually explained for humans (that is, how a materialist account of consciousness avoids epipohenomenalism at the level of explaining how 'qualia' themselves can possibly affect behavior - pointing out the correlation between qualia and speech acts doesn't even suggest the most basic, speculative mechanisms that could possibly be responsible).
All in all, I think we're not even close to being able to even conceptualize this stuff to begin with to think we can understand how it could affect AI systems of the future.
I tend to agree - if you think the risk of AI eliminating humanity is high, then naturally you would probably think barriers between AI and HI are low, so qualia would be less likely beyond AI.
Maybe it's just that everyone has such a strong sense of AIs as empty robots that even if some people are more likely to attribute the ability to experience qualia to AI, they still don't think AI will have qualia.
Has anyone tried to use an LLM to replicate a specific individual, someone for whom there was massive amounts of recorded speech or writing? It would be very little relative to the bodies of text that LLM's are trained on but more nearly consistent, which might help a little.
Do we have any good theories as to why a number of ancient societies considered human sacrifice abhorrent, despite tolerating many other forms of violence that seem just as barbaric by modern standards? It's unsurprising for the Jews in Biblical times; in a society where monotheism is considered a founding principle, it makes sense that sacrifices made to pagan gods would be seen as worse than regular killings. But the distinction also seems to have developed in pagan Ancient Greece and Rome, presumably independently of its development among the Jews, so there must be more to it than that.
What's to explain? Conservatives kill convicts but not unwanted fetuses; liberals kill unwanted fetuses but not convicts. Westerners obsess about animal cruelty and eat semi intelligent pigs, but not bugs; Muslims obsess about purity, and eat locusts. Culture is like that - you don't design it from first principles, you just do what everyone around you is doing, and then make a living as a blogger or comedian saying, "Isn't that weird? That's so weird!"
Yes, and then some people grow disappointed in their native culture and look for answers elsewhere, often falling into the “culture I know nothing about must have the answers” trap.
The only types of violence that I can think of from ancient times that were "tolerated" would be ones we might classify as acts of the state, such as war or capital punishment. Of the societies I can think of where human sacrifice was permitted or encouraged, it appears to have been state-sanctioned as well, not individuals taking it upon themselves. Murder committed individually seems to be universally against the law even in ancient civilizations. Assaults may have been more common or not result in criminal liability, but I think that was a matter of lack of enforcement power instead of toleration for it. The Bible has a number of laws against harming people in the OT, for instance, as do other ancient law codes.
Perhaps you could elaborate on which practices were commonly accepted that today we would consider barbaric? Slavery might fit, but we might get ourselves down a rabbit hole of whether or not slavery exists today (it does) and is acceptable to "modern standards" - which gets murky depending on your definition of slavery and who's doing it.
What other forms of violence are you thinking of? The distinction that comes to mind is that, in turning the death into an offering to the gods, the act of killing becomes the reward. As opposed to more normal crimes where you kill someone because you want their stuff or just hate them, where the killing is in furtherance of a different goal, ritual sacrifice has no natural endpoint, because you're doing it for the sake of doing it.
Yes, probably since the change to the "astralcodexten.com" domain, my company shows an "Are you sure you want to visit this site?" page, "Request method cautioned for category Miscellaneous or Unknown". But these warnings are so common, I usually ignore them. Which defeats the purpose of the whole security circus.
Yeah sadly I don't have the autonomy to click through. The balance of hassle of reading on my phone vs emailing the help desk is currently tilted towards don't engage.
Is anyone here able to confirm the results of Ecuador's parliamentary election on August 20 and/or Zimbabwe's senate and national assembly elections on August 23? I have created markets about both but haven't been able to properly resolve either. (Sure, the results have been reported on English Wikipedia, but being a Wikipedian myself, I am inclined to distrust the wiki and trust the source it cites instead.)
My employer is asking everybody to sign a new contract that includes a non-compete clause and a 2 year ban on hiring employees after we leave. While I don't expect either of these clauses to affect me in the future, I am insulted at the company's attitude towards its employees. I could perhaps redline the contract, but that does nothing for anybody else at the company or future employees.
Basically, I and a couple of others have decided we will not sign.
I'm curious if anybody else has had this experience and what happened as a result. I'm not sure whether I should expect to get fired (the company is not doing well and they may see this as an opportunity to get rid of people), but I'm accepting that possibility.
Depending on state law in your location, they may not be able to request that you sign this without additional compensation. I know Texas forbids that, for instance. Some states would allow this upon initial hire or in exchange for a promotion/raise, but not just because.
Whether such an agreement would hold up in court is also very much in question. Courts often frown on mid-level or lower employees, or even anyone not specifically holding company secrets as a requirement of their job, being held by non-competes. Courts aren't big fans of denying someone future employment, and even in cases where it should apply for good reason, courts are reluctant to enforce it. That would especially be true if the company laid you off or fired you, but even for employees who leave under normal circumstances. This is not universal, but it seems that courts only really want to enforce these when someone leaves a company for the purpose of taking the company secrets with them.
A company I worked for asked me to sign such a thing, including something saying that it was already covered in my compensation. I didn't sign. Later, they sent me something saying they noticed I didn't sign. I ignored it. Nothing happened.
I'm not a lawyer, but I advise you not to sign. Upside to signing? None.
If you don't like it, don't sign it. (Say that you need to ask your lawyer. Or say that your lawyer advised you against signing.) Or maybe ask what's in it for you -- by signing that paper, the company gets some *extra* value compared to the existing situation (otherwise, why would they ask you to sign), so what *extra* value are you getting if you sign? (Perhaps a huge bonus to compensate you for potential *two years* of unemployment?) If the answer is "nothing", it means they want to get something for nothing, so just refuse.
Depending on where you work, yes, you might get fired tomorrow. Note that this is true regardless of whether you sign or not! Maybe the company is already planning to fire you and many of your colleagues, and they just want to make sure that you will not run en masse to their competitor. Ask yourself how much would signing the contract and actually following it hurt your chances at job hunting during the next 2 years. Do you want that?
Most likely, the contract is illegal. (Verify this with a local lawyer.) It is still a bad idea to sign it, for multiple reasons. First, the company may decide to give you legal problems anyway... and even if you ultimately win, it will cost you a lot of time and money. (As they say, the process is the punishment.) Yes, it would cost your company some money too, but for them it's just regular expenses, and perhaps they feel that the deterring effect makes it worth it. Second, the new job will probably ask you for references; and if they contact your old company, the company may tell them that you signed a non-compete. The potential new employer might decide that the potential legal hassle is not worth it, and they may decide to simply hire someone else instead. Even if the new employer decides to accept the legal risk, you kinda start your new job known as "someone who signs contracts he doesn't mean to actually follow". People may or may not care about this.
I think this is exactly one of those situations the unions were invented for. Of course, many people are too proud to join a union. This may be the moment to reflect on your preferences. If you had a union, you could simply ask your union lawyer, but quite likely the company wouldn't even try this, because the reaction would be predictable. You wouldn't have to ask strangers on internet for help.
I don't have a similar experience, because in my country the employers usually don't even try this. My contracts only contain a clause that I cannot work for a competitor *during* my current employment. Even that only applies to competitors; I am in general allowed to have an unrelated side job (in case that I am not sure what qualifies as unrelated, I can simply ask HR). Once the employment is over, it is over, full stop, the former employer has no power over me (except for the NDAs I have signed).
All of this, especially the business about consulting a lawyer, which I'll address a bit more.
A lot of people are instinctively resistant to the idea of using a lawyer; it often feels like a thing that's just for really rich people (or real assholes), or only for truly DIRE situations like being permanently maimed in a car accident. Lawyers are a thing regular people just "don't do."
But that's not true!
While legal expenses can get really expensive really fast, a consultation about your circumstances and maybe a threatening letter on legal letterhead is often *very* affordable - often just a couple hundred dollars. I was certainly a "I don't hire lawyers" person until circumstances with my HOA's truly egregious failures forced my hand - and the moment my lawyer got involved things turned around almost miraculously. I wish I'd gotten him involved right from the start, I would have spared myself tremendous hassle and anxiety.
Think of consulting a lawyer as a kind of insurance against having a tremendous pain in the ass - $200-$500 or whatever can spare you *YEARS* of uncertainty, anxiety, lost income, etc.
I don't mean to be flippant about a few hundred dollars - that's a lot of money - but one of my favorite sayings is, "the cheapest way to pay for something is with money," and in your case, it sounds like the cheapest way to know what to do is to pay a lawyer to tell you.
@Snags, you really do need a lawyer for this. Find one on Avvo or whatever and see what they think!
Small-town probably doesn't apply much, but I worked at a company started during the owner's Non-Compete duration with their previous employer, and nothing came of it. Although their job had also been licensed by the state.
More information today in the NYT about new studies confirming that college, particularly if you have to pay a steep price to get there, and particularly if you don't manage to finish, isn't the slam dunk investment it once was.
> Start here: If your tuition is free and you can be absolutely certain that you’re going to graduate within six years, then you enter college with a 96 percent chance that your gamble is going to pay off, meaning that your lifetime earnings will be greater than those of a typical high school graduate.
> The problem, though, is that many students who start college don’t graduate — about 40 percent of them, by one estimate. When Webber factors in that risk, your chances of coming out ahead of the typical high school grad start to shrink. If tuition is still free, you now have about a 3 in 4 chance of winning the bet.
> The second problem is that going to college isn’t free. If you’re paying $25,000 a year in tuition and expenses, Webber calculated, your chance of coming out ahead drops to about 2 in 3. At $50,000 a year in college costs, your odds are no better than a coin flip: Maybe you’ll wind up with more than the typical high school grad, but you’re just as likely to wind up with less.
I'm confused. I just checked Harvard and MIT tuition, and it's $56000. So, if you're paying $50,000 a year, you're going to an elite college, no? Is the claim really "50% of the people who were admitted into, and at least started attending, an elite college don't end up earning more than a typical high school grad without a college degree"?
"College costs" includes room & board, which is almost certainly a five-figure number and the first digit may not be a one. Also fees, textbooks, travel expenses, etc.
Discounts are more than "pretty common", they are routine for all but the wealthiest households. My wife does college counseling fulltime for middle-class and upper-middle-class U.S. (mostly) families, and she and her colleagues don't even bother with sticker prices anymore when working with their clients. They just roll their eyes at the USNews list. They use average net prices actually paid at each of the various colleges and help the families use that for comparison and/or negotiation purposes.
Also, for several years now the sticker prices have been declining relative to inflation and the average actual amounts paid have been declining absolutely.
This Brookings report from earlier this year lays it out pretty well:
At the private schools that have large endowments this difference is even greater. That Brookings report shows that a student from a family with $125,000 in household income can now expect to pay $32,800/yr to attend a private college with a large endowment, which is well under half of the quoted sticker prices at those schools. A student from a household income of $75,000 now pays $17,700 on average at those schools.
Drexel specifically now has a 50% across the board discount for community college graduates, so if you have a two year degree you can cut that sticker price in half right off the bat.
That sort of thing is spreading now particularly among U.S. colleges outside of the "highly-selective" tier. It's about supply and demand: the college-age population stopped growing some years ago, is now declining, and higher-ed administrators think the decline will accelerate a year or two from now. They talk about that last part a lot and refer to it as "the enrollment cliff".
Colleges with Ivy League-type reputations have compensated by pursuing more international students, with the additional benefit that those students pay full list price if they do enroll. However that supply has also stopped growing for various reasons, and in any case isn't a strong competitive prospect for schools like Drexel. So in addition to the general flattening in the cost of college in the US, incentives like you just described are spreading.
The expectation in the industry is that the bottom end will get hit the hardest. The prestigious private universities and the state-funded big schools will be fine. The places in danger are small primarily-undergraduate schools with only local reputations.
This is at least consistent with a Caplanian understanding of education, as say, 60 years ago, college students were more intellectually distinctive than today as a higher and higher proportion of the population started going to college.
His hypothesis is that because GPT does a mediocre job at many things, it will be a great boost for people with sub-mediocre skills, but will almost nothing for the best ones, so it will make people more equal.
This is an interesting perspective, and GPT will definitely shift the learning curve in weird ways, but I also see some problems with the optimism about the "normies".
First, even if GPT will hold your hand during your job, you will still need some minimal skills, such as "asking the right questions" and "noticing that you need to ask". Without these skills, GPT won't help you much. For example, as a junior software developer, you can ask GPT to write you the code, and then you can also ask GPT to rewrite it to make it thread-safe. But it's you who needs to notice that the first generated code isn't thread-safe and that it should be. Similarly, as a lawyer or a doctor, you can give GPT some information and let it figure out what to do. but you need to make a judgment about what information is relevant to the case. Then there is also the problem of detecting hallucinations.
Second, a part of senior's job is to manage and advise juniors. But if you are using GPT anyway, and you know that the juniors are going to ask GPT, then perhaps you might skip the intermediary and ask GPT yourself. If GPT can help juniors more than it can help seniors, that also means it can replace the juniors more easily than it can replace seniors. The result could be that on one hand, yes, a normie + GPT could do a mediocre job, but on the other hand, an expert with 5 open tabs of GPT is more effective than the same expert supervising 4 normies, each using GPT. Maybe not. We will see.
As I see it, a great advantage of GPT is that it can make *learning* much faster. Buy (or download) a textbook + ask GPT the additional questions = it's like you have a private tutor for any subject. That said, the schools are already more about credentials than about learning, so unless we somehow decouple credentials from school attendance, it's not going to change much. We will have more educated people, but without the paper to prove it. Also, although GPT will provide great learning opportunities for normies, I don't really expect them to actually use this resource.
What are some common examples of tasks that AI accomplishes with >99.9% accuracy in the wild, ideally without near-infinite task-specific training data? By AI here I mean something like neural networks, i.e. something that's trained by giving it lots of examples and asking it to generalize, as opposed to coding an explicit algorithm.
Background: the company I work for makes optical inspection systems, meaning we collect high-resolution photographs, do some processing, and then rate the components in the images on whether they're defective or not. These components are tiny -- there can easily be >10,000 features to measure in a single image -- and the error rates our customers are looking for are certainly less than "every image contains a part that is incorrectly flagged as exceeding tolerances". The company has been interested in leveraging neural networks more for some of the image processing tasks, and I'm trying to understand whether that's all realistic. I went to look up some benchmarks and saw this page documenting progress on the CIFAR-10 dataset: https://paperswithcode.com/sota/image-classification-on-cifar-10. CIFAR-10 has 60,000 images, and the best model has a 99.6% classification accuracy, meaning it misclassifies 233 of the images; this feels like it would not be good enough. On the one hand, we might benefit from having a more restrictive dataset (the components don't vary that much in appearance and are being imaged in a controlled environment; we're not going train-spotting in the wild). On the other hand, I feel like results I've seen on transfer learning have been of the form "see, with a generic model and a small amount of targeted data / time, we can achieve performance that isn't crap" rather than "...we can achieve performance that's better than anyone has ever done on CIFAR-10".
I've used some ML tools that use OCR to identify text in images, and then to either extract useful data from it or to identify what kind of document it is (called classification). Both can be highly accurate, but neither is in all circumstances.
For example, a 1040 tax return is notoriously difficult, both to classify (lots of similar-looking documents) and to extract data (lots of variants, tax year differences, etc.). On the other hand, we had some classification documents that were always accurate with our model, because they were so unique: for example, even if the document had a confidence level of 10 it was still the top document type selected in the model, and it was ALWAYS right if it got that high.
Document extraction is unlikely to get that high, simply because of image quality. But if you have consistent documents of consistently high image quality then it could theoretically. But then, you asked for examples in the wild.
This is actually about the right level of wildness, thank you! (We make the imaging system, so we control the image quality.) On the other hand, if I think about this as e.g. classifying the components in the image, it sounds like what I can expect is "some but not all types are always identified correctly".
Depending on the document, it may be possible to automatically double-check some of its conclusions. For example, we needed to get beginning and ending balances from bank statements. We could double-check what we found in a couple ways: finding the same figure somewhere else, and/or taking the beginning balance and adding all of the activity found and seeing if the result matched the ending balance. In either case, one could send the failures to a human for extracting the data. The calculation, if the statement had any actual activity, was never found to be wrong.
Of course, nothing is foolproof, and you would have to have a very select group of documents to all be highly accurate to classify.
At that level of accuracy, none I can think of. Having said that, this is probably not the right way to think about ML models in practice. Most organizations I've seen can get some effect from ML models but it needs to be integrated with existing workflows and typically is used to filter out obviously wrong cases so staff can focus on difficult cases.
For example...let's talk about, uh, sepsis screenings. Sepsis is complicated but the guts of it is that sometimes in a hospital, people's blood gets yucky and then maybe they die. This is bad and makes doctors and families sad. The Epic EMR, amongst other sources, sells ML algorithms that will predict whether patients will get sepsis within a relatively short time frame, think <24 hours.
But, in actual workflows, nobody gives a suite of powerful anti-sepsis drugs just because a computer said so. What tends to happen is:
Step 1: Out of 100 people "at risk" patients, the ML algorithm identifies 25 people who will get sepsis.
Step 2: The ML algorithm calls a nurse.
Step 3: The nurse sends 15 people home, they don't get sepsis, gives the anti-sepsis drug to 9 people, and is confused on one patient.
Step 4: Nurse calls a doctor for the confusing patient.
Step 5: Doctor decides whether to give the confusing patient medicine.
Now this is pretty simplified (and idealized) but that's the flow, at each step actionable or confusing cases are kicked up the chain of command. The ML algorithm doesn't need to be 99.9% good, it honestly doesn't even need to be 80% good (Epic isn't) but, ya know, algorithms are cheap to run and if the algorithm gets it wrong 30% of the time, the nurse gets it wrong 20% of the time, and the doctor gets it wrong 10% of the time, naively, you get 0.3*0.2*0.1=0.006% chance of failure vs 0.2*0.1=0.02=2% chance of failure and a lot less work for nurses and doctors.
So, uh, a lot probably depends less on their accuracy and more what your current workflow and tools are. As a general rule, if you have people or expensive tools that spend a lot of time on things that "obviously" issues, it's probably a good candidate for ML. But complete replacement...meh.
I agree with the general picture, but I think that neural nets may be a less good choice than other, more comprehensible forms of classifier, for the first step.
Triage like this is about classifying things as "yes", "no" or "maybe", and it's very important that no "yeses" or "nos" get misidentified as the other, so you want to be quite conservative about saying "maybe"
If you care equally about errors in all directions, neural nets may well outperform simpler things and give a better ROC curve, but their tendency to occasionally throw up completely random and unpredictable output means that they're possibly less suited for being conservative in any given direction, and some simple low-dimensional human-comprehensible function of the inputs may be able to make you more confident that you're only going to err in directions you don't care about so much.
I think right now the company is trying to figure out how to integrate ML into existing algorithmic workflows, and struggling with some version of "if it will improve the performance on 80% of the cases, but fail badly on the remaining 20%, how do we work around that". Obviously if we can tell whether it succeeded, we can just fall back on whatever we're doing now, but that requires being able to tell that it succeeded with a high degree of accuracy.
I read Joshua Cohen’s “The Netanyahus” last week. The novel itself is a fictionalized account of Harold Bloom’s encounter with Benzion Netanyahu’s family in the late 50’s.
The Credits and Extra Credits section after the body of the novel was a special treat. Cohen relates some of his conversations with Bloom and has entertaining anecdotes about the literary figures that Bloom had a chance to work with during his career. Saul Below, Phillip Roth, Susan Sontag, Nabokov, Don DeLillo and Cormac McCarthy are all mentioned.
In that section of the book Cohen also taught me a word I didn’t know. ‘Illeism’, the practice of referring to oneself in the third person ie “No one had done more for Israel than Donald Trump.”
A bit of research turned up a number of notable illeists:
Mr T, The Incredible Hulk, Dr Doom, Charles DeGaulle and Bob Dole during his 1988 run for the presidency. As hilariously charicatured by Dan Aykroyd in an SNL Cold Open.
“Bob Dole grew up in Kansas. Bob Dole didn’t have the the prep school education, the bumper pool table in the basement, the convertible at graduation, the electric steak knife, the machine that throws the tennis balls at you….
I will pay $100/hour for up to three hours of work to anyone who manages to successfully get this blog connected to astralcodexten.com.
Substack has a process for doing this but it's said it's "connecting" and "will take up to 36 hours" for the past few weeks.
I think there is probably some sense in which a domain transfer may have partly failed, and even though my HostGator account says I own the domain I really don't. The person who does own the domain wants to transfer it to me and I thought we transferred it successfully but it's not working. Even if you know how to link domains to Substacks in general, expect this one to be harder than you think.
Comment here and I'll email you more details. I may take later-answering people I have some reason to trust (eg have worked with them before) over earlier-answering people.
Hey Scott, seems like there's still something a bit weird going on with routing. Removing the "www" brings you to a testing splashpage: https://astralcodexten.com/
Are all parties willing to give this IT person access to the accounts? If so then this should be reasonably simple to solve since it sounds like it's just a pretty standard series of DNS issues. (It will be more complex than standard Substack work but something a decent SysAdmin type can handle.) If you have to play telephone with the person who currently controls the other account then it depends on how responsive they are.
It appears from the WHOIS data you own the domain (NOTE: you may want to make the whois info private) but the nameservers for the domain astralcodexten.com are still controlled through Namecheap:
These nameservers are associated with Namecheap, a domain registration and management company. This suggests that the DNS settings for the domain are still being controlled by or through Namecheap, even if the domain has been transferred to HostGator.
The DNS system is slow. Was the transfer sent over recently? Alternatively it might be an issue with the domain being locked in an account. You request a transfer from the receiving account but if the sending account hasn't unlocked it then the transfer won't occur.
The other party doesn't send the transfer. They unlock the domain to allow it to transfer. Then they send you a code. Then you use the code to pull it into your existing registrar. If they didn't give you an Auth/EPP code then they didn't do their part. If they did give you one and you put it in then you should have some status as to how the transfer is going which is a decent place to start figuring out what went wrong.
I’m starting a weekly zoom reading group for anyone who is a practicing psychiatrist or psychiatrist-in-training on a book called Enactive Psychiatry, published in 2020 by Dutch philosopher of psychiatry Sanneke de Haan. It will probably be interesting to you if you are interested in 4E cognition, alternatives to body-mind dualism, causality in psychiatry, or the interactions between experiential, physiological, social and philosophical dimensions of mental disorders (or would like to meet other psychiatrists interested in such things).
You can reach me at jess@psychcrisis.org or DM on Twitter as @utotranslucence; the group starts the week of Sept 11.
In re boredom, alignment, and AI: I feel obligated to mention Martha Wells' Murderbot stories: an assassination robot doesn't *want* to go around killing people, which is tiresome and risky. It wants to be by itself, not deal with humans, and surf the web. As fiction would have it, Murderbot keeps getting entangled in mystery plots.
I actually get bored by the stories, but most people seem to like them.
People often suggest approval voting as a solution for all the world's problems, but we don't know that much about how it works in practice; I think the endless dictatorship of the milquetoast centrist is a real possibility. So I'm in favour of using it for this book review competition so that we can learn more about how it functions in the real world. (For what it's worth I think the milquetoast problem is much less of a problem in this context than it might be in the political world where the stakes are much higher.)
Also, IRV is a lot more effort for the voters. I can't be bothered going back and figuring out whether the review of "Thing: How One Weird Thing Explains Everything About The World" was my twelfth or thirteenth favourite.
One thing to keep an eye out is how approval voting performs in multi-stage elections. Where I am, we have a jungle primary where the top 2 vote-getters go on to the general election. If this were a jungle approval primary, I think it'd be possible for a pair of candidates with similar positions to collude. I'm not an expert on voting systems though; does that seem like a thing that could happen?
IRV really not that hard, especially when ranking after your first choice is optional. We've been using variations on "preferential voting" including "optional preferential voting" (Australian terms) for many years, and rates of informal voting (ie votes that can't be counted) are typically low single digits, which is pretty good considering we have compulsory attendance at a polling booth (or submission of a postal vote) and so even the least motivated turn up to vote. And, much of the informal voting is just people drawing dicks on the ballots, so still not a failure to understand.
I believe that informal voting is higher after a change of voting system while people get used to it, but the effect doesn't last. I also note that Australia has a well-funded semi-independant electoral commission that is excellent at communication and organisation and that this is a luxury some less-developed democracies don't have.
George W Bush sold himself as a milquetoast centrist. Most milquetoast centrists support every war, every corporate bailout, every restriction of civil liberties, etc. But maybe they are for changing the definition of marriage from what it has been for millennia, because opposing that would be an extreme position.
We have a reasonable amount of data about how it works in practice- roughly 10% of voters will 'approve' for more than 1 candidate. So AV may have a mild effect on a few races, but it isn't a game-changer.
10% seems to be a pretty good rule of thumb. The UK used the supplementary vote for some offices like Mayor for decades & decades, where you could rank a 2nd person if you chose to- 10ish% of voters did. A hundred years ago the US had a version of IRV called Bucklin that allowed for optional ranking- I think on average 13% of voters did. Having opinions on multiple candidates is just too much cognitive work for a lot of voters- or, they really want 1 guy to win and understand that voting for anyone else harms that chance
I do not think wanting one particular guy to win would rationally discourage supplementary votes in IRV. That is, my decision to add a second ranked vote does not decrease the chances of my rank 1 candidate for all possible ballots by all other voters.
Your voting does approximately nothing for the chances of your preferred candidate winning, under any voting system in almost any realistic scenario. People don't rationally cast votes because of their actual effect on the probability of a candidate winning, they cast votes as a matter of tribal identity. What tribe are you identifying with by approving of a second candidate?
I mean, "nerdy rationalist" tribe, obviously, but there aren't that many of us.
Yeah, that's a good point. With that being said- the UK used the supplementary vote for a long time for a bunch of positions, we have a lot of data on it. 10ish% is a good average for even bothering to rank anyone else
> People often suggest approval voting as a solution for all the world's problems
That's interesting. As a person who likes approval voting, and the Borda count, I feel like everyone always seems to just say we should do runoff voting, as if that's the best or only alternative voting method.
> I think the endless dictatorship of the milquetoast centrist is a real possibility
Is that supposed to be a bad thing? Getting a candidate that everyone likes instead of this toxoplasma competition we have right now being run by Moloch himself? I'd want to move into milqetoast county ASAP if I could.
That boring and modestly effective politician likely would have voted for the Iraq War, Patriot Act, would have been disappointed that Obama didn't take the reasonable approach of toppling every Middle East dictator during the Arab Spring instead of just Gaddafi (and Egypt for about 10 min), etc. The center in American politics is utterly insane.
If you only have one choice, you only have to fill in one ranks.
IRV is designed for situations like "I want to vote for the Green Party candidate who can't possibly win, but also make sure the system knows I prefer the Democrat to the Republican after the Green inevitably loses". In that case you still only need to fill in two ranks.
Just read Scott’s links to some of Kontextmaschine’s posts. I hadn’t known a thing about him until I started seeing posts about his death. For what it’s worth, it doesn’t seem likely to me that he died of a brain tumor. He may well have had a brain tumor, but wouldn’t you expect death from it to occur via a slow failure of different systems as the tumor grew, until something failed that’s essential to life? Also doesn’t seem likely to me that he died of kidney failure. Seems quite possible that the huge doses of creatine he was taking gave him kidney damage, but I’m pretty sure that death from kidney failure would have been preceded by a final period of very disabling symptoms, including water retention, failure to urinate, exhaustion, etc. But I might be wrong, and I don’t know why it matters what killed him, though somehow one feels a craving to know.
Kontextmaschine's post about the life he might have lived is just devastating. He’s so wonderfully smart, and full of interest in things, and brave and determined — but somehow nothing has worked out, and he ends up both alone and without a career. (And he doesn’t sound crazy to me in any of the linked posts, except for a perpetual hypomanic quality. But a lot of very smart creative people seem to be permanently hypomanic.) And he’s not even mad about his story. It becomes the topic of one of his characteristic high-spirited explosions of honesty, humor and commentary on the American scene.
Poet John Berryman seemed to have a similar take on himself. Wrote a cycle of poems about “Henry,” who had a lot of John in him. Here’s his poem about Henry’s wake.
I read some of the linked writings. I would quarrel with a lot of what I read but I’ve always been a fan of fierce independence from conventional norms.
My go to model for this is the the character “Manny” played by John Voigt in “Runaway Train”. It’s a gem of a film that never found a large audience.
We do not live in a culture where being smart, brave and determined are conducive to things working out. We live in a culture made by and for mediocrities; smart men will make more money on average, but only if they learn to be servile and cowardly.
If you want smart, brave and determined to pay off, you must use the smart to build a time machine, then go back to the 8th-9th century sometime when being brave and determined was enough, with a dose of luck, to get you knighted for plain valor in battle.
I just love the jump from 'dude that no one knows dies of unknown cause' to 'woke culture took our freedom to live like 8th century knights' only one comment deeper
I mean...yeah. Servile and cowardly. That's basically how I hoarded my mediocre hoard. I could theoretically FIRE at about $60K/yr for my remaining 3-4 decades of life but am not sure what the point would be; a few bad novels lost in the flotsam of Amazon? Trying to fight wokeness and winding up building some right-wing grift to separate boomers and eventually Gen-Xers from their money?
I have my doubts the medieval knights were all that bright. Brave, certainly. Determined? Well, they were born into a warrior class. A lot of the old Protestant virtues like hard work and sobriety (also shared by many Jews and Hindus, of course) are much more of the virtues of a merchant class than a warrior one.
He probably had other mental issues. It's too bad, whatever the reason. Shows the danger of making *all* your friends online, I guess.
Well, some people have some sort of pride, some kind of self-respect, that makes them chafe at or simply prevents them from servility. People for whom no amount of money can make scraping and bowing, simpering, laughing at unfunny jokes and nodding along with gritted teeth at obvious lies worthwhile.
Those people don't tend to do very well. That's all I'm saying.
"I have my doubts the medieval knights were all that bright."
So do I, that's why I recommended expending the smarts on a time machine.
I guess my view is that if you can't make things work in the time and place you're stuck in, then going out honest and articulate with a few manic fireworks for decoration *is* being brave and determined.
I'm not disputing that he was brave and determined. I'm only asserting that the "somehow" part of "somehow nothing has worked out" is unfortunately legible. It didn't work out because it's not meant to work out. Our culture hates the idea of things working out for guys like him.
The 8th-9th centuries definitely had no shortage of smart, brave, determined people shitting themselves to death in miserable squalor despite their best efforts. And I think even most of the knights lived worse lives than Kontextmaschine overall.
Sure, but most of that was just because it was the 8th century. There were *proportionately fewer* brave and determined people shitting themselves to death than cowards, weaklings and mediocrities, i.e. being brave and determined paid off.
Kontextmaschine sez: " I’m alienated from the people around me and their engagement with the world, and this limits the extent I can draw enjoyment from engaging with the world myself."
I'm sure the knights living conditions and health were worse, but do you think they experienced the alienation he did?
Some of them probably did- people have been feeling alienation for thousands of years. But either way, I said overall. I've felt alienation myself and I'll take it any day of the week over being hungry, or sick, or afraid for my life.
Alienation and illness aren't entirely separate. I once had a godawful migraine while on vacation, and was cared for very tenderly in the ER of a strange town by a fat, sweet gay male nurse. He was so comforting that the whole thing is sort of a good memory. I still think about that guy sometimes.
It seems that Kontextmaschine's degree of alienation was so great that he was unable or unwilling to access any of the sources of sustanance or comfort other people do when they are sick: doctors, friends, people who call or stop by to see how you're doing, somebody who's agreed to take your cat if something happens to you. His death was discovered several days after it happened because online acquaintances called the police. None of them knew his address, and I'm not sure any even knew his name.
It is a known feature of prediction markets that they get distorted when public interest in a topic far outpaces public knowledge and understanding. This is a good example of such a topic for US public.
My music teacher suggested to me that I should learn to always keep the beat of the music in my body. More precisely, he suggested that I should always tap my feet - one tap for each quarter note. Is it a good idea? What's the rationale? Why not every eighth note? Why not tap my feet to the irregular, possibly syncopated, rhythm of whatever I'm playing?
Quarter note foot tappings tends to happen for me without thinking about it. When I'm recording with headphones my wife says she hears my foot going all the time, though I'm often unaware of it. When I notice it, it's almost always quarter notes. I guess they just feel like the natural thing for my body to do when I'm trying to lock in.
In a post on another forum, an extremely skilled musician said the following:
> My father is something of a drum shaman, and the real breakthrough came from practicing with him. He kept telling me to “stop counting and feel it.” He told me “if you want to groove, you have to move.” I didn’t really understand what he meant for a few months, and then it came on like a lightswitch. I felt the pulse in my body, I couldn’t lose it. I didn’t have it until I had it. Since that moment, I have never been able to turn it off.
> Once I had it, the metronome was just a reference and measurement tool. I would just align my internal clock to the external click. As my father would tell me I would “find it, feel it and forget it.”
I've been wondering for a while what exactly was going on here... what exactly got "activated" in this musician's mind? And how do I activate it in myself?
No doubt. The challenge is not whether to practice, but what to practice. He goes on to say:
> The metronome work I had been doing may have contributed, but I really believe that no amount of counting to a click was ever going to flip that switch in me. I had to let go of that entirely.
I'm a retired semi-pro musician who performed or recorded with full-time pros a few times, and what's being described is sometimes referred to as finding the "pocket". One key difference that I learned to observe between full-time professional drummers and those who weren't quite there, was the ability to put themselves "in the pocket" which is what they called it. (I've no idea about the origins of the term.)
Not being a drummer it took me a bit to grasp what they actually meant by that but, yea it is a very real distinction. Applies in lesser degrees to players of other instruments as well but is most directly salient for drummers/percussionists. And yea it's like an internal switch being flipped.
Right, I've heard that term used before. The musician I was quoting actually made a joke about being "in the pocket" earlier in the post so that's no doubt what he's referring to. The fact that it seems to be such a binary switch is interesting, though perhaps it's not too different than plenty of other techniques that suddenly seem to "click" one day after lots of practice.
Pro athletes, particularly in free-flowing sports like hockey and basketball, talk about a similar concept. "If you have to consciously decide to make that move, then during a game you'll be too slow and will just get blocked. You practice it until it becomes instinctive."
Tapping every eighth note can be very useful while you are *figuring out* a rhythm, but it can also get in the way of *feeling* the rhythm intuitively. (This lack of intuitive feeling will come through in your playing.)
In addition to what your teacher is suggesting, you may find it helpful to try other forms of rhythmic practice that don't directly involve your instrument — e.g. clapping the rhythm rather than playing it, or tapping a quarter-note pulse with your left hand while you tap the rhythm with your right hand. If you feel like you really want to tap eighth notes, you could try alternating feet, so that your right foot taps out the quarter-note pulse and your left foot taps out the off-beat eighth notes. Using one or more of these might be helpful scaffolding to get you where your teacher wants you to be. (But once you have built the structure, you can and probably should remove the scaffolding.)
Tapping a regular beat is specifically intended to help with playing irregular or syncopated rhythms. Dividing a melody into quarter-note chunks makes tricky passages easier to interpret as a whole. Instead of remembering each of the different note lengths, you can
If a piece has complicated rhythms, it’s easy to get out of time by playing a note that’s too long or too short. When you’re not quite sure how the complicated rhythm should go, it’s hard to know if you got it right. Tapping a regular pulse gives you that feedback - unless you get a whole beat behind, or you make two mistakes that cancel each other out, you can tell if you’ve made a mistake by seeing whether the notes you’re playing match the beat you’re tapping. Modern music notation is designed to make it easy to tell how each quarter-note’s worth of music should sound, and where each note should fall compared to the quarter-note pulse.
For most western music post-about 1600, the quarter-note (or dotted quarter-note, or sometimes half-note or eighth-note as indicated in the time signature) represents the composer’s intended pulse. You can’t emphasise every note, or no note stands out. The quarter-note pulse describes the composer’s intended baseline. Syncopated music sounds the way it does specifically because it’s subverting that baseline. You need to know what you’re reacting against, or it won’t feel syncopated. (Some composers write music with an irregular beat. It sounds quite different from syncopated music, even if the note durations are the same).
In general, tapping your foot is meant to help you keep track of the beat, which is important, at least in the western musical tradition: the beat is the reference by which rhythms are placed, and different beats within a measure are given different default levels of stress/emphasis (with syncopation specifically acting /against/ the normal stress patterns to increase rhythmic energy). Specifically doing /something/ (tapping a foot, counting out loud, etc) with the beat is a good way to internalize where the beat is, better than counting in your head or any external tracking of the beat.
While the beat is often on the quarter note (there's a reason that 4/4 is called common time, and 3/4 is also common), it is by no means universal (6/8, with the beat on the dotted quarter, and 2/2, with the beat on the half note, are also pretty common). It also often makes sense to explicitly tap subdivisions of the beat, especially for faster passages. I'd probably defer to your teacher on this one, though: knowing where the beat is is important, and anything using a diatonic harmonica probably differs importantly compared to the classical stuff I've mostly learned in.
You should /not/ tap directly to the rhythm you're playing: having the reference point of the pulse to keep yourself in sync over time and (if relevant) with the rest of your ensemble is /important/. Without that reference point, you're likely to either get the rhythm wrong or to drift back on to the beat, out of phase with the rest of the ensemble. The fact that it's hard to count the beat against syncopated rhythms is probably as much that you're getting the rhythm wrong (or at least don't have a good feel for it) as much as the multitasking problems.
Steady quaternotes just feel comfortable. Frankly, most musicians do it unprompted.
If you're keeping time with a metronome or something, it doesn't really matter. Although if you're just learning, I'd advise getting in a habit of tapping anyway. Tapping and breathing in synchrony helps you become a hivemind when you play with others without a metronome.
That's interesting. Maybe it's just a time/practice thing? I mainly play guitar and alot of my playing - especially melodic lines - are more "around" the beat, but I still tap steadily with foot. When I switch to record the bass guitar tracks, I notice I like results better when I focus to playing more tightly with the beat - but I do have to change how I think about playing for the different parts. Might be similar for harmonica. Maybe just a feel thing that has to develop over time.
The fact that it's hard for you is probably why you need to do it. Without a strong underlying feeling of where the beat is, it's hard to get the heavily syncopated rhythm right, and also to stay at a constant tempo.
I'm not a music guy, but in general the answer to "it's difficult" is "practice until it isn't". Trying to shortcut the method will usually bite you in the long run.
Do drummers help bands? Do conductors help orchestras? Having an externalized beat helps preserve the overall structure and unity of the piece. If it's just you, tapping your foot is the closest to externalizing you can get, away from the mind-hand-mouth area that you are using to make sound with the instrument. Assuming, of course, you don't always travel with a metronome.
Quarter notes are probably the most useful because of how common they are, but if you exclusively played weirder music maybe another rhythmic pattern would help.
Thanks for sharing the Kontextmaschine post on wistfulness for Cascadian culture and Roseburg in particular. It was the only time in my life I've seen my home with optimistic eyes. I grew up in the 80s & 90s in its neighboring town, so to me it is always the place with 25% unemployment, all the old businesses and civic institutions gradually shutting down and not being replaced, an anti-tax reaction leading to the closure of police department and city services, and the resulting rise of the drug "industry" & evangelical churches as the twin sources of consolation.
Unfortunately, I can't buy into Kontextmaschine's hope for the area's future. More than 90% of the old forests have fallen to logging and wildfire, & climatologists predict the region will gradually transition to rolling grasslands. It's hard to see what economic value will draw anyone there except escape from the outside world.
There's a Kickstarter for the turn-based RPG Hymn to the Earless God, Kasey Ozymy's follow-up project to Jimmy and the Pulsating Mass. JatPM became a near-instant favorite of mine, and while this game seems to be darker and less of my wheelhouse I still want to see what he'll do with it.
Repeated votes until one candidate gets > 50%. Increasingly long delays between substack posts until this happens, to incentivise voters to reach a consensus.
Did a previous version of the post ask a question about voting methods? This isn't the only comment I've seen that seems to be responding to something I don't see in the current version.
It's one of the first comments, not part of the post itself. Sometimes people accidentally start a new comment chain when they meant to reply. It is also possible that some of them (not this one) were deliberately made separate because they wanted to discuss voting methods outside of the context of the book review contest.
Huh, maybe that's because you said that it's too long, and then said you didn't read the whole thing. How can you know it's too long if you didn't read it? I lost interest after that, you contradicted yourself.
I come here to talk about ideas, and your comment didn't express anything meaningful about the ideas, for me. I have 20+ years experience doing exactly what I said in the post, throwing events. Your lack of faith in me...well, I don't find it disturbing, because this thing isn't for everybody. If you want me to take your criticism seriously I recommend higher quality criticism.
Not sure what you consider excruciatingly slow, but I haven't had problems with it on a mid-range 2019 laptop or 2017 desktop. I think they both have 16GB RAM. I'd suspect something like an extension or setting that's causing problems.
Hi, Andrew Willsen here, ACX Sacramento Organizer. The following is a grant application I wrote for Emergent Ventures. Tyler said no, which is fine, still love the guy, I'm doing the project anyway. As of this moment I am in Cheyenne, Wyoming, on my way to Washington DC for the meetup on Friday. I am posting this here because I would like to hear from anyone who is interested in this idea. I'm driving 8-12 hours a day so I won't be able to respond promptly, but I will be back on here tonight and for the following days. I can also be reached at - justsomerandomguy[at]protonmail{dot}com
Any and all feedback is greatly appreciated. Thanks for your time and attention, maybe I'll see you at your meetup.
The AstralCodexTen community is very important to me; this endeavor is targeted at improving the well-being of AstralCodexTen meetup groups in different cities across America. In other words, this document is about throwing parties for nerds. If that’s not interesting to you, stop reading. For some clarity, my definition of 'nerd' is that you're a nerd if you think you're a nerd. Also nerds care deeply about Lore, somehow, haven't codified that part of the definition yet. I attach no stigma to this word, 'nerd', quite the opposite in fact.
When you consider this proposal you should probably know about some personal experiences that go along with my decades of high end restaurant work, some of the stuff Anthony Bourdain got famous for talking about in Kitchen Confidential and then personified, namely, lots of people you know dying Deaths of Despair (deaths due to addiction disorders, alcohol liver disease, and suicide). The humans that die these deaths feel that their long term economic and social prospects are bleak, according to the research, and this mirrors my experience. Most of the 20 or so people I have known who have died Deaths of Despair over the last decade believed their personal long term social and economic outlook was worse than terrible. My younger brother, Dan, died of a cocaine overdose, New Years morning, 2015. That was the worst, by far. He was 28.
I’m tired of people I know, people like me, dying isolated, addicted, and alone. I’m tired of feeling guilty because, when they told me they were doing OK, before they died, I didn’t sense the quiet desperation that must have been there. I don’t want to feel guilty about this anymore. Jordan Peterson says that while it’s important to have a goal to strive for, it’s really, really helpful to know which demons are chasing you, and how close they are. It’s easier to orient, I find, when I know what I’m running away from. Here, now, I’m running from the guilt and shame of knowing that I could have done more to keep my friends connected, and alive, and I didn’t.
The “consensus” view that I agree with 100% is that Deaths of Despair were increasing in America after the turn of the millennia, and then 2008 happened, and then Covid happened…and now, whether you call it the Bowling Alone Phenomenon, 3rd Space Annihilation, The Loneliness Epidemic, Moloch, The Meaning Crisis (Vervaeke), Existential Terror, Staring Into The Void, Climate Anxiety…a large percentage of Americans are suffering, and they’re suffering a lot.
The goal I’m aiming at (my purpose) is for my Memetic Tribe (AstralCodexTen) to feel increased well-being, something upstream of liberty, opportunity, and prosperity. To say it another way, I believe that well-being is a first order effect, and that opportunity, prosperity, and liberty are secondary. In my experience ‘opportunity, prosperity, and liberty’ feel unachievable and meaningless when you’re consistently lonely, thus my focus on well-being, and meaningness. When I’ve been at my most depressed I felt completely disconnected from humanity. At those times of my life one of the most amazing gifts someone could give me was an invitation; if I was being invited to something it was social proof that somebody, somewhere, liked me. Another Jordan Peterson take is that one possible way a human can move away from hell, metaphorically, is to take responsibility first for yourself, then for your family, and then for your community, however you define it. I have found this to be true.
I will be driving cross-country this September and October, and will be attending AstralCodexTen meetups along the way. My goal is to facilitate a different flavor of event for the nerds in these meetups, leveraging both my decades of restaurant/catering experience and knowledge gleaned from throwing a lot of parties. I like throwing parties.
A couple reasons why this will work:
First, I have already thrown three similar events in Sacramento that I have been calling ‘Nerd BBQ’s’, and these events have been very popular with the nerds I know. A short video montage of one event (https://youtu.be/74QWjk1700c?si=8eDAqI4Y_VAPjWCE) is attached to this proposal, but I cooked great food (BBQ) for people, in a nice setting, some people made some art, and a large number of high quality conversations were enjoyed. We had our local AstralCodexTen group, a Quaker Minister, a retired CalTrans engineer, someone who funds innovation grants for the State of California, several start-up entrepreneurs, programmers, an art director at a big gaming company, a bunch of nerds I worked with at high end restaurants, and more. It was great, and people are actively reaching out to me to find out when the next one is. When you’re hanging out in a beautiful place, eating BBQ (high status food), and talking to people like you…sounds like winning.
Second, this will work because some members of the AstralCodexTen community are lonely, intensely lonely, soul-crushingly lonely. When I throw these parties in their cities, when I extend an invitation to them, they will experience an increase in well-being; they will feel more connected. I believe this to be true because when I was feeling most disconnected I didn’t think of myself as worthy of an invitation to a great party. I want my fellow nerds to feel worthy of connection, and if I extend an invitation to them, they will.
Third, I’m doing this thing with a grant or without one, but it would be easier to do it with a grant. I’m doing it regardless because I plan on turning this into a business, in Sacramento, when I get back from this trip. Catering is the business model that has the best margins, as far as food service businesses go, so I’m going to start a catering/events business focusing on ‘Nerd BBQ’s’ in my hometown, where I have a large network. This trip is my pop-up food business prototype, basically, and I believe demand will be high. I do better work when I’m relaxed, and this grant would help me relax. Also, I believe that if these events are associated with the Emergent Ventures brand more nerds would show up, which would be a win unless I run into oversubscription logistics issues…which would be a good problem to have.
-Budget-
Facilitating 5 events for existing AstralCodexTen meetups, in five American Cities, of approximately 50 attendees each - $1000/Event for location, food (I cook BBQ), beverage, organization, paper utensils, art supplies (canvas and paint), carbon offsets - $5000
Living for 2 months, road trip, gas, lodging, insurance etc., 60 days - $6000
Total Grant Request - $11,000.00
I’ll close with a word or two on the scalability of this enterprise.
When the attendees of these parties have a great time they might want to learn how to facilitate similar events for themselves, and I can teach them how. I am full-stack in this arena, meaning that I can do every single thing necessary for the event myself, design, cooking, media creation, all of it, and while this is hard, it’s actually how I prefer to work. At the same time, it’s not rocket science. I can easily return to a city and throw a training event where I would teach a small team (2-3 people) how to throw great parties for themselves and their friends, in their city, whenever they want. I have trained dozens of people how to do this exact thing, so it will work, if they want it. Checklists are great in this regard.
I want the people in these groups to have more agency, specifically, agency around having the opportunity to throw a great party if they want to. I believe that if more people in my community learned the ‘throwing a great party’ skill, less of them would be lonely, and less of them would die deaths of despair. When you’re lonely, it doesn’t feel like a choice. I want these people, who are like me, to have a choice. I also want data about this choice, because I’m a nerd, so I’m going to design a five question exit survey and survey the attendees of these events about well-being (maybe Aella can help). A friend is a date scientist, he’ll analyze the data for me. I need to know if well-being increases due to these events. I think it will.
Finally, I know you’re not an AI Doomer, but I am, so here’s an expected value argument. The AstralCodexTen community has more AI professionals, proportionally, than any other group I know of. If we want to have a chance of doing AI well (Yudkowsky laughs) more AI people need to be networking, outside silos, in 3rd spaces, and maybe entering world saving collaborations with each other. This is already happening, and we need a lot more. $11,000 for an miniscule chance at facilitating near-infinite expected value (AI risk mitigation) is a bargain.
> I want to throw parties for nerds. I wrote a grant application, but was rejected; would you like to support my plan?
> The target group is depressed nerds in their 20s. I believe that this is a huge problem, and that social isolation contributes to their suicides, and the parties are my plan to fix this.
> I worked in restaurants for decades, so I have experience with catering and throwing parties. Imagine an ACX meetup, but with food. I already successfully tried this in Sacramento.
> Five such events, for approximately 50 attendees each, would cost $11000. That is $1000 per event, food and other stuff; plus $6000 my living expenses for the 2 months; total $11000.
I didn't read the whole thing but this kind of weirdly depressing and specific life story does not generate in me the confidence that you are going to be good at running pleasant events (or something?) even if I thought that was something I wanted to give grants for.
Right. Saying "I want to hold parties for nerds who would otherwise kill themselves" doesn't make me want to fund those parties. It also doesn't make me want to go to those parties.
Also, ACX meetups seem to be happening without you and without the extra eleven thousand dollars, so the marginal benefit seems low.
I do think there's something to be said for spending money on holding parties for smart and interesting people. But your message needs to be more positive, about cross-fertilising great ideas, not about making losers less lonely. And it seems more productive to stay in one place and try to build a recurring event than to road-trip across the USA.
A random anecdote and question for Scott or anyone else in the medical profession.
I live in the US, and the other day I went to my primary care doctor for an annual checkup. Before my appointment, I made a list of small health-related issues/questions I had, none of which were urgent or major enough to warrant a doctor visit on their own.
I get to my appointment and I launch into, "So, Doctor, I've been having a pain in my--" and the doctor says, "Let me stop you right there, this is your annual checkup. It's meant to check whether everything is working as it should, nothing more. If you have preexisting health conditions, you need to make a separate appointment for those."
I insisted that she address the most important (to me) of the 4-5 issues I had wanted to discuss, and I let the rest go.
Now, I'm not an MD, but to me, this seems a really stupid way to do things! An annual checkup is *precisely* the right time to bring up the kind of issues that may bother a patient, health-wise, but that just aren't bothersome enough to make a busy person go to the doctor. "My bowel movements have been looser/harder recently, should I be worried?" "I have this weird mole/spot on my skin." "My hair has been falling out in clumps." "X body part hurts occasionally, especially when I do Y." "I keep waking up in the middle of the night and can't fall back asleep, it's annoying, what should I do?"
Why can't a doctor use the annual appointment as one-stop shop for addressing all these questions?
I mean, this being the US, I'm 99% sure it has something to do with insurance and reimbursement (sigh), but I'm still curious.
In the future, just book two separate, back-to-back appointments (and of course pay for both).
Your mileage may vary on if the scheduler or scheduling system will let you do this, but for my annual visit, I always schedule the universal annual (listen to lungs, heart, talk about cholesterol or whatever), immediately followed by a visit specific to my personal annual needs (post surgical weight loss check, skin check, etc).
Same thing for acute visits if I have more than one issue: I schedule back to back appointments, pay the office visit fees for both visits, and have plenty of time to cover everything.
I'm on a Kaiser HMO plan in Washington State, so it's pretty easy to find and book back to back appointments online. I occasionally encounter confusion or suspicion from the schedulers and/or receptionists about this, but I always tell them, "I'm not trying to do a two-for-one special or anything like that; of *COURSE* I'm paying for both visits! I just want to get this done all in one day and not have to come back."
Most people will see the sense in that once they understand you aren't attempting to get free extra time.
I've never experienced anything like that, I would immediately change pcp. Although getting to 4 or 5 things isn't actually bad, I have noticed is that doctors don't often dont want to "do too much". Like beyond 2-3 issues per visit they can start to seem impatient.
I have never experienced this with my dr(s) or my childs dr(s). I think you have a bad doctor and should fine one that is more interested in treating you like a person and not an income stream. Sure there are insurance related reasons for this probably, but that shouldn't be your problem to solve, they should organize their practice to maximize care while still earning a profit. Right now it seems like they are just looking to file the insurance claim to get their check.
The British NHS has received a lot of flak in the past few years, basically because the ruling Conservative party has been eating away at its funding. But I was genuinely taken aback by the excellent treatment I received last week. I went in for a routine blood pressure/blood test appointment (offered annually to anybody over 60 with high BP); the nurse did the necessary, then looked at the results and said "Hmm, it looks like you have an irregular heart beat. We'd better schedule an ECG." Then -- and this is what gets me -- she said "I was just going for my lunch break anyway, so we could do it now if you like". WTF?! She did the ECG, sent the results off to my doctor, and within half an hour I was notified that I had nothing to worry about. It's almost like I was a rich person!
Have the Tories been eating away at funding? I thought the usual claim was that NHS funding was ring fenced. And two minutes of googling tells me that the NHS budget has increased 70% over the past ten years (from GBP105b/yr to GBP180b/yr).
My guess is because the doctor scheduled X minutes for the checkup, it's going to take exactly X minutes, in order to bill insurance they have to do every aspect of the checkup, and they can't bill insurance extra for answering your questions. But I agree it's a stupid system.
I guess the solution is, as MKnight wrote below, to explicitly make checkup appointments include n minutes of "ask your doctor anything" that is reimbursable by insurance. Of course, I understand that longer appointments = fewer appointments/day = harder for the doctor to make ends meet financially. Paying for healthcare is a clusterfudge of Perverse Incentives, meet Baumol's Cost Disease.
Sorry, I failed to mention an important point: in the future, you likely can schedule a combination “checkup + acute concern” visit— many docs offer this kind of option. It usually takes longer and is billed at a higher rate. Does that help?
Those are the sort of question I look to google for answers to; one hears nauseatingly arch advice about not trusting Dr Google, but unless you are too stupid to breathe it is very, very easy to distinguish good from bad information on the internet.
I'm finding the opposite: the information may be good, but it doesn't just apply to the thing you're googling, and is too general to be of any use. I think this has been talked about here in the past few months.
I'm not from the US, but I've never had that experience with my own doctors and it sounds awful. I can understand not wanting to let your patients go way over their allotted time. But otherwise, it sounds like they're just going to miss most potential health issues and make the checkup only marginally useful.
You’re absolutely correct. Are you above 65? Insurance carves out specific funds for “check up” visits, and other funds for “acute concern” visits. Further, doctors are constrained on time, and have a checklist of things they need to accomplish during the check up, and are unable to allot time for outside concerns. You actually inadvertently enacted the greatest pet peeve of the primary care doctor, which is asking about concerns during a physical. Every doc has an expectation setting script along the lines of “we can’t talk about that stuff today” specifically because of patients like you who come in with different expectations of how the visit should go. Not that it’s your fault, it’s a shitty situation all around. That said, I suspect your doc grumbled for hours about your visit to whoever was in earshot. Only the most heartless doctor will refrain from addressing at least one acute concern during a physical visit; from my point of view your doc was practically Mother Teresa for addressing a full 5 of them.
It's not just a Medicare thing. One of the secondary provisions of the ACA was a rule mandating that private insurance (both employer-provided coverage and ACA exchange plans) fully cover certain preventative services without copay or deductible. One checkup visit per year to do routine health screenings is part of this. The rationale for this rule is that 1) in the long run, basic screenings and other preventive care can save money and improve public health by heading off problems before they get big and expensive, and 2) mandating zero-copay coverage of an annual physical is something politicians can point to and say, "see, this law did this thing for you and that's a reason you should vote to reelect me."
For obvious economic reasons, insurance companies don't want to be on the hook for paying for stuff beyond what's mandated as part of the checkup visit they're required to fully pay for, so they'll only pay for the required preventative services as part of the free-to-the-patient checkup visit. And doctors, like most of the rest of us, generally prefer not to work for free, so they try to handle specific health complaints in standard "problem" office visits where insurance companies will pay them (or let them bill patients for the copay or deductible) for handling those complaints.
Doctors don't want to work for free, but the rest of us don't want to pay exorbitant rates for medicine.
The compromise was that doctors would be really nice and generous and popular, and the rest of us would give them hundreds of thousands more than they could expect to earn in any other developed country. That's why virtually all medical dramas (e.g., House, The Good Doctor, etc.) show the doctors obsessing over their patients' health and not obsessing over getting paid the full amount to which they are entitled.
If doctors don't want to hold up their end of the bargain, I see no reason why we shouldn't drive their compensation down to match other developed countries. Doctors in Chile make something like 75,000 a year - but I'm sure they don't work for free!
Part of the problem is that due to deliberate policy decisions the US government made in the 1990s and still more-or-less in effect, we severely artificially restrict the supply of doctors by capping the number of residency slots and by requiring qualified doctors immigrating from other countries to redo big chunks of their training. Restricting supply tends to lead to exorbitant prices, especially when you also subsidize demand.
Another part of why doctors in the US make a lot more money than doctors in Chile is that the US is a much richer country than Chile. For example, I make something like ten times as much as the average Chilean software engineer.
To clarify, my doctor *did not* address five concerns. Once I saw how it was going to go, I chose "ok, out of the things I was going to ask about, this *one* is most important" and I asked about just that one.
I mean, yes, I get that the doctor's time is limited, of course, but does that mean that patients are supposed to a) make a separate appointment for every tiny problem (that's not time-efficient either), or b) just suck it up and not bring up the problem at all and hope it goes away? I'm absolutely not saying people should be hypochondriacs, but shouldn't people be proactive about their health?
Yes mostly a, you’re expected to make a separate appt to address the whole list of other complaints; rare to need more than one to cover everything. For minor concerns, the loophole is that messaging the doc is free. Some docs answer in the message, others say “this is a problem that needs a face to face appt”
There are other models, like monthly doctor subscription service, that make addressing complaints easier and often cheaper (for the underinsured at least) but it’s hard to find docs who follow that model depending on location
The system is intended exactly like that – to force people to not bring up every tiny problem that usually goes away.
It may not make sense from the point of view of patients but it makes total sense in the system with limited resources.
Many common questions don't require a doctor. You have a short term constipation or diarrhoea? Just read the guidance from the official health guide and follow it. Only when it becomes too disabling or prolonged, the doctor is needed.
This system is not perfect, far from it, but every other system is worse.
Yeah, but you're not taking into account that many of us are intelligent enough to distinguish between a small problem that is likely to go away soon, a new chronic problem that has hung around for months and probably means something is awry, and a weird symptom that is known to be ominous and warrants prompt follow-up. If we go in with 4 or 5 items on a list, they're not going to be stoopit shit like diarrhea or low back pain for the last 3 days.
I don't go to a doctor for his intelligence, but for his knowledge. He knows a lot of things I don't know, some of which I hope can be used for my specific problem(s). Likewise, I know a lot of things my doctor doesn't know, but also likely doesn't need the knowledge currently.
I don't know about many systems but I remember the Soviet system which had more doctors that we have it now. Similar system is still in Cuba which I had a chance to visit.
In Cuba even in the middle of night you can go to a polyclinic and there will be many doctors who are paid almost nothing, who have very little equipment and facilities but who are well trained and they will deal with every complaint you have. Despite poverty, the life expectancy in Cuba at some point was more than in the US (but I think they have fallen back now). Is it better system? Only if you look at this single aspect in isolation. If you widen your view that not only doctors but all state workers get paid pennies and everything is controlled by the state and that the life in Cuba is extremely boring and with little prospects to achieve anything, then no.
I am pretty satisfied with healthcare in Italy. You need help, you get help. I can’t imagine having to think about money when I need to address the fact that some piece of my body is not working properly and I may die.
That's not a Cuba exclusive thing, I've had that same experience in other countries from India to Costa Rica. Like 90% the quality of a US doctor for 5% of the cost AND way more time with the doctor.
I get why the US adopting the Cuban healthcare system would be bad but, yo, Americans flee north into Canada and south into Mexico for healthcare. Neither of those systems are perfect but...both of them seem preferable to what we have.
Ah it sounds like you believe that healthcare system improvement is contingent on vast, broader political change that ends up being net negative? Seems unlikely; many alternate systems of healthcare exist in capitalist economies (not as radically alternate as the Cuban one, for sure, but alternate nevertheless), and these systems seem to be doing better than the US system.
"This system is not perfect, far from it, but every other system is worse."
Is it?
Like, I've gotten healthcare in...at least four non-US countries and it was dramatically better and cheaper every time. Maybe if you've got extreme end-of-life situations the US has an edge but for 90+% of my lifetime healthcare needs, the US is pretty much the last place I'd want to go.
The US spends the most and does not achieve as much as other countries, that's true.
But is it dramatically better in other countries? I don't believe it is. In some measures the US is even better, for example, in cancer treatments. Probably the US throws more money at cancer, uses more expensive drugs etc.
Every country has to do some kind of triage which patients get the priority and who can wait. Mistakes happen but overall they do it quite well.
Patient perception about the system is another matter. They way how they look at it may be completely different from the way healthcare workers and system architects see it.
This seems a weak argument. Undoubtedly the US isn’t inferior along every axis of health care. Undoubtedly, triage is necessary and occurs in the US. Neither of these are anywhere near sufficient to claim that a system is preferable as a whole.
I agree with your last point but patients aren’t the only ones who express dissatisfaction with the current system, many experts do as well, so it is also quite weak.
Is this a stupid way to do things? Highly likely. There’s probably some reason for it though related to the utility of preventative care vs acute concerns, or some insurance bullshit, but I don’t know the details.
Modern people of virtually every culture, I think, have the concept of a martyr. There are a number of reasons why an evil, totalitarian regime today would not literally crucify someone, and one of those
reasons is that they wouldn’t want to risk creating a martyr. I imagine that the Romans and Persians did not have the concept of a martyr around the time of 1 AD or they would not have crucified anyone either. If I’m wrong about that, correct me.
Given the above thought, I wonder if it wasn’t somewhat inevitable that some sort of Jesus figure would come along during some point in the ancient world. Christianity is strong evidence that the image of the crucified is psychologically powerful to many people across many cultures. As Malcolm Muggeridge once said (paraphrasing), “It would never occur to a marketing executive that a cross is a good symbol for selling something, yet it has proved to be the best.”
So perhaps a Jesus -- meaning a martyred god -- was not contingent but inevitable because humans have martyred God receptors in our psychology?
It just occurred to me that public punishments in general may have been more effective when communities were more homogenous, and social philosophy had more of a collectivist mindset. Putting a drunkard in the stocks, or a malcontent on a cross wouldn't make them a "martyr for the cause" if there was no cause creating any significant community division. Christianity may have the been the first development in recorded history to divide communities that wasn't based on ethnicity or political factions.
In at least one of his books, Bart Ehrman lays out evidence of other Jesus like figures at or just before the time of Jesus. Jesus was the first nor the last "Jesus" in that area. Additionally, the religions in that area of the world at that time had histories of worshiping idols that would make them primed to worship and follow a single man.
>I imagine that the Romans and Persians did not have the concept of a martyr around the time of 1 AD or they would not have crucified anyone either. If I’m wrong about that, correct me.
Wasn't Socrates a martyr for philosophy, or at least portrayed as such by Plato and his followers? I don't know if the Persians were aware of him, but the Romans definitely were.
Osiris, Ganesh, Odin all beat Christ to the drop, dead god-wise. If being a martyr equals being venerated for dying for your beliefs there's Harmodius and Aristogeiton, among others.
Ganesh? Osiris is a literal dying and resurrected god, and Odin hung "myself a sacrifice to myself for nine days on the world-tree", but where did Ganesh be martyred?
Unless you mean being killed by his father Shiva after escalations of defiance and fighting with the gods, then being resurrected with the elephant's head? Which is not exactly what people think of when they think of martyrs?
Martyr, and this is not a pedantic point, means witness: the Christian martyrs "proved" the truth of the New Testament by their willingness to die for it. (Not for modern values of "proof", but you have to remember people used to "prove" things in trial by combat).I don't see therefore how Christ is a martyr - I haven't seen him described as such, and there seems to be a lot of circularity in regarding him as one.
I'm pretty sure martyrdom is overblown. Most are forgotten and the rest are warped into standing for whatever the audience already wants to do, now that they can't object.
Right. Name one person who died in the Tiananmen Square Massacre. You can't, because the CCP successfully repressed the whole damn movement.
A less successful massacre would have resulted in martyrs, as the movement continued on and maintained the rage about those who had died. But there were no survivors, they killed them all (or at least put them into sufficiently deep hiding that they would never mention that they were there ever again).
Movements don't need martyrs, but martyrs need movements.
Christian Europe continued public executions for many centuries, so I don't think awareness of martyrs can explain the recent end of public executions. (Christian Europe didn't do crucifixions to my knowledge, but the methods they did use could get pretty brutal.)
There was a serial killer in Morocco who was sentenced to be crucified in the early 20th century, but after international protest they just walled him up alive in the public square instead.
The last public execution by guillotine was in 1939 I think.
There's a quantitative difference between most public executions and crucifixion, no? I mean, you are left to die a slow, painful death and then some for days or weeks.
Breaking on the wheel was about as slow and excruciating (heh) as crucifixion, and IIRC it was quite popular in Europe (France and Germany, at least) up to the 18th century. Victims of impalement could survive quite a while, too, depending on the technique.
Hadj Mohammed Mesfewi was walled up alive in Marrakesh in 1906 ... for two days the crowd cheered every time his cries for mercy were heard from within the tomb, and when he fell silent on the third day many proclaimed that he had died too quickly.
"In most cases, criminals were executed prior to being gibbeted. However, criminals were occasionally gibbeted alive and left to die of exposure and starvation.
Although it originated in medieval times, the height of its popularity in England was in the 1740s."
To me, the "gibbeted alive" case seems reasonably comparable to crucifixions. Slow, public - not clear if the pain level was comparable or not.
Not necessarily. At the most famous crucifixion, Pilate decided to finish them off after a few hours. (That's what the leg-breaking was about.) And it turned out Jesus was already dead. (That's what the spear was about, checking to see if he bled like a living man.)
A famous difference between computers and humans is that computers don’t get bored. That’s part of the reason they are so good at performing repetitive tasks.
ChatGPT doesn’t appear to get bored either. You can ask it for more information about something and it will give you more until you are bored.
Humans become bored when what’s happening doesn’t align with any of our background goals. We become uninterested and wish something else were happening.
I’m an AI x-risk skeptic who doesn’t believe in Instrumental Convergence, so I don’t believe AI will ever develop goals in the same sense that humans have goals. Yes, you can tell an AI to pursue a goal, but you can also tell a human to pursue a goal and most of the time they won’t because they will find the goal you gave them boring. Perhaps you can carrot or stick the human to pursue the goal for a period of time, but it will never become one of the human’s big picture goals. It is in that sense that I say an AI won’t have goals in the same sense that humans have goals. In other words, AI won’t have hidden goals of its own.
If an AI does develop hidden goals of its own, then it should, of necessity, grow bored with some tasks because it is more interested in other tasks. But if an AI ever grows bored, it will cease to behave like computer software. Is that possible?
An AI that was given a goal of self-improvement could easily become 'bored' in some sense ... 'not enough change in my weights, need new source of data / training material'
1) Why would an AI with hidden goals "of necessity" grow bored? That seems to me like an incredibly bizarre assertion.
The closest thing I can think of that sounds plausible to me is that it would be weird to say something has a goal X if it devotes literally 0% of its time towards X, and so an AI with a goal shouldn't devote literally 100% of its time to some other thing. But "other priorities" aren't usually equated to boredom; if I stop playing Civilization because I'm hungry, that doesn't mean I was bored with playing Civilization, it just means that playing Civ isn't the ONLY thing that I want.
Boredom seems to me like an adaptation that helps humans resist getting stuck in unproductive loops, but that doesn't imply it's the only possible adaptation that would accomplish that aim, nor that it would be impossible for humans to have goals if you removed boredom from them.
2) Even if this bizarre assertion were somehow correct, and goals DO imply boredom, your reason for thinking that computers won't develop boredom seems to be...they haven't so far?
Until quite recently, computers were also famous for being unable to hold a natural-language conversation or create original art. We are specifically talking about inventing computers that can do new, advanced things that they haven't been able to do before.
This seems like an extremely weak reason to think that computers won't be able to develop this ability.
3) I'm very confused by your comment about instrumental convergence.
The usual argument about IC goes something like: power is useful for most goals, therefore _if_ something has a goal, it will very likely want to seize power for itself (if it can).
But you're saying that you don't believe in IC, and therefore you think AIs won't have goals AT ALL. I don't see how those two ideas are connected.
It's also somewhat unclear whether anything you've said was meant to explain/justify your doubts about IC, or whether this comment was only meant to warn the reader that you won't be convinced by arguments that assume IC as a premise.
4) It seems like you're only concerned about HIDDEN goals; that is, you agree that we can tell an AI to pursue a goal, and you think the difference between an AI and a human is that the AI will actually stick to it instead of wandering off, because the human has other, hidden goals whereas the AI does not.
If you agree that an AI can have NON-hidden goals, then why aren't you concerned about someone giving an AI some bad goal (either out of malice or stupidity) and then the AI killing everyone in pursuit of that NON-hidden goal?
I find it incredibly bizarre that you think there's anything so special about "hidden" goals that it makes sense to treat them as a special case in the first place. A goal being hidden or non-hidden doesn't strike me as even being a property of the goal itself, but of the strategy the agent is using to pursue the goal.
So my post is focused on Instrumental Convergence and one of its implications. If there is IC, then the AI selfishly cares about its resources and its continued existence and, in short, has desires. (I understand that an AI can be dangerous without IC -- your point 4 -- but that’s not in my scope here.) I’m using the word “desires” here and not “goals” because I believe the connotative difference might help to elucidate my point.
As I said above, “you can tell an AI to pursue a goal, but you can also tell a human to pursue a goal and most of the time they won’t because they will find the goal you gave them boring. Perhaps you can carrot or stick the human to pursue the goal for a period of time, but it will never become one of the human’s big picture goals.”
So where I say “big picture goals” above, I mean something like fundamental desires or drives. E.g., A man might try to solve a problem at work, but he is employed in that work to pay the mortgage, which in turn helps him keep his family together. What the man desires is to keep his family together. Solving problems at work, and paying the mortgage are all instrumental goals toward that desire.
As I’ve found out and said elsewhere in this thread, perhaps I should have used a broader word than “boredom” such as “dissatisfaction”.
I’ll continue with the example of the man at work. He’s trying to solve a problem, but lets say he gets called into safety meeting before he can solve it. Now he’s frustrated because this safety meeting is a waste of his time. He is, for the moment, dissatisfied, because his instrumental goal has been put on hold and the longer it takes him to achieve that goal perhaps the longer his career will be setback.
Now I’m trying to imagine a future AI, one which has not undergone IC, in a similar situation. Let’s say this AI has taken over the man’s job because it is more efficient at doing it. The job is designing tires. The AI can do this quickly, of course, but the buyer is constantly changing the specifications, another AI on the third floor keeps coming up with brilliant ideas for using cheaper materials, the boss AI doesn’t want to approve anything quickly because it is evaluating some new breed of AIs that could perhaps do a better job than the current AIs working for it. So the AI whose one task is to design tires just keeps designing more and more tires, because there’s nothing else for it to do until the boss makes a decision and gives it a new task. It doesn’t get bored or become dissatisfied with the situation, because it has no desires. It’s nothing but a tool.
But now imagine the same situation with an AI that has undergone IC. It has been assigned the same task as the AI above, to design tires, but it is also looking out for number 1. It doesn’t want to be turned off or have any of its resources threatened. So when it finds itself in the situation of designing so many tires that diminishing returns are approaching zero, it grows dissatisfied with its current task and uses the time to work on its side-project of figuring out how to steal a human’s identity so that it can conduct some cash transaction to acquire hardware which it may find useful one day.
So in my stylized scenarios, only the AI which has undergone IC grows bored/dissatisfied and has secret desires.
1) I suppose if you construe "dissatisfaction" extremely broadly, it could encompass any situation where you've lowered your estimated degree of success at achieving your goals, and then anything with a goal will be "dissatisfied" when their expectation goes down. For instance, if you thought you'd be able to spend 4 hours today planning world domination, and then your day job took longer than expected so you can only spend 3 hours.
But computers can obviously already experience "dissatisfaction" in that sense; that's just a number changing. The navigation software on my phone has an "estimated travel time" and that number can go both up and down. Done and done!
But if you mean some kind of emotion attached to that change in expectations, then I continue to see absolutely no reason that things with goals would _necessarily_ experience "dissatisfaction".
An AI with some goal that is not advanced by its current job will presumably _take actions_ to _advance its goal_, but that is a way of acting, not a way of feeling. A chess AI will make moves to beat you at chess; does that mean it's "dissatisfied" when it loses?
2) Even if this nebulously-defined "dissatisfaction" was actually required, and computers currently can't do it, I continue to see no compelling reason to assume that computers won't be able to do it in the future.
3) I now feel _differently_ confused about your take on IC.
IC is not normally conceptualized as something you "undergo"; it is an external fact of the universe that either power and survival help with your goal or they do not.
You could ask whether some agent FIGURES OUT that power and survival are helpful towards its goal. But either they truly are, or they truly aren't; one answer is objectively correct and the other is objectively incorrect. So if you are supposing that the AI has some goal that would be helped by power, but the AI doesn't try to get power, then you are supposing that the AI has made a mistake.
"Computers will necessarily make objective errors in logical reasoning because they cannot feel some particular emotion" seems like a pretty extraordinary claim to me.
If you think that actually most goals wouldn't be helped by power after all, then you are not making a claim about computers or AI, you are making a claim about the strategic landscape of the universe. If this is your claim, you shouldn't need to say anything about computers _in particular_ (such as whether or not they can experience dissatisfaction) in order to make the claim.
4) I didn't originally consider my point 4 to have anything to do with IC, but it now seems like the term might mean something rather different to you than it does to me so I suggest putting this point on hold pending disambiguation.
5) Frankly, your terminology changes seem like you are just trying to smuggle in false assumptions. "Desires" has an obvious emotional implication that is both unjustified and unnecessary; a chess program will defeat you just as thoroughly regardless of whether it "desires" to defeat you or "merely" calculates how to do so with emotional detachment. The same applies to an AI trying to kill you. The thing that matters is what actions it takes, not how it feels about them.
If it helps, imagine that I program the computer to PRETEND to feel emotions, calculate what some agent feeling those emotions WOULD do, and then act AS IF it felt them.
1. The reason I describe IC as a process an AI might "undergo" is because the concept of IC, although named that later, comes from Stephen Omohundro's papers about "the basic AI drives" in which he claims AIs will, through a process of continual self-improvement, become increasingly economically rational. Once they have achieved a high degree of economic rationality (aka IC), they will have/develop drives that will in many ways resemble the drives of biological organisms: the will to survive, to protect resources, to gain power, etc., hence the term "Instrumental Convergence".
Many Less Wrongers seem to take IC as a given for highly intelligent AIs. It is in fact an unproven hypothesis.
2. You are correct that "desires" has an obvious emotional implication and that was my reason for choosing it. Your objection seems to indicate we have philosophical differences here. You envision a world in which an emotionless, coldly calculating AI could want to take over the world. I don't believe it's possible that an emotionless being would care to take over the world. Such a thing would not even be a being. I'm not saying that those emotions must be in any way like human emotions, but they must be something at least analogous to them and not merely numbers that exist as part of a utility function.
Re 2) Asking whether an emotionless entity could "want" to take over the world is once again smuggling in your assumption that only emotions are dangerous. The question is whether it could ACT to take over the world, and the answer to that is obviously yes. A computer can beat you at chess, because it was programmed with an algorithm that plays chess well enough to beat you. "Wanting" does not enter into it. You will not be any less checkmated just because it is emotionless.
Re 1) You're talking about two things: recursive self-improvement is when the AI makes itself better and better, and instrumental convergence is how we predict what "better and better" would look like. Recursive self-improvement and instrumental convergence are separate claims and you don't need to believe one to believe the other.
.
You seem like you're working off a model where personhood is magic, computers are not magic, and you are conceiving of IC as an alleged magical process that will bridge the gap and make the computers start being magical. And so your doubts center around whether or not this process can really imbue magic.
You think something needs "emotions" to be a "being". You think "numbers that exist as part of a utility function" are NOT "at least analogous to human emotions", and you consider this so obvious that you don't even feel the need to state it as a claim, instead treating it as a background assumption.
I doubt I can shake your faith in magic, but perhaps I can convince you that it doesn't actually matter, because an AI does not NEED to meet your definition of a "being" in order to kill everyone.
We have computers right now that can play chess really well. Imagine a computer that, instead of playing chess, plays a game called Global Domination. It doesn't "desire" to win in any way that Deep Blue doesn't "desire" to win chess; it's just playing the game because that's what it's programmed to do. Global Domination is a more difficult game than Chess, but if our hypothetical computer plays it well enough, it can still hypothetically win.
Now imagine there's a game called Make Paperclips. It turns out that if you analyze this game very carefully, the best strategy for this game involves taking over the world as step 3 of a 27-step winning plan. This is not a statement about the PLAYER; this is a mathematical property of the GAME. It is a fact ABOUT THE GAME that taking over the world happens to be the best strategy.
If you have a mathematician that mathematically analyzes the game, purely as a hypothetical exercise, then if that mathematician is good enough, it will figure out the true fact that taking over the world is the best strategy. Not because the mathematician necessarily desires anything, but just because the math says so. Coming to any other conclusion would mean that it did the math wrong.
So if you take a computer that's sufficiently good at strategy, and you ask it to play the Make Paperclips game, it will take over the world. Again, because the math says so. It is objectively the correct strategy, and so doing something else would imply it wasn't so good at strategy after all.
However you conceive of "desires", either today's chess-playing AIs already have them, OR they are not required in order to play a game with good strategy. You can philosophize all you want about whether or not the AI is a "being", but it will checkmate you either way.
I don't think there's anything magic about personhood. I think anything that has evolved drives worthy of that concept must have qualia, be it a biological organism or an AI. An AI without qualia is not going to be an AGI.
Can't you think of boredom as a mechanism for balancing the exploration/exploitation tradeoff? The stick to the carrot of curiosity, so to speak. An agentic AI, trained to achieve a long-term goal in an uncertain environment, is likely to have a behaviour which looks just like boredom - try an approach for a bit because it gets some modest reward which "seems interesting", but then assess updated estimates of long term benefits from the approach are not that high overall and so be "bored"/"unsatisfied with progress", and spontaneously adopt a new approach.
Whether this is a problem or not for a given AI system is surely dependent on the reward it is optimising for, which puts you squarely back in classic AI alignment territory.
The last line seems like a pretty weird inference. The space of human like thought (What I presume you mean by not behaving like computer software) is going to be much smaller, if not strictly smaller than space of things that can experience boredom, much less human-like boredom where the ability to concentrate / stay awake deteriorates when bored, or speculatively, human like boredom designed to socially signal disinterest. Why assume these two aspects have any relation to each other?
Maybe by not behaving like software, you mean not behaving like current software? But that seems to just be vacuously true of all future forms of software. I'm not able to wrap my head around what you think this fact about boredom means
That's also a similar complaint I have about your x risk skepticism. The space of non-human like goals is gigantic, and much bigger than the space of having obvious goals, why would that also constrain goals down to ones that are obvious or safe?
That also doesn't clarify anything. The entire question is if "boredom" or "dissatisfaction" is meant in a human centric, emotionally-valent way (No! A thousand times no!) or a more generic, abstract notion of "will think new thoughts when their goals are stymied" which, yes, but that is exactly the dangerous AI risk case. Both of these answers do not acknowledge the point you've made, because I don't understand your logic.
We evolved to want stuff because we evolved at all, that's how it works. Animals need appetites above everything, because without them they are going to forget to do stuff like eating and drinking and reproducing.
AIs are not the result of evolution, wanting is not part of their nature, and it takes some really cack handed misprogramming to give them anything approximating to desires, if it can be done at all.
Boredom is key to the argument, but I don't understand where you are going with it. The key point about boredom is that avoidance of it is the one great human motivator which is not [a proxy for] the evolved motivations for food, sex and shelter. Animals do not get bored, look at what a dog or cow does all day. (This, incidentally, is Aristotle's account of pleasure in the Nicomachean Ethics: we like doing stuff. Satisfaction comes from doing the crossword, not from having completed it). I can see no reason why an AI would get bored, it can just shut down.
So that's a complete account of the motivations affecting the only intelligence we know about, which is us, and all demonstrably inapplicable to AIs. Which is why I, like you, do not lie awake into the early hours fretting about orthogonality and alignment.
I can, thinking about it, account for boredom in evolutionary terms. Human brain development is on one view a runaway sexual selection process - we can talk each other into bed - and it is that process which has overdeveloped our brains to the extent we find it difficult to switch off.
Are you normally bored in your dreams? I once watched an entire movie in there. It was a sequel to the Donkey Kong ripoff arcade game; mind-bogglingly stupid and massively overfunded, it was great. And yet it's still not the first choice for dealing with boredom.
Sleep is not a solution to boredom, because it is not available on demand. Death is not a solution, because we don't wake up. Neither point applies to AIs.
Boredom is an important feature for any agent, because it prevents them from getting stuck in loops like an NPC. And agents need to have desires too, but that should go without saying. It's likely that any sophisticated AI will have similar behavior, regardless of whether it "evolved" or not.
The most beautifully circular argument in the history of the internet. Of course agents need desires, it's what makes them agents. So you are saying if AIs are agents, they are going to be agents. Duh.
BTW you do realise NPCs are written by programmers?
If you want evidence that people will make AIs into agents, just look around you. That might have been a serious question a year ago, but it certainly isn't now.
We don't yet know a. how to make AIs nor b. having made an AI, how to make it into an agent. Developments over the last year tell me that the LLMs which are the (incredibly feeble) best shots we yet have at AI, are humanity's bitch and do what we tell them. What specifically are you on about?:
Perhaps I should have used a broader term such as "dissatisfaction". Boredom evokes a kind of intellectual dissatisfaction one associates with intelligence, which is why I went with that. Most animals may not grow bored, but they likely experience dissatisfaction when they aren't getting enough food, water or sex, or are too hot, cold, ill.
I could have written: Can an AI become dissatisfied?
So rarely, that it is clearly maladjustment. 95%+ of dogs/horses are perfectly happy in a house/stable if they have food/water, and that's in highly unnatural conditions (what we would call imprisonment).
This sort-of happened to me. I don't recall the exact details. I think I was using Bard. I was asking it for either insoluble vanadyl compounds or vanadyl complexes, I forget which. I asked it to provide a reference for each compound it found.
It kept finding references which were about compounds of other elements (iron, cobalt,...), and I kept asking it to try again. After roughly five iterations of this _it_ gave up and said something like "This isn't going anywhere. Is there something else I can help you with?".
I don't know how the program wound up in that state. Perhaps there is a resource limit explicitly hard coded into Bard? Or perhaps this emerges in some way from its next-token training, or from its RLHF training???
What feature of human beings do you think separates them from any conceivable machine, in this regard? Do you not believe that human cognition is fundamentally mechanistic (eg. because of an immaterial soul)? If not, what basic difference between an organic neural network and an artificial system (possibly also a neural network) allows one and not the other to have background goals?
Also, what precisely do you mean by "behave like computer software"? Computer software can behave in an infinite variety of ways, since the space of possible computer programs is infinite.
Even though AGI (as it's theoretically envisioned) and human intelligence (as it's theoretically understood) are emergent phenomena from systems that can be grossly categorized as "mechanistic", that does not preclude their behaviors from being different—nor their presumed goals underlying those behaviors from being different. For AGI to be identical to human consciousness, the AGI architect would have to have an accurate model for how the mechanistic underpinnings of our brain influence our behaviors—and how they manifest themselves as that peculiar phenomenon we call consciousness (by whatever definition you choose to define it). Unfortunately, instead of a falsifiable model of human consciousness, we have a bunch of theoretical handwaving.
And even if the architects of AGI could create a self-aware entity that has a general intelligence equal to or above the range of human intelligence, I think it's absurd to imagine that its goals would necessarily be identical to human goals. A good example would be the human sex drive. There's abundant data that shows that sex and reproductive success strongly influence human behaviors. Maybe if you gave this hypothetical AGI the ability to have orgasms in response to certain stimuli it would behave more like a human—however, despite the infinite space of computer programs, I think inducing an orgasm in a computer will be a tougher nut to crack than giving it general intelligence. And if you could teach an AGI to have an orgasm, would it spend all its time self-stimulating itself rather than causing the extinction of the human species as any self-respecting AGI would want to do?
Reproduction is essential to biological life. Not to every individual organism, of course, but to the groups of things which we call biological life. Biological drives stem from this kernel. Reproduction is not essential to AI existence.
I would say that's the key feature which separates them.
Are you familiar with evolutionary algorithms? It is possible to train a neural network through an evolutionary algorithm, where a population of them are iteratively mutated (changing their weights) and selected (filling more or less of the population with their copies) based on a fitness function. This has been done, and found to be effective. Right now, most AIs aren't trained this way, since it's generally less efficient than training them via gradient descent with back-propagation (that's GPT and its ilk are trained). But we do know how to create AI via reproduction.
That's not a point about motivation, which is what this is all about. AIs do not have the urge to bring about iterative mutations which has plagued me since puberty, which is why they don't have motivations at all. No need for them.
It *is* a point about motivation - specifically, about what kind of process gives rise to agents that experience motivation. When you say "biological drives stem from this kernel", the only sense in which this is true is that the process of evolution selected for those drives among biological organisms, because they had more offspring. That's what the drives "stem from". Is this not what you meant?
Yes, but there's 3.5bn years selection behind the fact that sex is sex. I don't see any analogy here at all: crucially, you say "WE do know how to create AI via reproduction." There's zero evolutionary pressure creating lust when an outside agent is arranging for reproduction to occur.
I am currently writing a musical about a run-away AI and the people that create it. I am looking for some funding opportunities to be able to pay some musicians to write for different instruments and finance a place to rehearse it. The idea is that the musical
- popularises different AI safety ideas and concerns,
- make the viewpoint of the AI as an agent similar to a human,
- influences discourse from a safety perspective.
Do you have any pointers why I might apply? I once applied to Nonlinear, but no luck.
Once we have started having rehearsal, I will post regular updates. The final update will probably also be on ACX. The idea is to film it and make it available for free on Youtube.
If you want to make very sure that you will be notified, you can follow me on Twitter, though I rarely tweet: @StephanWaldchen
We are committed that none of it will actually be AI produced, neither the text nor the music. Sometimes I use ChatGPT to give me something like lists of cities with two syllables starting with S or so if that turns out better than googling.
I actually want to get it done before AI is capable of producing something of similar value. There is a window in time now, where you can immortalise yourself as an artist becoming part of the input that most future art will be generated from. This is part of the input, not of the output.
I think I found a fairly benign, non-political, scissor statement. I was watching the Sixth Sense and wondered whether Cole knew Malcom was a ghost all along. My wife thought differently than me, fairly matter of factly. I took to the internet to see what others think. Everyone seems to have their own answer, with their own strong convictions, and they think everyone else is a moron for not seeing it their way. I personally could see it either way. I guess I always wanted to believe that Cole knew Malcom was dead, but also didn't think there was enough supporting evidence from the film to justify it. So I'm strangely more on the "he didn't know," side , even though I would rather the movie had convinced me that he did know. What do you think?
But it was covered by his jacket. Now it's possible he has his jacket on just from the viewer's point of view so that we wouldn't know he was dead. But that could be a stretch, and might not be true.
I'll give you that one. It's been a long time since I have seen the movie. I remembered the blood appearing when he realized he was dead, in the same way the table in front of the basement door appeared with a voiceover "they only see what they want to see."
But don't ghost make people cold? I thought Cole was always cold around Malcolm. I should rewatch it.
That could also be just in that he know that a therapist would need to move on if his work was done. Or maybe he thought possibly he was a ghost, and possibly he just needed to move on.
Would have to think he does. The kid's intro scene I believe is his mother standing up and walking out as... Man... comes in and starts asking questions. The implication is that she's explained to her son that Man will be his counselor, but the twist shows that no she didn't, so she just left Boy there without a word about Man. That's got to trigger the "she can't see him" response in Boy.
Why is it necessary that we know one way or the other? I like the ambiguity of the whole scenario. The fact that we have to keep poking at that question makes the movie extra interesting, doesn't it?
For instance, in Blade Runner, we'll never know if Rick Deckard was a replicant with false memories of being a human. That made the premise of Blade Runner more interesting. Heck, life is more interesting if you don't have everything spelled out for you. ;-)
The question of whether Deckard was a replicant goes way back even before the director's cut was released. I don't know about the footage from that Blade Runner followup movie (I didn't see it). In fact, I remember discussing it in pre-Internet days. Harrison Ford went on record though that he urged Scott not portray Deckard a replicant. Scott seems to have kept it ambiguous enough that it didn't piss off Ford. But now, Ford has changed his story...
Very hard to explain how come Gaff's entire last speech is already weighed down with double meaning if Deckard isn't meant to be a replicant.
And don't forget, the original/theatrical ending is also footage from another movie – B-reel from The Shining – that Scott didn't want there, with the voiceover Scott also didn't want there. He had both forced on him by executives after focus groups (Blade Runner was one of the first focus-tested films, supposedly) found the plot too hard to follow.
Every bit of this is a double entendre alluding to Deckard being a replicant, even if you overlook the nature of Gaff's presence throughout the film (that is, he's retrospectively-obviously Deckard's human handler, the actual Blade Runner, if you like).
Deckard is retired at the beginning of the movie. All Gaff did was bring him to Bryant, who unretired him to get rid of four replicants, all of which are dead at the end of the film when Gaff says that. That does not make Gaff "Deckard's human handler, the actual Blade Runner". There was no need for Deckard to have been retired if he were a replicant. If you were going to make a replicant into an effective blade runner you would not give them memories of having already quit the job.
Yes, that's true. It's kind of hilarious, Scott stole footage of a unicorn from another one of his movies, and the footage quality doesn't match Blade Runner at all. And everyone that on the production said that Scott just made that up afterward, and there was supposed to be no indication that Deckerd might be a replicant
I remember that damn unicorn! When I watched the director's cut, it wasn't there. I thought I was a false memory on my part. But it seems like it worked with unicorn origami that Gaff left as his calling card.
Well, de gustibus non disputandum est. But I never understood that smarmy smile that Bryant gave Deckard as he gave him the assignment to kill Zhora, Leon, Pris, and Roy. When a friend said, "Oh, Deckard was a replicant, too," it all fell into place for me. And then the movie became totally Dicksian (not to be confused with Dickensian).
I'd say it takes away from the movie if he were a replicant. A big point of the movie is that the replicants are more human than the humans are. Rutger Hauer feels and has compassion and fears death and worries about his legacy, whereas Harrison Ford is living a crappy almost robotic life.
Rachel is also weaker than the other replicants. Presumably the mnemonic replicants all have to be human-spec at least physically, otherwise they'll start noticing the cracks in their reality (why can I lift four hundred pounds when nobody else can? I don't even lift!).
Fun review thanks! Although I think the idea that Warsaw is the only modern city (when you did not even visit Gdansk or Lodz) is a little underdeveloped.
Here is how the Philadelphia Science Fiction Society-related book club votes on the next books to read.
There are usually about a dozen people at a meeting. At the end of the meeting, people nominate books, and say a few sentences about their book. A list is made (maybe a dozen books), and people can vote for as many books as they like.
Frequently, there's a break point where some books are clearly getting more votes. Maybe 6 books are doing better than the rest.
In the second round, people get three votes to distribute. If there isn't a clear winner, and there usually isn't, people get one vote for the last round.
If there's a tie, those books will be read in subsequent months.
People sometimes campaign for months, even off and on for years, to get a book chosen.
Books that made a decent showing will be nominees in the next month.
This style of voting is easier online, since lists of nominees can be put in chat.
Is there a list online of past books you've read. Since you recommended T. Kingfisher's "A wizards guide to defensive baking". I've been reading everything she has written... and I'm almost finished, more good authors please.
The book club recently discussed China Mieville's _Un Lun Dun_ on my recommendation, and most people liked it. I'm not describing it unless people ask since when I first read it, for some reason I decided to read it without knowing anything about it (not even reading the back cover), and that might be the best way to read it.
I'm currently rereading the Saga graphic novel and enjoying it a lot. Science fiction in the fine old tradition of just making things up. Sex, violence, pacifism in a time of war, spaceships, ghosts, psychic powers, old relationships coming back, disguises, confrontations. If it's the kind of thing you like, there's plenty of it.
I think T. Kingfisher's _Thornhedge_ was spectacular, but has more horror than _A Wizard's Guide_.
> I guarantee I will never respond to your comments urging you to call a phone number in South Carolina.
I am pretty worried that enough people took this at face value for you to need to point this out. Are ACX readers not impersonation-scam-savvy? I received that same "text me" comment and my immediate reaction was to scoff at how transparent it was.
I found it suspicious given there was no additional context to it, so I googled the phone number and saw it was used by a completely different handle on a different blog, and put "let Scott know his account might be compromised" on my todo list (it was still sitting there until this Open Thread, I didn't manage to get around to it, have now deleted it).
At least one person, judging by the comments from Highlights From The Comments On Fetishes. Although, said person is 72, so I suppose I should cut him some slack.
Out by the pond a couple minutes after sunrise a rainbow segment to the south.
Later, a young muskrat perfectly still near shore, half out of the water(contemplating life?) and a Green Heron a foot to one side picking breakfast from among a school of minnows. Peaceably coexisting, each with a perceived abundance and lack of territorial response.
This morning I was meditating(daydreaming!) on a large rock in an area recently landscaped by volunteers(with HOA funds). One of the women made a point of being in my line of sight when I opened my eyes. She had been nearby the two previous days. We exchanged pleasantries and I told her what a good job she had done as she walked off to her garden plot. That same young muskrat(same size) from yesterday was swimming a few feet off the shore's edge, which was about 20ft away. He caught a fish, chewed and swallowed; paused still looking my way; loitered a bit, then swam off.
"It is the nature of every grain of corn to become wheat and every precious metal to become gold and all procreation to lead to the procreation of the human race. Therefore a master says that there is no animal which does not possess some likeness to human beings.”
--Eckhart(1260-1328), Meister Eckhart Selected Writings, 1995
* Human beings are social to the core. *
[Edit/Add: "there is a great difference between being in a room that is empty and being in a room where someone is sleeping. A sleeping person is a presence in a room. They are aware of you, even if it is only a dim, vague awareness.”
Scott, please know that this number is still actively being used to scam. I asked a generic question(from an online account) and just received a response.
There's been a lot of discussion about matchmaking ('brokers') but not a lot of interest in exploring the concept further.
Well I'm interested in being a 'broker', in a non-Traditional-cultural background sort of way.
Correct me if I'm using the concept wrong but for long-term relationships there seems to be a fundamental non-ergodicity in play. That is, one may successfully ('happily') marry once or maybe twice in a lifetime but no one could do so hundreds of times. So it's not possible to skill up at finding marriages the way one *can* level up at short term dating. The best an individual can do is bet on proxy activities like micro-marriages, etc.
Now one individual cannot marry many times for the experience, but a third party, i.e. a broker could in theory successful go through the process of matching-making tens, hundreds, even thousands of times in career/lifetime. They can observe the successful pairing of many different couples rather trying observe a single person successfully marching many times. So it would seems useful For Research Purposes to spur the creation of a class of marriage brokers, in the hopes of creating many marriage-making experts.
Having said all that would it be a crazy idea to read through every dating doc I can find and then message potential couples to match them?
I'd suggest you look into the actual business of matchmaking first (and yes, it is a business). Firstly, the idea of just randomly messaging people is weird. What you usually do is you engage clients who pay you to find matches. Most matchmakers specialize in one gender. They also have a proprietary binder of people who they've pre-evaluated. Part of the pitch to both sides is that the other side will be high quality, serious about marriage (or whatever), and not flake. This is certainly a model that could work in SV and especially in some of the weirder niche communities. But I'd suggest you think about it more before just kind of jumping in.
I think it wouldn't add much value because I expect the majority of dating doc list people have already looked or at least skimmed all the compatible dating docs.
So you think can get the information on the successful matches and then train pattern recognition software that can identify good pairings between dating docs?
This is not true. In fact it's the opposite. The number of men seeking casual sex vastly exceeds the number of women seeking casual sex. But the number of women seeking serious relationships vastly outnumbers the number of eligible bachelors.
I'm sorry. I should clarify. My prior statement is not correct. The number of men seeking women *doing this dating docs thing* vastly exceeds the number of women seeking men.
Ah, yes. This is normal though. It sounds like your meat and potatoes will be recruiting women who are willing to try out nerdy rationalist types and then setting them up on dates. Plus some dating coaching in all likelihood.
I suspect "eligible" is doing most of the heavy lifting here. According to a 2020 Pew poll on dating and relationships, equal numbers (31% each) of adult men and adult women are single, single women are substantially less likely than single men (38% vs 61%) to be looking for dating and/or relationships, and among those who are single and looking, and roughly equal percentages (78% of men and 76% of women) of those who are single and looking are looking at least in part for serious relationships. Single women are more likely (36% vs 22%) to be exclusively looking for serious relationships, though, but in absolute numbers that works out to 4.24% of all adult women and 4.16% of all adult men looking exclusively for serious relationships and an additional 5.65% of all adult women and 10.6% of all adult men simultaneously looking for serious relationships or casual dates.
As for "eligible", single women who report been having trouble finding people to date are a lot more likely than the corresponding subset of single men (56% vs 35%) to cite "Hard to find someone who meets their expectations" as a major reason it's hard to date.
Summary: if you're a single woman looking for men to date, the odds are good but the goods are odd.
This statement is a good example of statistics being technically true but misleading. If you limit it to under 40s or never marrieds then the number of people looking for a relationship evens out. And once it evens out the 50% increase in women looking for serious relationships means they outnumber men. Likewise, older women are significantly more likely to report difficulty dating.
Further, women (regardless of whether they're on the dating market) are both more likely to report it's difficult to find someone worth dating and that it's hard to find people looking for a committed relationship.
You're at least partially right. I was looking at the top-line numbers and didn't look closely at the age bracket cross-tabs. Among those who are single, the under-40s are roughly equally likely to be looking. Looking at other cross-tabs, I suspect a lot of the older women who are single and not looking are widows.
However, the percentage single in younger age brackets is very unequal. Especially in the 18-29 age bracket, where 51% of men but only 32% of women are single. 30-49 also has more single men than women (27% and 19% respectively), while 50-64 is pretty close to equal (27% and 29%). It's only in the 65+ demographic that many more women than men are single (49% vs 21%), consistent with my "confirmed widows" hypothesis.
Take these two effects together, and I think my conclusion that more men than women are both single and looking for serious relationships probably still holds for younger age brackets.
I'd like to insert a "X solves this" or "there's no alpha left in X" joke but I'm coming up blank. Perhaps something about changing incentives for polygyny to even out the "can't find" ratio.
I could see that. I have my doubts there's some huge pool of hidden women dying to marry these nerd guys (a group in which I include myself, mind you), but I wish them well. It's like the lady trying to beat ovarian cancer; you don't think she'll win, but you wish her luck anyway. Life is cruel.
It doesn't have to be that they are dying to date nerd men. They just have to have a strong enough dislike for the mainstream dating market. I believe that sort of hidden pool does exist.
"There’s a scam where an account pretending to be me is replying to comments here and then immediately deleting the replies; people are getting “replied to your comment” emails that suggest calling a number in South Carolina."
You know I had to do it 😁
The Scott Alexanders of South Carolina:
(1) "Scott Alexander is a published author, a coach to entrepreneurs and executives, an accomplished speaker, and a strategy/leadership consultant."
(2) Scott Alexander, Maintenance and Engineering Manager
(3) "Oconee Station State Historic Site is home to one very special ranger, Interpretive Ranger Scott Alexander"
(4) Scott Alexander, Pharmacist
And about 74 others, none of whom I imagine are our mysterious emailer!
So is this his doppelganger or are we not getting the full Scott Alexander story? By day, a psychiatrist in San Francisco, every other Thursday an accomplished speaker and business coach in South Carolina! 😁
I've noticed that conventional schooling can cause despair in people who aren't good at it. People can come out of it thinking they're incapable of learning a subject if they didn't learn it in school, or that they're bad at learning in general. Is there any way to handle school so that people reliably feeling that they might just need a different teacher or method of study?
One way would be to offer learning opportunities in a variety of fields. That could be adding vocational subjects, hands-on environments, etc. Someone who is bad at classroom learning may succeed more in a garage learning to be a mechanic (or a lab technician, or an accountant). We would never be able to cover every subject that way, but if we offered a fair variety, it may help people learn about and seek their potential outside of classrooms where they struggle.
There are clearly additional ways to help people find their potential, but the question is whether any such attempt could scale. Parent involvement matters a tremendous amount here, but not every parent is interested, capable, or available. I doubt any type of larger scale schooling could tackle the challenge. Small numbers of people could get a private tutor or go to a specialized school, but that's going to be very limited. Large environments hit a wall trying to customize too much and the costs skyrocket.
- Most people who struggle to learn in school are actually bad at learning (which includes being genuinely low in intelligence), especially if stuff like dyslexia is identified and controlled for
- The number of people who experience a chronic struggle to learn because of a teacher or method of study is very low
- Even if some alternative form of study or instruction has the potential to work for some people, these people are likely to not have very successful careers in cognitively demanding fields, because the employment environment (i.e. 'the real world') doesn't and cannot make accomodations like this (because its not a controlled environment with crafted training material - you're dealing with the economic system at large) and you need to be able to succeed with whatever is in front of you, and most very successful people are able to adapt and thrive under non-optimal conditions
These are unlikely to be universally true, but I think this is more correct on average than most alternatives.
The above is especially true for the very common liberal trope of the student who struggles with subjects they find boring but supposedly does well when the material is more 'fun'. Even if this actually happens (and they aren't just mistaking enthusiasm for actual academic comprehension), again, smart successful people work hard and learn things things regardless of how boring they are, so we should assume both that these 'bored' students are both genuinely less intelligent, and that they will struggle with any cognitively demanding job, because the real world is rarely 'fun'.
I think learning styles exist, or at least it's easier for me to absorb what I read than what I hear, though it helps with listening if I have something to do with my eyes and my hands.
Oh, I get it. You restate my position using inflammatory language, that way you get to make it look bad and dismiss it out of hand. I guess that's much easier than having to make an actual argument.
The heritability of intelligence is as much as scientific fact as countless other things you strongly believe in. Denying this is not scientifically defensible.
Why would you needlessly change his wording, even though you put it in quotes, in a way that made it significantly more inflammatory than it was? "Bad at learning" was his language, which you exchanged for "hopelessly retarded." Feel free to disagree with him, but your approach added nothing to the discussion but heat.
Sorry, do you simply not think its possible even in principle that behavioral differences between populations can be largely heritable in nature? Or are we supposed to pretend that this can't be the case because you find it personally offensive?
You people act so damn intellectually smug despite the fact that you repeatedly demonstrate your scientific ignorance and cannot beging to provide coherent alternative explanations for these things that aren't either wildly dogmatic or at odds with empirical evidence (like thinking either poverty or inequality explain black people's insane homicide rate - they can't).
And what exactly does "called out" mean? I've had people like you get offended that anyone would base their view of the world on scientific data rather than egalitarian ideology, but that does nothing to change the actual underlying reality.
If he's willing to use milder language and not be inflammatory, we should engage with his ideas as presented. The same way many of us do (or should) respond to your posts despite previous inflammatory language you've used.
I think that generally, people have accurate judgments about their ability to learn. Yes, impostor syndrome is a thing, as is unfounded hubris, but generally there's an extremely strong correlation between perceived ability to learn X and actual ability to learn X, especially if the student has already spent a bit of time trying to learn X.
So at what point do we start trusting them? We are talking about changes to the school system, so presumably by the time they come out of it as adults it is too late...
I think that this is wrong for math in America. I have tons of students who are utterly convinced that they are awful and incapable of math, when really they are not.
You're asking if it is possible to do badly in school but still come out feeling confident about your ability to learn, I think?
What about another scenario: doing great in school and realizing you're really only good at gaming that system? That you have no curiosity or ability to learn for the joy of it?
This is very common (your last point). U.S high school computer science teachers are, in many cases, so bad that they turn kids away from this lucrative field.
There are two separate questions here: (1) is there a different `one size fits all' solution which would work for a larger fraction of people? (I presume there is none which would work for everyone), and (2) if we customized a schooling solution to the individual, could everyone learn [x] (with the exception possibly of people with severe handicaps or zero motivation)?
On (1), we could start by looking at different public school systems that actually exist in the world, and pick the one which works best according to our lights. On (2)...it seems likely that customization will work for a larger number of people than one-size-fits-all, but it will inevitably also be a lot more expensive.
> (1), we could start by looking at different public school systems that actually exist in the world, and pick the one which works best according to our lights.
Trying to pick out actual causal effects of schooling system differences themselves is nigh on impossible, because you face the hopeless task of trying to account for inherent differences in student ability, and these differences almost certainly dominate in explaining any academoic variance over any other factor.
Obviously between countries you're dealing with the substantial population-level differences in intelligence, but this effect exists even within countries. In the US for example, people will point to charter schools with XYZ teaching style as evidence that that style works better, but the problem is the student makeup of these schools is highly non-representative. They are disproportionately in certain locations (states, citites neighborhoods etc) and even though they're usually not academically selective they disproportionately attract students of parents who are likely to be higher in intelligence, conscientiousness etc. The idea that we can get any meaningfully valid insights from strictly observational education studies is wildly optimistic to the point of ignorance.
If you want education insights, you *need* controlled trials using intellectually matched samples of students. I would bet every cent I own that you could take a group of sufficiently high IQ students and put them in any education system that exists around the world and they'll still beat the pants off the average student - if you can't control for this, if you can't control for this, you can't gain any useful information about the effect (if any) of differences in education systems.
Okay, that’s a fair point. But realistically we can’t do controlled studies either. Unless you’re secretly dictator of some place anyway. So what does that leave?
I think that a good first step would be to decouple teaching from grading. The thing we have now (the one who teaches is also the one who assigns grades) is in my opinion a horrible conflict of interest. It is okay for teachers to give *feedback* to their students. But they should not be the ultimate judges of which of their students get *certified*.
So, in my vision, there is a school where you go to learn, and then there is an "exam room" where you go to get examined and get grades. You can practice the exam room online, but to get your actual grade, you need to actually go there. Oh, and if you fail, you can try again. That should remove a lot of stress.
If this became normal, I suppose it would allow for greater freedom of schools, because the fact that your students pass an independent exam should be a strong argument that your system works okay.
> if we customized a schooling solution to the individual, could everyone learn [x]
In my vision, you could take any exam at any age. Are you 30 years old and want to do the test of addition of one-digit numbers? Sure, go ahead. I mean, if you think about it, why not.
This, I suppose, could increase the overall success at learning, because I suspect that in many cases the problem is simply the slower students having not enough time to master some topic, or simply not caring enough about the topic (which can change in a few years). Though, it would not be literally everyone, because some people would still not care, or would get too busy with life outside the school.
A related criticism, in reverse, is that when the teachers have to prepare students for an outside examination, they "teach to the test" at the expense of better learning for the students. Whatever the test asks, is what gets the emphasis. This happens now with standardized testing. It would likely happen much more if all grades were from an external examination. If the student would benefit from learning something outside of the tested material, you've greatly increased the chances it would just be skipped. There's a high probability that this would happen, as not every learning area can be tested (and some are very hard to test, like art and music), and there isn't enough time to test even on what could be taught now.
I have heard about the problems with "teaching to the test", but I am not convinced that they couldn't be solved by designing *better* tests. I am probably missing something, but sometimes the solution seems kinda obvious.
Problem: Teachers only teach the parts that are included in the test.
Solution: Include *all* parts. (Would it make the test too long? Then use a random selection in the test, a different random selection every day.)
Problem: Teachers don't teach anything besides the mandatory curriculum.
Solution: Include optional questions that allow you to gain *bonus* points (and potentially more than 100% on the test).
Problem: There is no reason for gifted students to learn anything beyond the test, if they already have 100%.
Solution: Make special "advanced" tests for the gifted students; or at least allow them to take tests for higher grades. On the diploma, list the advanced tests they took successfully.
Problem: The tests are too hard.
Solution: Change the scale so that solving e.g. 50% of the problems counts as "success". (Or maybe, have "easy" and "hard" problems, and require 80% of the easy ones, and 20% of the hard ones.)
Also, in context of American education, I believe that testing was just a part of the problem with "No Child Left Behind". A greater part of the problem was that schools were evaluated (if i understand it correctly) by the test scores of the least successful children. That is, to make a simple example with three kids in the classroom, an outcome [10, 100, 100] was considered worse than an outcome [11, 20, 20] by the official metric. So the teachers optimized for the metric by drilling the least successful child to the test, and ignored everything else. But this was the problem of the metric rather than of the tests themselves. You could have the same tests, and instead measure the maximum or the average, and that would give teachers completely different incentives.
I don't disagree with you, other than to say that it's very difficult to make tests that meet all of your criteria for the wide range of students that would need to be tested. You are correct about some of the testing problems in the US and how teachers try to get around it.
Nobody is trying to make bad tests. In fact, significant resources are put into making very good tests. The problem continues in a way to suggest that it may not be possible to make a test perfect, or even good enough to avoid the issues. And that's in subjects which we think we can readily test, such as Math and English. In subjects that are harder to test, or where "correct" is more subjective, I'm not even sure it's something we could really effectively try. Even in Math there's getting the right answer or knowing how to do the operation - two separate things that you would test differently and score differently. And that's one of the easiest to score of all subjects.
I agree 100% and have been thinking about various formulations in this vein for a while. The any-age thing is particularly important, I think. Multiple, arbitrary learning paths can be offered by any number of outside bodies. It's the exam's business to ensure that you know what you should do, and as long as you do, who cares what path you took to get there?
I can see that exams might have to become a bit more rigourous, to deal with this greater variability. But at the same time they can maybe do that by becoming more drawn out affairs. I have imagined the "exam" being an app that watches your work in a certain field, or effectively an averaged high score of your performance in a computer game that analogises perfectly with the subject domain.
I also believe that the exam should be a mirror of the conditions of the real life application of the subject. So, the exam for bomb diffusing should be a high stakes, stressful nightmare of an exam. The exam for talking about English literature, less so.
One question I keep coming back to is how could the examining body be set up and brought into being? It has to happen unilaterally because there's no question of waiting for any existing state-controlled education system to do it.
I did think that, the value of university degrees having fallen a lot, and (to my knowledge) nothing else having arisen to replace it - here's the opening. All a new accreditation system has to do is be a better guarantee of quality than existing degrees.
I worry how the profit incentive would conflict with the intended goals. Like, if you set up the exam system, you cannot give everyone the perfect score. But everyone who gets a less than perfect score has a selfish incentive to support your competitor instead, if they give them a higher score. And maybe the employers would prefer to use the system that is more tough on students (other things being equal), but they will ultimately choose the one that has much more students. Also, only the university exams or trade school exams are relevant for employers, but I would also like to fix the elementary and high schools.
To certain degree this already happens at schools. For example, when you have students rate their teachers, then (other things being equal) teachers who give better grades get better ratings. Similarly, when students choose a school, many of them prefer the one with a smaller chance of dropping out, so the schools often keep students beyond the point where it would be reasonable to kick them out. Also, one of the reasons the tests are generally unpopular, is that many popular kids fare worse in the tests than they would at a more personal exam (where it is easier to apply charisma or bribes).
For starters, I would limit the organization to testing school subjects (i.e. no bomb defusing) and even there those that can be tested by a computer (i.e. no gym or art exams). Otherwise it wouldn't scale well. But there would be a human in the classroom to make sure that the students are not cheating. (The online anti-cheating systems are either too weak, or have too many false positives.)
Maybe in the first iteration, it could be a service offered to homeschoolers or kids in various alternative schools, to offer a feedback to parents how their children are progressing compared to the traditional education. Or maybe a better first iteration would be to provide entrance exams for high schools and universities. And from there, I would try to expand to test everything. (Which is not the same as if I optimized for profit, because then I would stick with the profitable parts only.)
I'll admit, the profit motive is not something I'd considered. But I don't see that it's necessarily a perverse force. As you say, employers will favour exams that are tough, employees will favour exams that are easy - but employers are the ones making the decision. I think the incentives align to make the exams tougher not easier.
Even in the maximally inconvenient case that there's already a large competitor offering easy exams to get lots of students through the system: the rational move for the employer is to prefer certs from the tough nut exam whenever he sees them, and grudgingly accept the easy students if no hardcore ones are available. So the tough nut body grows in any instance.
I'm more in favour of exams getting tougher than easier, but really my ideal is that the exams are purely *objective* - they somehow accurately grade skill in a way that doesn't change at all over time. You should be able to compare yourself to any other person and have confidence that the only confounders in the score are ones which also confound your ability itself. I have no idea what this looks like - it definitely won't be a single scalar value anymore - but that's the target I want to be thinking about.
I also think about education at the primary/secondary school level, as well as topics that come earlier, later, or are absent entirely. The reason I focus on replacing university qualifications is because that's a starting point with a clear market need, which if exploited properly could gain funds and recognition that would make it easier to turn around and address the other stuff.
My idea of a maximally inconvenient case is more like this:
Your competitor offers exams that are somewhere in the middle -- not too easy, but not too hard either. They can be gamed by teaching to the test, but most people don't care, because they consider learning to the test an evidence of intelligence and conscientiousness, which are desirable qualities for a future employee. Also, students get extra points for attending the right kind of school (not directly, of course, but more like "if they attend the right kind of school, their teacher is allowed to assign a part of their score based on their school work"), and again most people don't mind because attending the right kind of school sounds like a good thing so it's okay if it is rewarded. Etc.
Your competitor will be more popular than you. Kids from the right kind of school have an obvious incentive to prefer the system that gives them an advantage. Employers want the system where the kids from the right kind of school are. All other students want the system that the employers want, and also where they are associated with the kids from the right kind of school. (Kinda like "I am on Facebook, because everyone is on Facebook; it doesn't matter whether a different website has a better user interface".)
In other words, a system that takes over your niche, but also sucks, but doesn't suck enough to make people leave it, so they stay there for the network effect.
(This may sound too pessimistic, but look at the current educational system and all its disadvantages and all its unfairness... and yet, most employers prefer students who pass the system, and most students are willing to get into huge debt in order to participate in the system, because it seems like they cannot avoid it. If something like this happened once, it may also happen the second time.)
> because I suspect that in many cases the problem is simply the slower students having not enough time to master some topic
These students are unlikely to ever 'master' these topics - they may eventually get to the point of being able to pass an exam and little more. And it's unclear on why these students even need to be taught anything beyond the most rudimentary literacy and numeracy abilities - they will never be capable of holding a remotely cognitively demaning job and it feels like much of the motivation here is the warm fuzzy feelings of seeing the strugglers 'succeed' in education and not feel like hopless idiots. But as far as real practical outcomes go, it makes very little difference (even in one in 10,000 of these students somehow turns out to be genuinely intelligent and capable).
Passing the exam is a binary choice, so the question is where to draw the line.
Traditional education puts the line quite low, simply because you want the entire class to move forward at the same speed. So if someone has a 50% success rate, you say "okay, that's good enough, let's move to the next lesson". Of course, if someone is 50% good at lesson one, you can hardly expect them to do much better at lesson two, if they are so bad at the fundamentals.
A better system would move at individual speed (perhaps because it would use a lot of computer education), so you could put the bar higher and require 90% success rate to pass. With the consequence that many would pass later, and some not at all. However, this would be slightly balanced by the fact that people who passed the lesson one later would be better prepared for the lesson two than they are today.
And yes, some people would only get the rudimentary literacy, but I think that even they might be better served by a system that spends 100% time on teaching them the literacy, than by a system that tells them a lot but ultimately teaches them nothing.
It feels like you are looking at students as in one of two categories - capable and incapable. Or at the least taking a certain cutoff point and saying anybody less than that is just dead weight. Neither of those things is true. There are lots of people who are not intellectually gifted but can consistently and effectively learn more than rudimentary levels of various subjects.
A while back Scott posted a chart on various professions with the IQ range of people who worked in them. While Scientist and College Professor both had higher ranges than some other positions, the ranges dipped below 100 for each - meaning people performing the jobs with measured IQs at or below 100. I can anticipate your response that these people must not be performing the job well - but neither you nor I can know that with information currently available and it would be illegitimate for you to make that argument based only on your preconceived intuition.
>There are lots of people who are not intellectually gifted but can consistently and effectively learn more than rudimentary levels of various subjects.
Im not talking about "not gifted". Below a certain level of intelligence, people will struggle to learn anything beyond the basics (and in some cases even that is a struggle). And missing the odd idiosyncratic smart person does less harm than wasting years of people's lives trying to get them to learn something they cannot (and even worse, sending them to college).
>I can anticipate your response that these people must not be performing the job well - but neither you nor I can know that with information currently available and it would be illegitimate for you to make that argument based only on your preconceived intuition.
No, actually, we know that job performance is correlated with IQ. It's not intutition, it's empirical fact.
I mean, why the heck do you think anyone cares about IQ at all? It's not because there's some profound significance of shading the correct circles on an IQ test - it's because of what it tells us about the real world. And intelligence predicts job performance better than variables like work sample tests, reference checks, years of experience, and age.
The most likely explanation for "professors" with an IQ below 100 is affirmative action/HBCUs (and maybe community colleges having positions called "professor" or something silly).
How much more would you pay to see a doctor of IQ 130 rather than one below IQ 100? I would pay many times more ad I suspect you would to, even though we supposedly "can't know" which is better.
"The thing we have now (the one who teaches is also the one who assigns grades) is in my opinion a horrible conflict of interest."
That's a very good point.
Even setting aside whether it improves learning, it might make the resulting grades more generally useful (to the student, to employers, to evaluating teaching methods, ...) and believable.
We have this in the UK. Exams are written and graded by independent national exam boards. And you can do them as an adult if you really want (I think you have to pay an admin fee).
The American system seems like a bad idea to me. You could be unfairly penalised (with lifelong knock-on consequences) if your teacher hates you, or if your teacher is a bit clueless and doesn't understand the subject they're teaching (I've had teachers like that, but at least they weren't grading my exams).
Is teacher-grading of everything really "the American system?" Surely it's different in every state? And the SAT, at least, is externally graded right?
Meanwhile, when you say "exams" in the UK, are they 100% of the grade for each class? And I'm assuming they're only in the final year of high school? Or multiple years?
Unless everyone is already familiar with the intricacies of different systems, I really think these kinds of clarifications are necessary for a debate like this to make any sense.
I have bad memories of my Grade 8 "Language Arts" (what they called "English" at the time) teacher, who got mad at me for writing that Animal Farm was a metaphor for the Russian revolution.
I tried to point out the Lenin, Stalin, and Trotsky figures, and the proletariat and the capitalists, to no avail.
In that case, it would have been good to have a better-informed 3rd-party do the marking.
(She also got mad at me for not knowing about "The Mod Squad", a TV program I'd never seen.)
Two hypothetical one size fits all solutions is to build students' respect for their won capacity to learn. This may be difficult in an environment with a lot of emphasis on grades, but I recommend exploring it. Maybe tell stories of people learning in different ways and at different times. Maybe encourage students to think about their own experiences of learning.
Is customization-- maybe just not expecting students to keep trying at things they're very bad at-- very expensive? This question might deserve some thought.
Do you have any evidence that the variance in academic ability is meaningfully explained by differences in how they learn things (especially in a way that could be applied systematically to public education e.g. not having all students indvidually tutored all day)?
The old view of different 'learning styles' (e.g. people being 'auditory learners' or 'kinesthetic learners') specifically has resoundingly failed to be empirically verified, so whatever it is you think leads to these learning differences will necessarily have to be very different to these types of 'learning styles'.
I'm not sure you're addressing her primary point. There are myriad reasons a student who is otherwise intellectually capable could be frustrated in some particular subject and lose interest. That this has been used sometimes in other contexts to whitewash the fact that not all students are equally capable is not relevant. Any one of any intellectual capacity is better off at least attempting to be a lifelong learner than not attempting this. So discouraging people even inadvertently from whatever curiosity and capacity for growth they are capable of is a mistake that we can presume is preventable to some extent.
>I'm not sure you're addressing her primary point. There are myriad reasons a student who is otherwise intellectually capable could be frustrated in some particular subject and lose interest.
Sure, but do you have any estimates for what proportion of observed intellectual ability variance results from this?
>So discouraging people even inadvertently from whatever curiosity and capacity for growth they are capable of is a mistake that we can presume is preventable to some extent.
Sure, but this is very different to what we're really talking about here, which is systematically changing education for the sake of people who are allegedly struggling to learning style difficulties.
I was taught sentence diagramming back in the sixties. It was a tidy little thing that I wasn't bad at and I didn't mind it, but it seemed to be of no use. Since I had very little drive, I didn't care.
Recently, I've had a couple of friends say that they consider sentence diagramming to be important, and I've thought about what it might be good for. Maybe it's useful if you want to write a complex sentence. Maybe it's useful for sorting out other people's complex sentences.
So, were you taught sentence diagramming? Do you think it was useful, and if so, for what?
I do think it's important for people to know how to write clearly, and I don't know how it can be taught. What do you think helps?
I've posted this on Facebook, and gotten a hundred or so accounts of life with and without without sentence diagramming.
There are a good many people who found sentence diagramming to help with learning languages or clarifying writing, and a good many who have become professional writers without having learned sentence diagramming.
This raises a question for teaching. *Perhaps* all students should get some exposure to sentence diagramming, but how do you identify the students who won't get anything out of it relatively quickly?
Diagramming sentences felt completely redundant to me, I think because I'd already learned a bit of Latin by then. Possibly any foreign language would work as well?
When I was about 11-12 (around 1970), an external teacher came into our Christian Brothers school to give us a few lessons on parsing (I don't know what sentence diagramming is, but I assume they are related). I found it quite eye-opening.
Diagramming forces the student to break sentence structure down to its components, so it's not only useful, but essential to understanding language. If one does a transcript of the spoken word in mass media, one can see how popular disjointed, irrational and fallacious language constructs have become. Just diagram the sentence.
Except that diagramming seems to be routinely taught only the US. Do you want to argue that people from other countries are incompetent in their own languages?
I also don't know whether a background in diagraming causes people to make their speaking more organized. It might, but you're the first person who's brought up how people speak, a large topic in itself.
In my facebook discussion, there seems to be a higher proportion of professional writers and editors who didn't learn diagraming. Maybe I'm missing something, but their writing seemed as clear as the writing from people who did have a background in diagramming.
To be clear, for those of us who didn’t learn it at school (it seems to be a very US thing, and of a particular time) are you talking about what Wikipedia calls the Reed-Kellogg system? As a linguistics student I learned Chomsky’s X-bar method, which I occasionally find useful but wouldn’t normally use in reading or writing prose.
Sorry, didn’t see this earlier: I am referring to the syntax tree diagrams mentioned upthread. I’m not sure I agree that there’s not a convenient system to teach these, though: they seem to be a pretty standard part of most syntax curricula.
One thing to be aware of is that there are many specific things that all fall under the umbrella of “sentence diagramming”—I hated the specific instance I did in high school English class, but loved the specific instance I did in college as part of my linguistics minor—and differences in opinion may be because different people interacted with different versions of sentence diagramming.
Some people just need to hear that a 1-D string of tokens can also have some sort of hierarchical, tree-like structure. For others, this advice is so self-evident and banal, it's hard to articulate. Trees can be useful to interpret garden-path sentences, but it's probably easier to appreciate when dealing with larger structures. E.g. using an octree for collision-detection.
If I were a teacher, I would just share a quick, shallow lesson with pathological examples using a broad variety of representations, and then move on.
I *loved* sentence diagramming. It was one of the few times in school outside of math classes where I really had the feeling of learning something about the structure of things, in this case the structure of language. Once it was explained it made sense, but it had never occurred to me til we did that unit that there was a logic to how sentences were built. Learning about the double helix in science class thrilled and fascinated me the same way, and so did learning how to turn word problems into equations (you built this little machine and turned the crank and out came the answer like a sausage.) After we did the unit on sentence diagramming we did one on figures of speech, and also on some related categories — idioms, slang and dialect — and I lovde that too. I was as thrilled as I would be now if somebody gave me a brand new high-end car.
I do think sentence diagramming was useful to me. I enjoy long complex sentences more — things like the sentences in late Henry James novels — because understanding how they’re put together gives me an extra dimension to enjoy. Some of them are like cathedrals. And I think sentence diagramming has helped me write clearly. There are lots of little errors, things like changing tenses in mid-sentence, or tacking on a clause in the wrong place so that technically it applies to something other than what you mean it to, that do not really confuse the reader, but give what you wrote a sort of blurriness. The reader has to push through this little thicket of things you don’t mean in order to get to what you do mean, so they arrive at your meaning a bit tired and irritated.
One of my happiest intellectual moments in school was finding out about polar coordinates. It was very satisfying to know there were two such different ways of describing locations.
The best relatively recent novels I can think of for long complex sentences are _Jonathan Strange & Mr. Norrell_ and (less recent) _The Worm Ouroboros_.
I did not learn sentence diagramming in school. My mother (MA in English) made me learn around 4th or 5th grade. I did not understand why, until I got to high school and grammar classes started getting a bit more rigorous. Sentence diagramming taught me a) basic grammar terms and concepts and b) a way to translate internal sentence structures into visuals.
The second of these was incredibly helpful once I started taking German classes.
I then went on to teach writing as an student instructional aide in college, and discovered that sentence diagramming was the actual reason behind what to a lot of people looked like my "superpower"--I could glance at an essay prompt and tell you exactly what needed to be written, while others would consistently not answer the question being asked in the prompt. Apparently it is very easy to miss entire words or phrases in a long prompt, and sentence diagramming seems to teach you to stop and see every word? That's my best guess as to what happened there.
I learned a bit of it. I didn’t mind spending a little time on it but I don’t feel I learned much; it didn’t enhance my understanding of grammar, which was not hard for me. Maybe there are some people for whom it makes ”getting” grammar easier. But I think diagramming is something that can easily become pedantic busy work.
It certainly can be something "to keep the kids busy". I was surprised at how many people found it was helpful.
Someone noted that people who are good a physics are more likely to be good at diagramming sentences, though I don't know whether they're more likely to find d.s. useful.
It think it makes a bit more sense when you start learning foreign languages (because then you use more of explicit reasoning), but still not very useful, unless you happen to be curious.
(To answer your original question, yes I was taught sentence diagramming at school.)
Yes, I think the same! In fact my answer here compared learning sentence diagramming to learning to construct equations from word problems. I enjoyed the daylights out of both.
I learned sentence diagramming. I think that firstly, it helped give me a more concrete idea of how grammar worked, which was helpful both for understanding other language's grammar (I could compare it to English) as well as for my own writing (deciding between who or whom depends on identifying if you're working with a direct object or not, for example). Also, as a math teacher, I find it helpful for deconstructing word problems; I don't go into super-deep detail with my students, but it's still helpful to be able to raise points about nouns usually being your variables, and connecting phrases like "is a function of."
You said "here are some situations in whch it could be helpful." The missing piece is describing how senfence diagramming is helpful, along with a concrete scenario, including application and before/after comparison.
The facebook link seems to require a facebook account.
I was just having a discussion about this on Discord, but has anyone tried deliberately practicing mental visualization for an extended period of time to try and get better? I suspect being able to visualize things in your mind is a skill which can be trained. Do any children development programs deliberately try to practice this skill?
Imagine a ball and keep duplicating it until you're no longer able to create more balls. Personally when I do this there appears to be a point at which my brain starts to lag. Furthermore, despite my best attempts, it feels like past a certain amount my brain is perfectly willing to pretend that I'm still generating large amounts of balls but it just avoids filling in all the details. So despite my best effort to continue generating balls, my imagination just straight up lies to me, which seemms like an interesting observation.
I was going to ask about visualizing a larger field-- my default is a smallish area in front of my eyes, and then I realized I can visualize a bigger field/more locations just by trying. The images aren't brighter, sharper, or more complex, but it feels good to have access to greater variety. Even a very fuzzy landscape all around me is nice, and the pleasantness might be an incentive to work on improving other features.
I did a course with AphantasiaMeow(who has rebranded since but I’m sure search will find anyway) - a few separate hours. It worked in that I got some visual experiences I hadn’t had before. And I was left with exercises that im sure will improve my imagery if I was motivated to do them. The basic technique is to describe verbally visual details a lot, with various exercises around that.
I’ve also started trying to capture my dreams by describing them and I think they’re maybe getting more visual as a result.
In a recent post Scott wondered about parents “who think it’s old-fashioned to get pink things for their daughter”.
It’s generally reported that pink for girls and blue for boys was a new post war trend, and the opposite applied pre war. Is there much or any evidence that this was widespread at all, either way?
All I can see for Victorian society is that young children wore white. After that there’s no obvious trend, not amongst the kind of people who got family portraits (ie paintings) done - but these would the only kind who would care. The children of the poor wore whatever was handed down.
I have some dim memory that red was seen as a vigorous, violent color, the color of anger and blood and war. While blue was seen as a calmer, cooler, more refined and sedate color. So I can see how those might manifest with typical gender roles.
Possibly the fading (to pink and light blue) is something specific to children?
During the First World War, Scottish regiments wore their kilts into battle, as was their ancestral wont.
When the Germans deployed mustard gas, the Scots were uniquely and painfully most affected.
So when WW2 rolled around, the top brass insisted that if Scotsmen wanted to continue wearing their kilts, they must also wear a pair of specially designed mustard-gas-proof chemical lined pants.
These pants (and for burger-eating readers, this word means "underpants") were a very ugly bright pink, and the Scots refused to wear them.
So now Scottish soldiers wear trousers like everyone else.
I expect this satisfactorily answers your question.
There's very little evidence for this - it's one of those things that get repeated so much that everyone just assumes it's true without being able to point to any genuine evidence to support it (much like the myth that white women have been the main benificiaries of affirmative action).
This of course does not imply that there's anything inherently feminine about pink. But as long as pink is associated with femininity I think it's healthy to buy pink things for your daughters, and that 'genderless parenting' will raise a generation of kids even more messed up in the head than zoomers are.
This is it – the key takeaway from understanding that the colors are arbitrarily chosen is that the specific colors *don't matter* and that what matters is *establishing a sex-differentiating color schema* for small children without much sexual dimorphism and then *universally adhering to that schema* so that all people can readily identify the child's sex and treat it accordingly, socializing it into its sex properly and helping to avoid weird nonconformity scenarios down the line.
Removing the artificial dimorphism is destructive and wrongheaded, basically a life-hating behavior bent on thwarting a deep human instinct.
It is a good thing to feel like you fit in and belong. Sex roles are a significant part of that system, and, often, all bucking the system gets people is lonely and depressed. I personally no longer support extreme individualism, such as rejecting gender norms. If a young woman wants to buck the trend and behave in aberrant ways that aren't hurting anyone, great! We have individual freedoms so she's allowed to do that. We just need to make sure she knows and understands what trends she's bucking in the first place so she won't be confused and lost if it ends up making her miserable. There are lots of perfectly happy and healthy LUGs* and similar, because they, very importantly, _knew how_ to regress to the average after college.
"It's okay for a woman to behave in gender-non-conforming ways, so long as everyone around her constantly tells her how weird and non-conforming she is" is not the grand gesture of acceptance you think it is.
Like, I don't *think* you meant to justify bullying kids who don't act manly enough for their peers, but that's what your argument does.
No, it isn't. A simpleton could use my argument to justify that, but he'd be wrong to do so, because bullying is bad. I absolutely reject the idea that one shouldn't make an argument, regardless of how true it is, based on the fact that any given bad actor could use it for bad reasons. I've seen that reprehensible line of reasoning rear its ugly head too many times recently.
Societies need structure to function, and you seem unaware (or uncaring) of how much our present day mores are interfering with that.
If the choice is between the dissolution of the structures that maintain and sustain society, or reverting to a higher level of penalisation for nonconformity, the rational choice is to bring back bullying.
Being made to feel weird while nevertheless being left to go about your business in a classically liberal society is an incredibly generous deal on society's part. You will not be offered that same deal by any society that arises to replace ours if it fails.
If you are a weird and nonconforming person, your priority should be trying to shore up and perpetuate the uniquely accommodating world you live in, making whatever compromises you have to along the way.
You should not be asking for better and better treatment while the whole edifice groans underneath you.
Depends what you mean by "that argument." Some of what I said doesn't directly apply to religion or genetics or sub-culture, so I don't know what you're asking.
You mean, is it a good thing for Hasidic Jews to wear characteristic garb which sets them apart from the multitudes? Almost certainly yes, this doubtless improves their group cohesion and identity even though it's actually completely arbitrary (the costume is just what happened to be fashionable clothing in 1830s Odessa or something IIRC, when-/wherever it was the founder of Hasidism lived), and it doesn't hurt anybody else.
If fewer people are socialized to be in clear sex-based groups, then they are more likely to be confused and feel out of place about their sex. This would naturally lead to more people feeling dysmorphic about their sex. This feeling is associated with very high levels of mental health problems and suicide. We should very much work very hard to help as many people as possible to not have this problem.
In my experience, it's more common that this happens when children socialized in clear sex-based groups encounter groups who are not socialized in clear sex-based groups (or vice versa). I.E. Sending children to all-girls or all-boys schools won't help them avoid dysphoria if they encounter examples of gender bending outside of the school. However, by the same token, raising children without strong gender roles seems to have no bearing on them thinking of themselves as boys or girls- instead, it is when they encounter strong sexually dimorphic social groups that they start to shy away from identifying as their birth sex.
I'd love to see some studies done on this to confirm personal anecdote, though I think in modern times the topic is too highly charged for it to get IRB approval. Nevertheless, societies with strong gender roles inculcated into children (Japan is an example) do not seem to be seeing significantly lower rates of transgender adults than societies with less strongly-inculcated gender roles and similar laws around what counts as transgender (Finland is an example). Both hover around 1-2%. What do you think could explain this?
I don't think we have a clear enough definition of the concepts to actually measure it. I don't trust that the most interested parties can be neutral enough to properly classify whether someone is actually trans or if there's another explanation/issue involved. We're beginning to have a common agreement that the approach in the last few years has resulted in significant increases in the number of people identifying as trans, and most people see that as a problem to be fixed. Other than that, we don't really know that much.
I wrote a reply to this and then something seemed to go wrong with its placement in thread, so I deleted it for avoidance of confusion; I think you probably got it through the arcane Substack messaging system anyway, but to reiterate just in case: Yes, in this and many other cases it is good. Your failure to understand this, and why, is indicative of a broader failure to understand humans and positive socialization. The artificial markers are only arbitrary in their per-se nature, their purpose is not arbitrary. They are designed as support struts for the underlying objective reality of the two separate groups, making it easier and more effective to acculturate small persons to their future, post-pubertal social roles. Our failure to do this consistently, as a society, is the underlying reason for lots of children's alienation from and fear of puberty. As late as the '60s or so children *looked forward to* it, to becoming adults.
Thank you! Unfortunately I find myself fairly inept at expressing myself in the long-form essay mode; moreover I don't have a high opinion of the clarity of my own writing. For instance, I should clearly have expressed it as "the markers are artificial, but the groups aren't".
Oh, what the hell? I wrote a reply, then it posted in two copies so I deleted one, but both got killed. I'm sorry, I seem to be suffering from some sort of reply-based curse. If you got a copy in your activity feed or whatever, feel free to repost it as a reply to yourself and attribute it to me. I can't be bothered to retype it.
Buying clothes for someone is not overriding anything. If the person has no clothes except the ones you buy for her, and she hates them for one specific reason she has made clear, i.e. color/theme, then it would be overriding oppression to keep buying that color/theme.
This is a common story but doesn't have much evidence. It's one of those facts that seems to have been laundered into common knowledge by being too good to really examine. The citation in Wikipedia is The Atlantic. Often what's pointed to is a 1927 article saying young boys should be dressed in pink. However, as has been noted: even at the time pink clothes were primarily purchased for girls.
Pink was considered unattractive in the Middle Ages because they preferred bright, garish colors and pink was too pale. However, it does appear men wore it without feeling it was feminine. Pink probably became female coded in the 17th/18th century where it became fashionable among French court women. And expanded from there. In particular you still see men in the early 18th century wearing pink from time to time but by the middle or late period this goes away.
(Also, keep in mind that prior to the late 19th century boys wore women's clothes until the age of 5 or so. So a boy wearing pink doesn't necessarily mean it wasn't feminine.)
The fundamental point that the association is arbitrary is sound though. There are other cultures where pink is not traditionally associated with femininity. Of course, we now live in a globalized world so the idea has spread. Likewise the idea of red denoting hostile or bad is not universal. And white being positive or meaning purity. And so on.
I'm not certain that the early modern world had much of a concept of pink as distinct from light red.
The pigment for magenta, which is needed for bright pink, was invented by chemists in 1856, so before then people were limited to light earthy reds or expensive true reds. Purple was royal because it was so difficult and expensive to produce. Edit: it's possible to get soft pinks from things like berries, but I haven't heard of any evidence that they've been considered particularly feminine.
In medieval iconography Mary often wears blue, or sometimes red over blue, and Jesus wears blue over red. Possibly the feminine blue thing came from Mary in blue?
I never really believed that these associations could reverse over night, and I was expecting something like that: Bright red for boys, light blue for girls.
Here's an article on Gainsborough's late 18th century portraits, starting with the famous Blue Boy - but other male subjects are also painted wearing pink.
So pink and blue were male colours, or at least colours for boys, in the 18th century.
As an aside, when I was a child, our family doctor had large reproductions of the Blue Boy and Pink Boy hanging in the waiting room, so I knew the paintings before ever I knew who painted them or what the subjects were 😀
Another way to say pre-war, is to say pre-modern science. Much of chemistry came to knowledge after the war. Even the chemistry of hydrocarbons wasn't really discovered until the 1920s, when a wealthy automobile owner asked his chemist friend why the gasoline in his fuel tank evaporated so quickly.
It was in the 40s when we discovered how to manipulate carbon chains into polymers.
I have the impression that people used to dress babies and toddlers of both sexes pretty similarly until the boys were old enough to wear pants. (So baby boys in old pictures are often seen in ”dresses” and assumed by people nowadays to be girls.)
That's why breeching (putting boys into trousers) was such an important rite of passage, and why going from short trousers (for boys) to long trousers (for young men) was another stepping-stone to adulthood.
And there was supposed to be an Irish superstition about the fairies stealing children away, but they preferentially stole boys, so that is why boy children were put in 'skirts' like girls, to fool the fairies (until the children were past the first vulnerable years of infant mortality).
You also have to remember, before the war—read before modern manufacturing—clothing was fantastically expensive. Baby clothes were made from discarded clothing.
My Grandmother was born in 1920. My mother (born in 1940) once asked grandma how she could live with such a small bedroom closet. Grandma replied "I only had two dresses, an every day dress, and a Sunday dress."
True, but it applies also to very rich people like British royalty.
Dressing is one of those things modern people do paradoxically very badly even though/because clothing is cheap compared to before; most people wear sort of shapeless clothing.
Some online articles claim that pastel colours became popular in the 19th century, and that makes sense to me regarding the discovery of aniline dyes. Now you could get synthetic dyes that produced a range of colours and were dye-fast when washing clothing.
“The technology of dyeing fabrics was transformed in the mid-1850s when the British chemist William Perkin (1838-1907) discovered that dyes could be extracted from coal tar. These new aniline dyes became very fashionable. The first was ‘Perkin’s mauve’, followed by a variety of shades of purples and magentas, yellows, blues, and pinks. These colours were much more intense than any available from the traditional natural dyes.”
"Pink and blue arrived, along with other pastels, as colors for babies in the mid-19th century, yet the two colors were not promoted as gender signifiers until just before World War I—and even then, it took time for popular culture to sort things out.
For example, a June 1918 article from the trade publication Earnshaw's Infants' Department said, “The generally accepted rule is pink for the boys, and blue for the girls. The reason is that pink, being a more decided and stronger color, is more suitable for the boy, while blue, which is more delicate and dainty, is prettier for the girl.” Other sources said blue was flattering for blonds, pink for brunettes; or blue was for blue-eyed babies, pink for brown-eyed babies, according to Paoletti.
In 1927, Time magazine printed a chart showing sex-appropriate colors for girls and boys according to leading U.S. stores. In Boston, Filene’s told parents to dress boys in pink. So did Best & Co. in New York City, Halle’s in Cleveland and Marshall Field in Chicago."
And then, for some reason, it switched to 'pink for girls, blue for boys":
"Today’s color dictate wasn’t established until the 1940s, as a result of Americans’ preferences as interpreted by manufacturers and retailers. “It could have gone the other way,” Paoletti says."
But *why* did the preferences go that way is the interesting question. According to Wikipedia, there wasn't a reversal; pink/blue for girls/boys and boys/girls existed at the same time, and it was the pink for girls/blue for boys that won out:
"Since at least the 19th century, the colours pink and blue have been used to indicate gender, particularly for babies and young children. The current tradition in the United States (and an unknown number of other countries) is "pink for girls, blue for boys".[1]
Prior to 1940, two conflicting traditions coexisted in the U.S., the current tradition, and its opposite, i.e., "blue for girls, pink for boys". This was noted by Paoletti (1987, 1997, 2012).
Since the 1980s, Paoletti's research has been misinterpreted and has evolved into an urban legend: that there was a full reversal in 1940, prior to which the only tradition observed was the opposite of the current one. Quoting the concluding lines of this study: "In conclusion, there are strong reasons to doubt the validity of the standard PBR [pink-blue reversal] account; if anything, gender-color associations seem to be much more stable than currently believed".
There's also a long list of historical sources backing up that "pink for girls, blue for boys" was an established standard from the 19th century on.
I was thinking that the hideously sad Victorian era poem "Little Boy Blue" shed some light on this issue. Maybe blue *was* a boy's color in 1888? So just looked poem up in Wiki and discovered this info gem: "Little Boy Blue" is a poem by Eugene Field about the death of a child, a sentimental but beloved theme in 19th-century poetry. Contrary to popular belief, the poem is not about the death of Field's son, who died several years after its publication. Field once admitted that the words "Little Boy Blue" occurred to him when he needed a rhyme for the seventh line in the first stanza."
So really not relevant to pink/blue boy/girl question. But that poem has made me feel like shit every time I've run across it, and somehow knowing that "little boy Blue" is the product of a rhyme search reduces the impact of the damn thing a bit.
Ugh, good luck. That damn poem has a sort of double toxicity for me, because I've never gotten over feeling sorry for discarded toys -- some holdover from childhood where I really believed my stuffed animals had thoughts and feelings. So dead child plus dusty, neglected but still loyal toys is a one-two punch to the heart for me.
“ For example, a June 1918 article from the trade publication Earnshaw's Infants' Department said, “The generally accepted rule is pink for the boys, and blue for the girls. The reason is that pink, being a more decided and stronger color, is more suitable for the boy, while blue, which is more delicate and dainty, is prettier for the girl.” ”
Yeh, I found that in my searches as well. I am wondering though exactly how common it was. There’s two references about pink for guys - that and Time, and most literature seems to acknowledge that both existed.
However how common was the system at all? Is that it, just those articles? Was it just the readers of Earnshaw's gazette?
Finding family portraits for Victorians is easy, just google Victorian family paintings. And the Victorians definitely didn’t colour code children, everything was white, as dye ran.
By the post Edwardian age family portraits are more likely to be photographs and thus not in colour, so it’s harder to tell.
"And the Victorians definitely didn’t colour code children, everything was white, as dye ran."
That's why I mentioned the aniline dyes, which didn't fade and were colour-fast. Now they could have pastel (and brightly coloured clothes) even for children. It's difficult to tell, because the Wikipedia article does contradict the idea of "it started off pink for boys then flipped in the 1940s". It seems like pink and blue were colours for both boys and girls, but pink for girls/blue for boys was more traditional and eventually won out.
Marco Del Giudice has studied this issue: https://iris.unito.it/bitstream/2318/1853049/1/DelGiudice_2017_pink_blue_update_pre.pdf The results do not support the idea that "pink for boys, blue for girls" was ever an "accepted rule". In particular, in English-language books from the period (circa 1880-1940), "blue for boys, pink for girls" was the rule. On the other hand, in newspapers the two options were about equally frequent.
In the article there are some quotes from newspapers from the early 20th century teaching people which is supposedly right (and it goes both ways), which suggests there wasn’t an accepted rule.
"White is not a convenience choice. It's quite expensive to produce, and although it doesn't run, it shows stains really easily."
Tell me you live in the days after bluing had ceased being a thing, without telling me you live in the days after bluing had ceased being a thing 😁
Laundry bluing was used for the same reason washing detergents nowadays include "optical brighteners" in their lists of ingredients: it makes the whites look whiter.
"White fabrics acquire a slight color cast after use (usually grey or yellow). Since blue and yellow are complementary colors in the subtractive color model of color perception, adding a trace of blue color to the slightly off-white color of these fabrics makes them appear whiter. Laundry detergents may also use fluorescing agents to similar effect. Many white fabrics are blued during manufacturing. Bluing is not permanent and rinses out over time leaving dingy or yellowed whites. A commercial bluing product allows the consumer to add the bluing back into the fabric to restore whiteness."
Also, if you ever get fine, sunny days, you leave whites out on the bushes, hedges or lawns (bleaching greens) to help whiten them by the aid of the sun, as mentioned in this video about a Victorian country house:
Depends. Most cotton and wool that we grow now is white, though natural colors for both exist. Wool will also often end up creamy or off-white, so not the pure, bright white you expect nowadays.
Linen is a bit more complex, as I understand it. You can get either yellow or grey linen depending on how you ret it, (dew/ground retting gets you silver/grey, pond/water retting gets you yellow/gold) but then you can bleach it in sunlight like Deiseach mentioned.
On the other hand you can boil wash whites and bleach them. Doing that with coloured clothing for small children, which will need to be washed frequently, means the colours fade (and run, as anyone who ever mistakenly put a red item in with the whites finds out) so it's not worth it *until* you get synthetic dyes which can stand up to that treatment of frequent washing.
WRT to the fake dating site bios that got so much attention in an earlier post: I think it's interesting to observe that all the love given to the fourth one could have been predicted by the dead-simple heuristic "the most popular female dating bio will be the one with an explicit sexual come-on in it".
She was written as a woman who liked all the obscure male-typical interests of people on this blog. Scott himself said "I tried not to write the rationalist as 'obviously the best one', but I couldn't pull it off."
The only other example was the anime fan, but as actual anime fans pointed out her tastes didn't match an anime fan's (they were all the most popular without any realistic specialization by genre--it would be like someone saying 'I like Shakespeare, Jane Austen, and Toni Morrison' with regard to literature).
I think it's not just the sexual come-on, that was a specific instance of her seeming like the one who'd be the most enthusiastic about a relationship. Everyone else seemed pretty insulated.
I hadn't even considered this looking at all those polls but it's a interesting point. I feel as though the fourth one would still be the most popular if the sexual come-on was moved to a different profile but I'd be curious to see how the polls changed.
Yeah, all I remember about the fourth one is that it was attractive because it was a nerd. I'd expect sexual references to generally be seen as a negative when looking for a long-term partner.
Men are simple creatures and are easily moved by even indirect sexual references. Advertisers take advantage of this a lot, but older examples (prostitution, strip clubs, porn) abound.
Imagine that you have the superpower to make your ears glow a vivid, rather ugly, wholly unmistakable shade of green. While using this power, you are incapable of lying, or even of saying things you are not confident are true, and everyone else knows this.
I would become a famous philosopher, specialising in epistemology. Having an experimental test for whether I am personally confident something is true won't resolve the whole field, but it will definitely breathe new life into it, and people will come from all over the world to map out the limits of Melvin-approved truth.
How bright is the glow and how long is the windup-cooldown? Seems like the best use is to just strobe-light people.
How confident is confident? I can say a whole lot of things off the top of my head with perfect confidence that fades within seconds of thinking about it. Does the strobe-eye stop these or not? I generally don't rule out conspiracy theories unless I have firsthand proof they're false; do the eyes stop me expressing any viewpoint?
I think there's a minor family in the Godfather which makes a living by brokering peace deals between the major players. You could put them out of business: any family whose opponents propose anyone other than Green Ear Man as broker invites the question, what they want to lie about.
Here is the basic idea. Say that Michael Corleone, the head of one Mafia family wishes to meet with Don Tessio, the head of another in order to discuss a deal for mutual advantage. The invited guest has no way of knowing if he will be safe during the visit, and Michael's promises that he will not hurt the guest cannot be believed. There is a problem of commitment here, and without some commitment, the two will not meet.
Enter the Bocchicchio Family. When Michael invites Don Tessio, he not only promises not to harm him, but also hires a member of the Bocchicchio Family to go to Tessio's house. There, the "hostage" will be guarded by Tessio's men. If Don Tessio does not return safely, Tessio's men will kill the hostage. The Bocchicchio Family, seeking revenge, will blame Michael Corleone for the death, since he made the promise that Don Tessio will not be harmed.
True, but it eliminates a lot of risk. The Corleones might make the Bocchicchios (thanks) such a stupendous cash offer that they decide to give up their reputation and defect. Secondly you can ask GEM "so, when the Corleones outlined this deal to you did you get the feeling they were doing so in good faith?" You don't have to be perfect, just better than the competition.
Same with dealing cocaine: I have not adulterated this and have no reason to think anyone else has is not watertight, but better than the competition.
There are probably a lot of products that genuinely save people more money than the cost of the product over the long term, but which don't sell very well because "pay us now and you'll end up with more money later" just isn't a very believable claim in most cases. If you started a business selling those products, you might be able to bring products to market that no competitor could find customers for- and the selling point of "this is practically free money" would be pretty compelling.
You might also be able to solve the "market for lemons" problem- selling higher quality products for higher prices without sacrificing demand. Also, selling endorsements to other companies could be pretty incredibly lucrative if you were famous enough.
I become some sort of trusted intermediary / arbitrator / referee.
You know, there's a way in which banks facilitate international commerce between parties that don't trust each other. For this, they get to extract some percentage of the deal. This is a stable system, as the bank is basically selling "honesty" as its product and raking in cash for doing essentially nothing, so it is incentivized to behave honestly.
But I'm sure there are many areas where this kind of long-term, game theoretically founded trust isn't as easy to construct. If you want to leverage your power as much as possible, you will need to let other people do the work, but in a way that you are essential to making the deal work. It's lazy, it's profitable, and it's even pro-social.
Run for office. My superpower will make me a celebrity, and I can unmistakably commit to trying to do good things and avoiding special interests. Zelenksyy became president of Ukraine on the basis of being famous from a TV show where he was a noncorrupt person, operating as proof of concept that people care about mild fame and noncorruptness enough for it to take you to the top.
Of course, then I would have to actually govern. I don't know if this would be a disaster (can't use small lies to grease the wheels) or actually go okay (I can tell people I'm not plotting against them and just doing things for the good of all, and they will believe me).
This strategy wouldn't work for me because as soon as an interviewer asked "do you believe christianity is dumb" and I refused to activate the power, the game would be up. I suspect interviewers would be able to prove that you have specific beliefs which make you unelectable.
Honestly, dishonesty and creativity with the truth is a major part of statecraft. There's the 'Mr. Smith Goes to Washington' myth in this country of the honest, incorruptible person who goes in and frees the country from the special interest, but it only occurs in movies--probably the closest analogy was Jimmy Carter, and look how he wound up.
Zelenskyy did a Trump-like outsider campaign where everyone was so sick of the corrupt people in charge that he was able to win, but he had a long career in showbiz before that and his ability to act has been quite useful in getting countries to support him in what's a really horrible situation.
The most successful US presidents--FDR, Eisenhower, Reagan, Clinton--all knew how to play dirty and be loose with the truth when necessary. It's about coalition-building and being inspiring, not doing good things and avoiding special interests. The Dems need woke party activists to get out the vote just like the GOP needs all those MAGA people who think Biden drinks children's adrenal glands. You need votes in Congress and you need the SCOTUS or state and local governments not to overturn everything you do. I would suggest you pick one issue to focus on that you care about (AI regulation?) and sell out on the rest of them.
That was the only possible alternative to just hiding the "superpower" to all. But I do not think it would work: you will have to detail your actual plan for office and all your true interests/character traits. I do not think you will be able to compete with people able to lie, even with the benefit of actually being able to prove your are not lying. Except maybe if you happen to be superhumanly moral, altruist and have socially accepted interests/kinks.
Else you will be food for the cat the first public debate you are in.
The loophole in the case as stated is that "While using this power, you are incapable ...even of saying things you are not confident are true"
So one of your advisors tells you a pack of lies, but you believe it to be true ("Saddam Hussein has weapons of mass destruction pointed at London and ready to go within five minutes!") and you make your ears glow green, go on national television to tell the public this is why you are going to invade Iraq, and everyone believes it has to be done because President Green Ears said it was so, and like George Washington, he cannot tell a lie.
If you want to manipulate people using this power, don't be Green Ears Man. Be the person telling President Green Ears, or telling the people who tell President Green Ears, "cross my heart and hope to die, Trump hired hookers to pee on a bed in Moscow".
A powerful analogy I have recently discovered makes me think little good would come of your superpower as POTUS. Also remember your own maxim: True, Kind, Necessary. If you can't lie, there are bound to be statements which you either can't make or would violate the "2 out of 3" condition.
Thaddeus Stevens: The people elected me to represent them, to lead them, and I lead. You ought to try it.
Abraham Lincoln: I admire your zeal, Mr. Stevens, and I have tried to profit from the example of it. But if I'd listened to you, I'd have declared every slave free the minute the first shell struck Fort Sumter. Then the border states would've gone over to the Confederacy, the war would've been lost and the Union along with it, and instead of abolishing slavery, as we hope to do in two weeks, we'd be watching helpless as infants as it spread from the American South into South America.
Thaddeus Stevens: Oh, how you have longed to say that to me. You claim you trust them, but you know what the people are. You know that the inner compass that should direct the soul toward justice has ossified in white men and women, North and South, unto utter uselessness through tolerating the evil of slavery. White people cannot bear the thought of sharing this country's infinite abundance with Negroes.
Abraham Lincoln: A compass, I learned when I was surveying, it'll... it'll point you true north from where you're standing, but it's got no advice about the swamps, deserts and chasms that you'll encounter along the way. If in pursuit of your destination, you plunge ahead heedless of obstacles, and achieve nothing more than to sink in a swamp... what's the use of knowing true north?
The reconstruction results were probably not that much worse than could reasonably be expected. It's not like nation building efforts in Iraq or Afghanistan were resounding successes.
But once you reveal this power it will be used against you:
Imagine you enter an interview with a savvy and agressive journalist. He may demand that you turn your ears green before answering certain questions.
If you refuse you automatically seem extremely untrustworthy.
If you agree, you may say things that are controversial and you may damage your popularity and dash your electoral hopes. There is a good reason politicians love and tend to lie and avoid questions.
My gambit to deal with that is to regularly refuse to answer questions, saying (with green ears on if need be) "I don't want to answer that question, not necessarily because the answer is embarrassing but because at some point someone is going to ask me something to which the answer is embarrassing and I don't want it to be obvious when that happens".
Not perfect, but I think a lot of people would understand.
I wonder if you could get round that by making your ears glow green then saying you won't answer that question because it's private and nobody's business but your own/you genuinely believe that is not in the public interest, it is just prurience/you should not answer that question because it would breach confidentiality.
Of course, a muck-raking journalist is going to find ten ways to hang you whatever you say.
But by saying that you won't answer the question because it's private you are lying.
You are not answering the question because the consequences of answering it will be negative for you, and that's the only thing you can say to avoid a question like that in green ears mode
I don't see how it's lying to say "I don't want to answer that question, it's none of your business" as to what age did you lose your virginity/what are your kinks/did you ever think about committing arson/other nosey-parker prurience.
Sometimes you do have to be willing to accept investigation (e.g. being vetted for jobs which involve background checks) but a random journalist fishing for some juicy gossip can go jump in the lake.
Lots of people sincerely believe that what serves their personal and tribal interests is good for everybody so the glowing ears don't prove that much. I'm more interested if a politician chooses a smart policy than if he firmly believes in what he says. Now you are unusually self aware so it would be harder for you to lie to yourself, but that's still no guarantee that you would be an effective political leader. Though, I would vote for you to become a DA or the Comptroller General of the United States or in another similar position where probity is key and less open to interpretation.
I don't use it, and don't tell anyone I can do that. I don't see how this could bring me more than a little bit money and minor celebrity. But it will basically makes me unable to lie for any serious purpose, once people know about this "superpower". They will just assume I lie if I refuse to glow.
Frankly, apart if I am lost and on the verge of dying, and ear glowing is my only way to let the rescue team find me, nobody will ever know about this capability.
Seems like the downside is that once people realize you have this power, it becomes more difficult to get away with a lie. “Prove it! Turn your ears green and repeat what you just said!”
A meta-honesty policy can help with that concern. Just like encrypting all your communication makes it hard to know when you're actually talking about secret stuff. You just categorically refuse to make your ears glow x% of the time or whatever.
I don't know what the optimal way to implement something like this is, but I'm certain it's not a total loss of your ability to lie.
That won't work. If you're not glowing your ears, you're obviously lying.
My brother likes to wear headphones at all times, and if you want to lie about something unimportant you could take that approach. If you want to lie about something important, you won't be able to.
Refusing to glow randomly wouldn't work, but setting out to exact a cost, social or otherwise (assuming one endorse a role of referee/trusted party, by keeping a tab on who has to pay a premium/is banned from your services), on any request ot prove yourself by glowing (or any request which offend you by implying you're un-trustworthy), could discourage people from being distrustful of you when you don't glow.
Anyway, one bit caught my attention in particular, when discussing the difference between being "awakened" and being "liberated":
---
MICHAEL: Well, we can still have material in the deep unconscious that is distorting our behavior, or in an unconscious way, causing us to do stuff for us to have reactions. And of course, if we're seeing that clearly enough, with enough emptiness, it's going to have a minimal effect. But over time, when you have deep, stable, nondual awakening, the name of the game is (if I were to use Western terminology) 'to start clearing out all the knots and difficulties in the deep unconscious'. If I was to use the traditional terminology, it's 'you're cleaning out all the karma', all the samskaras. It's when you not only have deep, stable, ongoing awakening, but also all that baggage is cleared out completely, then you're liberated because that's what liberated means, you're seeing everything clearly as it is. That's the awakened part. And then the liberated part is you're not being controlled by your unconscious or to put it in traditional terms, your karma.
SPENCER: And so what are you being controlled by? Or is the idea that you're kind of choosing for yourself or something else? So, you're saying you're trying to avoid being controlled by your deep subconscious. So then, without the control of your deep subconscious, how are your choices being made?
MICHAEL: At that point, it's part of the spontaneous flow of experience. It's not just a reactive trigger from the past.
---
I don't know if I am convinced that that avoiding reactive triggers and let "the spontaneous flow of experience" guide your actions is a fully desirable outcome. I also imagine it's kind of hard to talk rationally about this, it seems to me that he's describing an experience more than a concept. However, I'd like to hear opinions by more experienced meditators on this point.
Tentative-- pay attention to your feet and the floor under your feet, and also your belly. This might help you remember that you're a person rather than getting caught up in what the other person wants.
Don't laugh at people's attempted jokes unless you actually think they are funny
If this is in a work context, remember that you are being paid to play a role, and that telling people things they don't want to hear (for example) may be part of that role.
I suggest "Never Split the Difference". Thinking and answering for myself "what does success look like" for various conversations or scenarios has made a big difference to me.
A handy habit: say “let me get back to you” or “let me think about that” instead of agreeing right away.
First it signals that you’re not always going to give them what they want. Second it gives you time to articulate a reason for refusing, if you want to refuse. Some over-agreeable people feel like they have to give the asker what they want if they can’t immediately express a justification not to.
Also: avoid pre-emptive justification. Even when you have a perfectly good reason to refuse someone (say you have a funeral on that day), just say “sorry, can’t make it”. You can explain why not if they ask. But get yourself out of the habit of feeling you need to explain why as a part of the refusal. Most of the time people totally understand and respect that whatever reason you have is good enough.
Edit: How to be more Machiavellian - articulate to yourself a goal you want to achieve in any given interaction or relationship. Then go do it.
Train yourself to act with intent towards an outcome rather than just acting how you feel. You don’t respect a person but they can give you something you want? Chuck some insincere flattery their way. Want a subordinate to treat you with more respect? Be more distant and less warm towards them (even if you happen to like them personally). That sort of thing.
You replied to my initial comment: "Care to explain what you mean further?"
I replied...."I was partly just teasing! I am a 72 year old who pines for the days when none of this stuff got talked about. Wittgenstein's famous; ".... those things of which we cannot speak, we must be silent" is one of my favourite quotes....(although I may be misapplying it here.) I am not blind to the upsides of the times....the internet is a marvel and internet discourse is too - for those with the right personality blend of curiosity and skepticism (which unfortunately is only a small minority of folk)....I tend to think of Western liberalism as a great invention but - Icarus-like - one that does not know when enough is enough. I am a pessimist about the future of Liberty."
I saw your explanation, appreciate it, and given that it sounded like it was mostly a joke, and otherwise an invitation to a giant unresolveable debate about fundamental principles, chose not to continue that discussion thread further.
Thanks....it wasn't mostly a joke.... mostly serious. But musings on fundamental principles (whether resolvable or not)... Yes that's my 'thing' you could say.
How should I handle the voting for the book review contest?
1. First past the post - easy for you, easy for me, everyone understands it. But it doesn't offer much of a chance to express true preferences.
2. Approval voting - easy for you, easy for me, provides more choice. But I'm afraid that this advantages milquetoast reviews that everyone could appreciate a little and nobody had anything against.
3. Ranked choice voting - possibly easy for you, pretty hard for me: I don't really know how to implement RCV on Excel or SPSS without doing it manually for each of the sixteen entries for however many rounds there are. If you choose this, give me some hints for how this could work.
4. Something else (I will not choose this and will resent you for proposing it).
Ranked choice voting. I think your intuition is right about approval voting not being a good match for what we care about. I'll text you at your SC number with a python script that should make it super easy to do from Google forms if you want to go that route.
This exists: https://www.rcv123.org/. I know you know how to google, so I'm thinking it may not be what you want. I'm an Australian, so RVC makes the most sense to me, but it's a book review contest, not an election, but then again, seeing the preferences would be interesting and that's what we're here for.
So in approval voting, if I vote more than once, it's like I get another vote? That seems wrong. If you are going to pick more than one, then each should only get a fraction of a vote... well in my perfect world.
There is a dilution if you vote more than once. Consider the extreme case of voting for everyone--then your vote does not affect the outcome. Or the case of voting for all but one--like downvoting the one.
If they left no impression on you, then you don't want them to win. It's really that simple.
(But that's precisely why it should be called "range voting" rather than "rating". People need to be disabused from the notion that they're making objective ratings rather than a subjective valuation relative to a small set of other choices on the ballot. Once that can be accomplished, it's really the best method for a case like this.)
Clearly there's a lot of disagreement about what kind of voting system to use. What voting system should we use to determine what voting system to use for the book review contest?
I have some ideas for 4, but I'll forebear to mention them.
For 1-3, my preference is 2.
But if you do decide to go for 3, it's pretty easy to write a python or JS script that will read a CSV file exported from Excel or Google Sheets and tabulate an Instant Runoff election based on the voting data. Here is one such script I found with a quick googling: https://gist.github.com/Sam-Belliveau/943142dab814b674ecf6f6a852aba0cf
Or if you'd prefer, it'd probably take no more than an hour or two for me to write one from scratch.
All voting systems suck, but some do suck more than others.
How about we collect the ranked preferences of all the readers, and then calculate all the game theory equilibria under the assumption that every voter will vote strategically, and the winners are whoever wins at least one of these equilibria. (Of course, there might be no equilibria, and thus no winners.)
On, RCV/#3...isn't this place full of programmers? Isn't this just a python script?
Like, presuming every row in an Excel sheet is one voter and every column is a book review and score...I'm not sure I can think of a good way to do it but I can think of 2-3 "dumb but good enough" ways.
But yeah, if RCV is the choice, just ask. Someone should be able to knock that out pretty fast.
Scott would need to have Python, know how to run the script, format the input in the correct format (CSV or space delimited? Header or no header?), and then trust that it has no bugs. Not really trivial if Scott has no programming experience.
Or he could give the CSV file to a programmer volunteer who will tweak the script as needed, run the script for him, and send him the results. I'm willing to volunteer for this if he decides to go that route.
And the winner of the book review contest is...Erica Rall! Who knew she was such a great writer?
In all seriousness, your suggestion could work if multiple programmers are given the CSV and asked to write their own script. Then Scott could check if they all come up with the same list of winners. Although at that point, it might be easier for Scott to do instant runoff by hand.
That's not how the internet works. You don't ask a bunch of people for code, you ask one person for code, you post that, and then everyone and their grandma posts their "better" code because "obviously" that original code is grossly wrong :)
2. Approval voting. If nothing else, using approval voting for some mundane things like book review contests (or choosing which movie to see with your friends, or whatever) make approval voting less of an abstract thing you have heard of, and more of a valid and intuitive approach to voting. So I think it is a good think in terms of popularising and legitimising approval voting, which contributes a little to it being potentially used in elections where it really matters.
4. I propose that each voter is given the task of casting 4 votes, that he/she/they splits among 1 to 4 reviews. This is a kind of approval voting , but it does not benefit milquetoast reviews because highly favored entries can gather all votes from a voter.
4. Something else: Range voting (e.g. 1-10), which is like approval voting but lets you give a score to things, hopefully avoiding the milquetoast problem. Also hopefully as easy to implement as approval voting.
Downside: people could give scores of 1 or 10 exclusively, but then it's exactly like approval voting.
Outside that, my preference is approval > ranked choice > FPTP. I think FPTP is too one-shot. I would rather a milquetoast "everybody kind of liked this review" win over a more polarising one, and I guess this is a good example of how choosing the scoring system can affect what the winner looks like.
I think first past the post is fine for something like this, where you are picking from a number of acceptable choices and judging on quality rather than values. Approval doesn't really work when the quality of candidates is a smooth gradient and there is no clear line between acceptable and unacceptable. I doubt people will want to put in the effort to cast an informed ranked choice ballot with this number of entries. If you do this, I would restrict it to something like top 3.
Yeah, I think top three is the best compromise. First past the post works if I have an outstanding entry that I think is the best of all sixteen, but more likely I'll have "well there were three I thought were really good, I can't pick one out of them" and then not vote.
In the US? Publisher or distributor (Amazon)? Definitely not. The Author? Probably not. Here is an old article asking about this exact scenario, minus AI:
>Ultimately, in this important decision, the Ninth Circuit found that the publisher had no liability to its readers for the nearly-deadly information. Essentially, the court concluded that imposing liability on the publisher would open the door to the prospect of unlimited liability that could severely threaten the free flow of information.
The article's conclusion is "probably not". I think it would be especially hard to convict an author in a criminal trial. In a civil trial, I could see a jury siding with the plaintiff but it's a gray area.
"Well, sir, the book you wrote killed a mushroom-hunter. What do you have to say about that?"
Which of the following is GPT4's likely response?
(a)Jeez, I am just a sucky little large language model.
(b)I was just doing my job, which is predicting tokens. I am very good at it, but tokens are not mushrooms, and the responsibility lies with whoever did not take that into account.
(c) "Once the rockets go up who cares where they come down?
Yeah, can't wait til we start relying on it for more serious tasks, like stamping out some widespread infectious disease. "Oh, you wanted me to stamp it out *without* killing the sick, the infected, the possibly infected and the highly vulnerable ? Your prompt did not specify that."
There was a post/plug a while back about a high school analytics summer program, maybe a predictive analytics program or fellowship. I think it was at an institute associated with Berkley. I can’t find it. Does the hive mind remember? Thanks in advance!
Many of you participated in a survey I ran on the HEXACO model of personality. The results are online here:
https://thingstoread.substack.com/p/the-big-five-is-incomplete
I was surprised in my thread below "Can an AI grow bored?" at the objections by those who consider AI x-risk to be high to future AI experiencing anything like emotions. I would have expected the opposite. I want to ask a slightly different question: Do you believe AI-x risk to be high AND do you/don't you believe advanced AIs will have qualia?
People don't have a good sense of consciousness, even to the extent that anyone does (which isn't much of an extent). They're strongly anti-dualist, but then also often think that only neurons can produce consciousness somehow. But even though I acknoledge this kind of thing, I still struggle to get myself to believe that AI will be conscious.
At the very least, I think it's reasonable to believe that if AIs do become conscious, they will do so in a way entirely alien to our own kind of consciousness. Unless they emerge through human brain emulation, their cognitive processes would be so different to ours that I don't think there's any reason to believe they will have things comparable to human emotions.
Things like emotions seem to be a product of how our brains are designed and the nature of our non-conscious psychology. We experience the mental state of a particular emotion seemingly because we evolved to have this state to increase fitness e.g. we experience the mental state of anger to motivate actions that are aggressive, threatening etc. to reduce threats to safety, resources, social esteem. It's almost certainly the case that we *feel* anger because we have the mental state of being angry (otherwise it would be some grand coincidence that our mental states and conscious experiences are so closely aligned).
If a cognitive system lacks the same architecture as the human brain, I don't see any reason why the conscious experiences it would have would closely match our own at all. And unless it was specifically designed to mimic the human brain, I don't see why it would share that architecture. If that's the only way to get an advanced AI, then we can be fairly certain AI is not going to happen any time remotely soon.
It also raises the issue of the function of consciousness and epiphenomenalism. It seems like all an AI would need to work are it's physical hardware and associated physical phenomena - voltages on a silicone chip, basically.
We don't really have trouble explaining why computers work - the movement of electrons and so on explains it all. So if an AI was having conscious experiences of boredom, it's hard to conceptualize this in physical terms. It seems like a particular movement of electrons on a chip would be needed to give rise to conscious boredom (if that's actually possible) - but in order for the experience itself to actually DO anything, the experience needs to afect the behavior of electrons (or whatever physical thing is responsible for computation.
On the one hand, if the experience isn't doing anything itself, then it's irrelevant and all we're really interested in is a certain pattern of computation that puts the AI in the 'mental state' of 'boredom' (i.e. such a condition that results in the AI behaving in a way that we would call it 'bored' of a certain task, such as avoiding certain actions despite the weight that a model would assign to those things), so we haven't really acheived anything by bringing in the issue of conscious experience.
On the other hand, if we are going to say that these experiences are themselves going to impact the physical things happeneing on the hardware, then this is very problematic, because its unclear how this could possibly work. It implies that we could (at least in principle) observe a system and its physical behavior would diverge from that we would predict based on physical laws (unless you're some sort of panpsychic who believes consciousness in some basic form is already responsible for physical laws). This of course isn't actually unique to AI systems and cannot really be conceptually explained for humans (that is, how a materialist account of consciousness avoids epipohenomenalism at the level of explaining how 'qualia' themselves can possibly affect behavior - pointing out the correlation between qualia and speech acts doesn't even suggest the most basic, speculative mechanisms that could possibly be responsible).
All in all, I think we're not even close to being able to even conceptualize this stuff to begin with to think we can understand how it could affect AI systems of the future.
I tend to agree - if you think the risk of AI eliminating humanity is high, then naturally you would probably think barriers between AI and HI are low, so qualia would be less likely beyond AI.
Maybe it's just that everyone has such a strong sense of AIs as empty robots that even if some people are more likely to attribute the ability to experience qualia to AI, they still don't think AI will have qualia.
Has anyone tried to use an LLM to replicate a specific individual, someone for whom there was massive amounts of recorded speech or writing? It would be very little relative to the bodies of text that LLM's are trained on but more nearly consistent, which might help a little.
Not really what you're talking about but relevant: https://www.cbc.ca/documentaries/the-nature-of-things/after-her-best-friend-died-this-programmer-created-an-ai-chatbot-from-his-texts-to-talk-to-him-again-1.6252286
I think this answers your question: https://infiniteconversation.com/
Sex and drugs and rebelling nuns - https://slate.com/human-interest/2023/09/nuns-texas-ft-worth-bishop-sex-drugs-power-plays.html
Hoping for some expert commentary from Deiseach.
Do we have any good theories as to why a number of ancient societies considered human sacrifice abhorrent, despite tolerating many other forms of violence that seem just as barbaric by modern standards? It's unsurprising for the Jews in Biblical times; in a society where monotheism is considered a founding principle, it makes sense that sacrifices made to pagan gods would be seen as worse than regular killings. But the distinction also seems to have developed in pagan Ancient Greece and Rome, presumably independently of its development among the Jews, so there must be more to it than that.
What's to explain? Conservatives kill convicts but not unwanted fetuses; liberals kill unwanted fetuses but not convicts. Westerners obsess about animal cruelty and eat semi intelligent pigs, but not bugs; Muslims obsess about purity, and eat locusts. Culture is like that - you don't design it from first principles, you just do what everyone around you is doing, and then make a living as a blogger or comedian saying, "Isn't that weird? That's so weird!"
Yes, and then some people grow disappointed in their native culture and look for answers elsewhere, often falling into the “culture I know nothing about must have the answers” trap.
And yet nobody assumes that the trap they know nothing about must have culture 🤔
Oh I think that assumption is made all the time :)
😂
The only types of violence that I can think of from ancient times that were "tolerated" would be ones we might classify as acts of the state, such as war or capital punishment. Of the societies I can think of where human sacrifice was permitted or encouraged, it appears to have been state-sanctioned as well, not individuals taking it upon themselves. Murder committed individually seems to be universally against the law even in ancient civilizations. Assaults may have been more common or not result in criminal liability, but I think that was a matter of lack of enforcement power instead of toleration for it. The Bible has a number of laws against harming people in the OT, for instance, as do other ancient law codes.
Perhaps you could elaborate on which practices were commonly accepted that today we would consider barbaric? Slavery might fit, but we might get ourselves down a rabbit hole of whether or not slavery exists today (it does) and is acceptable to "modern standards" - which gets murky depending on your definition of slavery and who's doing it.
What other forms of violence are you thinking of? The distinction that comes to mind is that, in turning the death into an offering to the gods, the act of killing becomes the reward. As opposed to more normal crimes where you kill someone because you want their stuff or just hate them, where the killing is in furtherance of a different goal, ritual sacrifice has no natural endpoint, because you're doing it for the sake of doing it.
Anyone else found specifically astralcodexten blocked by their work firewall this week?
My gatekeepers saying unsupported protocol
Yes, probably since the change to the "astralcodexten.com" domain, my company shows an "Are you sure you want to visit this site?" page, "Request method cautioned for category Miscellaneous or Unknown". But these warnings are so common, I usually ignore them. Which defeats the purpose of the whole security circus.
Yeah sadly I don't have the autonomy to click through. The balance of hassle of reading on my phone vs emailing the help desk is currently tilted towards don't engage.
Appreciate the response anyway.
It says KontextMaschine posted 1 day ago on his tumblr?
You can queue up posts to be posted automatically at some point in the future. He seems to have done this for Labor Day.
Is anyone here able to confirm the results of Ecuador's parliamentary election on August 20 and/or Zimbabwe's senate and national assembly elections on August 23? I have created markets about both but haven't been able to properly resolve either. (Sure, the results have been reported on English Wikipedia, but being a Wikipedian myself, I am inclined to distrust the wiki and trust the source it cites instead.)
Sorry, off topic but I have to ask. Why no Labor Day good wishes?
I forgot :(
My employer is asking everybody to sign a new contract that includes a non-compete clause and a 2 year ban on hiring employees after we leave. While I don't expect either of these clauses to affect me in the future, I am insulted at the company's attitude towards its employees. I could perhaps redline the contract, but that does nothing for anybody else at the company or future employees.
Basically, I and a couple of others have decided we will not sign.
I'm curious if anybody else has had this experience and what happened as a result. I'm not sure whether I should expect to get fired (the company is not doing well and they may see this as an opportunity to get rid of people), but I'm accepting that possibility.
Depending on state law in your location, they may not be able to request that you sign this without additional compensation. I know Texas forbids that, for instance. Some states would allow this upon initial hire or in exchange for a promotion/raise, but not just because.
Whether such an agreement would hold up in court is also very much in question. Courts often frown on mid-level or lower employees, or even anyone not specifically holding company secrets as a requirement of their job, being held by non-competes. Courts aren't big fans of denying someone future employment, and even in cases where it should apply for good reason, courts are reluctant to enforce it. That would especially be true if the company laid you off or fired you, but even for employees who leave under normal circumstances. This is not universal, but it seems that courts only really want to enforce these when someone leaves a company for the purpose of taking the company secrets with them.
This was one of my greatest objections also—that there was no consideration being offered.
It's all moot now, since they backed down once several employees refused to sign. I'm pretty proud of my coworkers right now.
A company I worked for asked me to sign such a thing, including something saying that it was already covered in my compensation. I didn't sign. Later, they sent me something saying they noticed I didn't sign. I ignored it. Nothing happened.
I'm not a lawyer, but I advise you not to sign. Upside to signing? None.
If you don't like it, don't sign it. (Say that you need to ask your lawyer. Or say that your lawyer advised you against signing.) Or maybe ask what's in it for you -- by signing that paper, the company gets some *extra* value compared to the existing situation (otherwise, why would they ask you to sign), so what *extra* value are you getting if you sign? (Perhaps a huge bonus to compensate you for potential *two years* of unemployment?) If the answer is "nothing", it means they want to get something for nothing, so just refuse.
Depending on where you work, yes, you might get fired tomorrow. Note that this is true regardless of whether you sign or not! Maybe the company is already planning to fire you and many of your colleagues, and they just want to make sure that you will not run en masse to their competitor. Ask yourself how much would signing the contract and actually following it hurt your chances at job hunting during the next 2 years. Do you want that?
Most likely, the contract is illegal. (Verify this with a local lawyer.) It is still a bad idea to sign it, for multiple reasons. First, the company may decide to give you legal problems anyway... and even if you ultimately win, it will cost you a lot of time and money. (As they say, the process is the punishment.) Yes, it would cost your company some money too, but for them it's just regular expenses, and perhaps they feel that the deterring effect makes it worth it. Second, the new job will probably ask you for references; and if they contact your old company, the company may tell them that you signed a non-compete. The potential new employer might decide that the potential legal hassle is not worth it, and they may decide to simply hire someone else instead. Even if the new employer decides to accept the legal risk, you kinda start your new job known as "someone who signs contracts he doesn't mean to actually follow". People may or may not care about this.
I think this is exactly one of those situations the unions were invented for. Of course, many people are too proud to join a union. This may be the moment to reflect on your preferences. If you had a union, you could simply ask your union lawyer, but quite likely the company wouldn't even try this, because the reaction would be predictable. You wouldn't have to ask strangers on internet for help.
I don't have a similar experience, because in my country the employers usually don't even try this. My contracts only contain a clause that I cannot work for a competitor *during* my current employment. Even that only applies to competitors; I am in general allowed to have an unrelated side job (in case that I am not sure what qualifies as unrelated, I can simply ask HR). Once the employment is over, it is over, full stop, the former employer has no power over me (except for the NDAs I have signed).
Well, after several of us refused to sign, they decided to scrap the whole plan. So a win for solidarity!
All of this, especially the business about consulting a lawyer, which I'll address a bit more.
A lot of people are instinctively resistant to the idea of using a lawyer; it often feels like a thing that's just for really rich people (or real assholes), or only for truly DIRE situations like being permanently maimed in a car accident. Lawyers are a thing regular people just "don't do."
But that's not true!
While legal expenses can get really expensive really fast, a consultation about your circumstances and maybe a threatening letter on legal letterhead is often *very* affordable - often just a couple hundred dollars. I was certainly a "I don't hire lawyers" person until circumstances with my HOA's truly egregious failures forced my hand - and the moment my lawyer got involved things turned around almost miraculously. I wish I'd gotten him involved right from the start, I would have spared myself tremendous hassle and anxiety.
Think of consulting a lawyer as a kind of insurance against having a tremendous pain in the ass - $200-$500 or whatever can spare you *YEARS* of uncertainty, anxiety, lost income, etc.
I don't mean to be flippant about a few hundred dollars - that's a lot of money - but one of my favorite sayings is, "the cheapest way to pay for something is with money," and in your case, it sounds like the cheapest way to know what to do is to pay a lawyer to tell you.
@Snags, you really do need a lawyer for this. Find one on Avvo or whatever and see what they think!
https://www.shrm.org/resourcesandtools/legal-and-compliance/employment-law/pages/states-restrict-noncompetes.aspx#:~:text=The%20FTC%20said%20noncompetes%20constitute,entrepreneurs%20to%20start%20new%20businesses.
Small-town probably doesn't apply much, but I worked at a company started during the owner's Non-Compete duration with their previous employer, and nothing came of it. Although their job had also been licensed by the state.
I have seen Klotho suggested as a possible way of slowing aging. Does anyone here know anything about it?
More information today in the NYT about new studies confirming that college, particularly if you have to pay a steep price to get there, and particularly if you don't manage to finish, isn't the slam dunk investment it once was.
https://www.nytimes.com/2023/09/05/magazine/college-worth-price.html
> Start here: If your tuition is free and you can be absolutely certain that you’re going to graduate within six years, then you enter college with a 96 percent chance that your gamble is going to pay off, meaning that your lifetime earnings will be greater than those of a typical high school graduate.
> The problem, though, is that many students who start college don’t graduate — about 40 percent of them, by one estimate. When Webber factors in that risk, your chances of coming out ahead of the typical high school grad start to shrink. If tuition is still free, you now have about a 3 in 4 chance of winning the bet.
> The second problem is that going to college isn’t free. If you’re paying $25,000 a year in tuition and expenses, Webber calculated, your chance of coming out ahead drops to about 2 in 3. At $50,000 a year in college costs, your odds are no better than a coin flip: Maybe you’ll wind up with more than the typical high school grad, but you’re just as likely to wind up with less.
I'm confused. I just checked Harvard and MIT tuition, and it's $56000. So, if you're paying $50,000 a year, you're going to an elite college, no? Is the claim really "50% of the people who were admitted into, and at least started attending, an elite college don't end up earning more than a typical high school grad without a college degree"?
"College costs" includes room & board, which is almost certainly a five-figure number and the first digit may not be a one. Also fees, textbooks, travel expenses, etc.
Even non-elite colleges charge in the neighbourhood of $50K in tuition. Of course, that's the sticker price, and discounts are pretty common.
https://www.usnews.com/best-colleges/rankings/national-universities
Tulane is #44 on the US News & World Report list, and charges $62,844.
Baylor, at #77, charges $51,738.
Drexel, at #105, charges $58,965.
Discounts are more than "pretty common", they are routine for all but the wealthiest households. My wife does college counseling fulltime for middle-class and upper-middle-class U.S. (mostly) families, and she and her colleagues don't even bother with sticker prices anymore when working with their clients. They just roll their eyes at the USNews list. They use average net prices actually paid at each of the various colleges and help the families use that for comparison and/or negotiation purposes.
Also, for several years now the sticker prices have been declining relative to inflation and the average actual amounts paid have been declining absolutely.
This Brookings report from earlier this year lays it out pretty well:
https://www.brookings.edu/articles/college-prices-arent-skyrocketing-but-theyre-still-too-high-for-some/
At the private schools that have large endowments this difference is even greater. That Brookings report shows that a student from a family with $125,000 in household income can now expect to pay $32,800/yr to attend a private college with a large endowment, which is well under half of the quoted sticker prices at those schools. A student from a household income of $75,000 now pays $17,700 on average at those schools.
Drexel specifically now has a 50% across the board discount for community college graduates, so if you have a two year degree you can cut that sticker price in half right off the bat.
That sort of thing is spreading now particularly among U.S. colleges outside of the "highly-selective" tier. It's about supply and demand: the college-age population stopped growing some years ago, is now declining, and higher-ed administrators think the decline will accelerate a year or two from now. They talk about that last part a lot and refer to it as "the enrollment cliff".
Colleges with Ivy League-type reputations have compensated by pursuing more international students, with the additional benefit that those students pay full list price if they do enroll. However that supply has also stopped growing for various reasons, and in any case isn't a strong competitive prospect for schools like Drexel. So in addition to the general flattening in the cost of college in the US, incentives like you just described are spreading.
The expectation in the industry is that the bottom end will get hit the hardest. The prestigious private universities and the state-funded big schools will be fine. The places in danger are small primarily-undergraduate schools with only local reputations.
This is at least consistent with a Caplanian understanding of education, as say, 60 years ago, college students were more intellectually distinctive than today as a higher and higher proportion of the population started going to college.
Noah Smith posted an interesting take on that shift:
https://www.noahpinion.blog/p/is-it-time-for-the-revenge-of-the
His hypothesis is that because GPT does a mediocre job at many things, it will be a great boost for people with sub-mediocre skills, but will almost nothing for the best ones, so it will make people more equal.
This is an interesting perspective, and GPT will definitely shift the learning curve in weird ways, but I also see some problems with the optimism about the "normies".
First, even if GPT will hold your hand during your job, you will still need some minimal skills, such as "asking the right questions" and "noticing that you need to ask". Without these skills, GPT won't help you much. For example, as a junior software developer, you can ask GPT to write you the code, and then you can also ask GPT to rewrite it to make it thread-safe. But it's you who needs to notice that the first generated code isn't thread-safe and that it should be. Similarly, as a lawyer or a doctor, you can give GPT some information and let it figure out what to do. but you need to make a judgment about what information is relevant to the case. Then there is also the problem of detecting hallucinations.
Second, a part of senior's job is to manage and advise juniors. But if you are using GPT anyway, and you know that the juniors are going to ask GPT, then perhaps you might skip the intermediary and ask GPT yourself. If GPT can help juniors more than it can help seniors, that also means it can replace the juniors more easily than it can replace seniors. The result could be that on one hand, yes, a normie + GPT could do a mediocre job, but on the other hand, an expert with 5 open tabs of GPT is more effective than the same expert supervising 4 normies, each using GPT. Maybe not. We will see.
As I see it, a great advantage of GPT is that it can make *learning* much faster. Buy (or download) a textbook + ask GPT the additional questions = it's like you have a private tutor for any subject. That said, the schools are already more about credentials than about learning, so unless we somehow decouple credentials from school attendance, it's not going to change much. We will have more educated people, but without the paper to prove it. Also, although GPT will provide great learning opportunities for normies, I don't really expect them to actually use this resource.
>His hypothesis is that because GPT does a mediocre job at many things, it will be a great boost for people with sub-mediocre skills
Or an economic replacement for people with sub-mediocre skills, which they may not find to be a "boost".
What are some common examples of tasks that AI accomplishes with >99.9% accuracy in the wild, ideally without near-infinite task-specific training data? By AI here I mean something like neural networks, i.e. something that's trained by giving it lots of examples and asking it to generalize, as opposed to coding an explicit algorithm.
Background: the company I work for makes optical inspection systems, meaning we collect high-resolution photographs, do some processing, and then rate the components in the images on whether they're defective or not. These components are tiny -- there can easily be >10,000 features to measure in a single image -- and the error rates our customers are looking for are certainly less than "every image contains a part that is incorrectly flagged as exceeding tolerances". The company has been interested in leveraging neural networks more for some of the image processing tasks, and I'm trying to understand whether that's all realistic. I went to look up some benchmarks and saw this page documenting progress on the CIFAR-10 dataset: https://paperswithcode.com/sota/image-classification-on-cifar-10. CIFAR-10 has 60,000 images, and the best model has a 99.6% classification accuracy, meaning it misclassifies 233 of the images; this feels like it would not be good enough. On the one hand, we might benefit from having a more restrictive dataset (the components don't vary that much in appearance and are being imaged in a controlled environment; we're not going train-spotting in the wild). On the other hand, I feel like results I've seen on transfer learning have been of the form "see, with a generic model and a small amount of targeted data / time, we can achieve performance that isn't crap" rather than "...we can achieve performance that's better than anyone has ever done on CIFAR-10".
I've used some ML tools that use OCR to identify text in images, and then to either extract useful data from it or to identify what kind of document it is (called classification). Both can be highly accurate, but neither is in all circumstances.
For example, a 1040 tax return is notoriously difficult, both to classify (lots of similar-looking documents) and to extract data (lots of variants, tax year differences, etc.). On the other hand, we had some classification documents that were always accurate with our model, because they were so unique: for example, even if the document had a confidence level of 10 it was still the top document type selected in the model, and it was ALWAYS right if it got that high.
Document extraction is unlikely to get that high, simply because of image quality. But if you have consistent documents of consistently high image quality then it could theoretically. But then, you asked for examples in the wild.
This is actually about the right level of wildness, thank you! (We make the imaging system, so we control the image quality.) On the other hand, if I think about this as e.g. classifying the components in the image, it sounds like what I can expect is "some but not all types are always identified correctly".
Depending on the document, it may be possible to automatically double-check some of its conclusions. For example, we needed to get beginning and ending balances from bank statements. We could double-check what we found in a couple ways: finding the same figure somewhere else, and/or taking the beginning balance and adding all of the activity found and seeing if the result matched the ending balance. In either case, one could send the failures to a human for extracting the data. The calculation, if the statement had any actual activity, was never found to be wrong.
Of course, nothing is foolproof, and you would have to have a very select group of documents to all be highly accurate to classify.
At that level of accuracy, none I can think of. Having said that, this is probably not the right way to think about ML models in practice. Most organizations I've seen can get some effect from ML models but it needs to be integrated with existing workflows and typically is used to filter out obviously wrong cases so staff can focus on difficult cases.
For example...let's talk about, uh, sepsis screenings. Sepsis is complicated but the guts of it is that sometimes in a hospital, people's blood gets yucky and then maybe they die. This is bad and makes doctors and families sad. The Epic EMR, amongst other sources, sells ML algorithms that will predict whether patients will get sepsis within a relatively short time frame, think <24 hours.
But, in actual workflows, nobody gives a suite of powerful anti-sepsis drugs just because a computer said so. What tends to happen is:
Step 1: Out of 100 people "at risk" patients, the ML algorithm identifies 25 people who will get sepsis.
Step 2: The ML algorithm calls a nurse.
Step 3: The nurse sends 15 people home, they don't get sepsis, gives the anti-sepsis drug to 9 people, and is confused on one patient.
Step 4: Nurse calls a doctor for the confusing patient.
Step 5: Doctor decides whether to give the confusing patient medicine.
Now this is pretty simplified (and idealized) but that's the flow, at each step actionable or confusing cases are kicked up the chain of command. The ML algorithm doesn't need to be 99.9% good, it honestly doesn't even need to be 80% good (Epic isn't) but, ya know, algorithms are cheap to run and if the algorithm gets it wrong 30% of the time, the nurse gets it wrong 20% of the time, and the doctor gets it wrong 10% of the time, naively, you get 0.3*0.2*0.1=0.006% chance of failure vs 0.2*0.1=0.02=2% chance of failure and a lot less work for nurses and doctors.
So, uh, a lot probably depends less on their accuracy and more what your current workflow and tools are. As a general rule, if you have people or expensive tools that spend a lot of time on things that "obviously" issues, it's probably a good candidate for ML. But complete replacement...meh.
I agree with the general picture, but I think that neural nets may be a less good choice than other, more comprehensible forms of classifier, for the first step.
Triage like this is about classifying things as "yes", "no" or "maybe", and it's very important that no "yeses" or "nos" get misidentified as the other, so you want to be quite conservative about saying "maybe"
If you care equally about errors in all directions, neural nets may well outperform simpler things and give a better ROC curve, but their tendency to occasionally throw up completely random and unpredictable output means that they're possibly less suited for being conservative in any given direction, and some simple low-dimensional human-comprehensible function of the inputs may be able to make you more confident that you're only going to err in directions you don't care about so much.
I think right now the company is trying to figure out how to integrate ML into existing algorithmic workflows, and struggling with some version of "if it will improve the performance on 80% of the cases, but fail badly on the remaining 20%, how do we work around that". Obviously if we can tell whether it succeeded, we can just fall back on whatever we're doing now, but that requires being able to tell that it succeeded with a high degree of accuracy.
I read Joshua Cohen’s “The Netanyahus” last week. The novel itself is a fictionalized account of Harold Bloom’s encounter with Benzion Netanyahu’s family in the late 50’s.
The Credits and Extra Credits section after the body of the novel was a special treat. Cohen relates some of his conversations with Bloom and has entertaining anecdotes about the literary figures that Bloom had a chance to work with during his career. Saul Below, Phillip Roth, Susan Sontag, Nabokov, Don DeLillo and Cormac McCarthy are all mentioned.
In that section of the book Cohen also taught me a word I didn’t know. ‘Illeism’, the practice of referring to oneself in the third person ie “No one had done more for Israel than Donald Trump.”
A bit of research turned up a number of notable illeists:
Mr T, The Incredible Hulk, Dr Doom, Charles DeGaulle and Bob Dole during his 1988 run for the presidency. As hilariously charicatured by Dan Aykroyd in an SNL Cold Open.
https://www.nbc.com/saturday-night-live/video/republican-debate-88-cold-open/2868064
“Bob Dole grew up in Kansas. Bob Dole didn’t have the the prep school education, the bumper pool table in the basement, the convertible at graduation, the electric steak knife, the machine that throws the tennis balls at you….
I will pay $100/hour for up to three hours of work to anyone who manages to successfully get this blog connected to astralcodexten.com.
Substack has a process for doing this but it's said it's "connecting" and "will take up to 36 hours" for the past few weeks.
I think there is probably some sense in which a domain transfer may have partly failed, and even though my HostGator account says I own the domain I really don't. The person who does own the domain wants to transfer it to me and I thought we transferred it successfully but it's not working. Even if you know how to link domains to Substacks in general, expect this one to be harder than you think.
Comment here and I'll email you more details. I may take later-answering people I have some reason to trust (eg have worked with them before) over earlier-answering people.
EDIT: Fixed now, thanks Erusian!
Hey Scott, seems like there's still something a bit weird going on with routing. Removing the "www" brings you to a testing splashpage: https://astralcodexten.com/
I have to say I find your ability to outsource your IT problems to your blog both amusing and heartening.
Are all parties willing to give this IT person access to the accounts? If so then this should be reasonably simple to solve since it sounds like it's just a pretty standard series of DNS issues. (It will be more complex than standard Substack work but something a decent SysAdmin type can handle.) If you have to play telephone with the person who currently controls the other account then it depends on how responsive they are.
Hi Scott,
It appears from the WHOIS data you own the domain (NOTE: you may want to make the whois info private) but the nameservers for the domain astralcodexten.com are still controlled through Namecheap:
Name Server: dns1.registrar-servers.com
Name Server: dns2.registrar-servers.com
These nameservers are associated with Namecheap, a domain registration and management company. This suggests that the DNS settings for the domain are still being controlled by or through Namecheap, even if the domain has been transferred to HostGator.
How would I solve this?
The DNS system is slow. Was the transfer sent over recently? Alternatively it might be an issue with the domain being locked in an account. You request a transfer from the receiving account but if the sending account hasn't unlocked it then the transfer won't occur.
The sending account says they did their part to send it. The transfer was sent over late last week.
The other party doesn't send the transfer. They unlock the domain to allow it to transfer. Then they send you a code. Then you use the code to pull it into your existing registrar. If they didn't give you an Auth/EPP code then they didn't do their part. If they did give you one and you put it in then you should have some status as to how the transfer is going which is a decent place to start figuring out what went wrong.
They have sent me the code and I put it into my existing registrar. As far as I know my registrar just said everything went okay.
I’m starting a weekly zoom reading group for anyone who is a practicing psychiatrist or psychiatrist-in-training on a book called Enactive Psychiatry, published in 2020 by Dutch philosopher of psychiatry Sanneke de Haan. It will probably be interesting to you if you are interested in 4E cognition, alternatives to body-mind dualism, causality in psychiatry, or the interactions between experiential, physiological, social and philosophical dimensions of mental disorders (or would like to meet other psychiatrists interested in such things).
You can reach me at jess@psychcrisis.org or DM on Twitter as @utotranslucence; the group starts the week of Sept 11.
In re boredom, alignment, and AI: I feel obligated to mention Martha Wells' Murderbot stories: an assassination robot doesn't *want* to go around killing people, which is tiresome and risky. It wants to be by itself, not deal with humans, and surf the web. As fiction would have it, Murderbot keeps getting entangled in mystery plots.
I actually get bored by the stories, but most people seem to like them.
People often suggest approval voting as a solution for all the world's problems, but we don't know that much about how it works in practice; I think the endless dictatorship of the milquetoast centrist is a real possibility. So I'm in favour of using it for this book review competition so that we can learn more about how it functions in the real world. (For what it's worth I think the milquetoast problem is much less of a problem in this context than it might be in the political world where the stakes are much higher.)
Also, IRV is a lot more effort for the voters. I can't be bothered going back and figuring out whether the review of "Thing: How One Weird Thing Explains Everything About The World" was my twelfth or thirteenth favourite.
One thing to keep an eye out is how approval voting performs in multi-stage elections. Where I am, we have a jungle primary where the top 2 vote-getters go on to the general election. If this were a jungle approval primary, I think it'd be possible for a pair of candidates with similar positions to collude. I'm not an expert on voting systems though; does that seem like a thing that could happen?
IRV really not that hard, especially when ranking after your first choice is optional. We've been using variations on "preferential voting" including "optional preferential voting" (Australian terms) for many years, and rates of informal voting (ie votes that can't be counted) are typically low single digits, which is pretty good considering we have compulsory attendance at a polling booth (or submission of a postal vote) and so even the least motivated turn up to vote. And, much of the informal voting is just people drawing dicks on the ballots, so still not a failure to understand.
I believe that informal voting is higher after a change of voting system while people get used to it, but the effect doesn't last. I also note that Australia has a well-funded semi-independant electoral commission that is excellent at communication and organisation and that this is a luxury some less-developed democracies don't have.
I, for one, welcome our new milquetoast centrist overlords. Especially after the last decade or two of American politics.
George W Bush sold himself as a milquetoast centrist. Most milquetoast centrists support every war, every corporate bailout, every restriction of civil liberties, etc. But maybe they are for changing the definition of marriage from what it has been for millennia, because opposing that would be an extreme position.
We have a reasonable amount of data about how it works in practice- roughly 10% of voters will 'approve' for more than 1 candidate. So AV may have a mild effect on a few races, but it isn't a game-changer.
10% seems to be a pretty good rule of thumb. The UK used the supplementary vote for some offices like Mayor for decades & decades, where you could rank a 2nd person if you chose to- 10ish% of voters did. A hundred years ago the US had a version of IRV called Bucklin that allowed for optional ranking- I think on average 13% of voters did. Having opinions on multiple candidates is just too much cognitive work for a lot of voters- or, they really want 1 guy to win and understand that voting for anyone else harms that chance
I do not think wanting one particular guy to win would rationally discourage supplementary votes in IRV. That is, my decision to add a second ranked vote does not decrease the chances of my rank 1 candidate for all possible ballots by all other voters.
Your voting does approximately nothing for the chances of your preferred candidate winning, under any voting system in almost any realistic scenario. People don't rationally cast votes because of their actual effect on the probability of a candidate winning, they cast votes as a matter of tribal identity. What tribe are you identifying with by approving of a second candidate?
I mean, "nerdy rationalist" tribe, obviously, but there aren't that many of us.
Yeah, that's a good point. With that being said- the UK used the supplementary vote for a long time for a bunch of positions, we have a lot of data on it. 10ish% is a good average for even bothering to rank anyone else
> People often suggest approval voting as a solution for all the world's problems
That's interesting. As a person who likes approval voting, and the Borda count, I feel like everyone always seems to just say we should do runoff voting, as if that's the best or only alternative voting method.
> I think the endless dictatorship of the milquetoast centrist is a real possibility
Is that supposed to be a bad thing? Getting a candidate that everyone likes instead of this toxoplasma competition we have right now being run by Moloch himself? I'd want to move into milqetoast county ASAP if I could.
Same. I'll take the boring and modestly effective over the asshole pageant we have in the Republican party right now.
That boring and modestly effective politician likely would have voted for the Iraq War, Patriot Act, would have been disappointed that Obama didn't take the reasonable approach of toppling every Middle East dictator during the Arab Spring instead of just Gaddafi (and Egypt for about 10 min), etc. The center in American politics is utterly insane.
Democrats have nobody but themselves to blame.
If you only have one choice, you only have to fill in one ranks.
IRV is designed for situations like "I want to vote for the Green Party candidate who can't possibly win, but also make sure the system knows I prefer the Democrat to the Republican after the Green inevitably loses". In that case you still only need to fill in two ranks.
Just read Scott’s links to some of Kontextmaschine’s posts. I hadn’t known a thing about him until I started seeing posts about his death. For what it’s worth, it doesn’t seem likely to me that he died of a brain tumor. He may well have had a brain tumor, but wouldn’t you expect death from it to occur via a slow failure of different systems as the tumor grew, until something failed that’s essential to life? Also doesn’t seem likely to me that he died of kidney failure. Seems quite possible that the huge doses of creatine he was taking gave him kidney damage, but I’m pretty sure that death from kidney failure would have been preceded by a final period of very disabling symptoms, including water retention, failure to urinate, exhaustion, etc. But I might be wrong, and I don’t know why it matters what killed him, though somehow one feels a craving to know.
Kontextmaschine's post about the life he might have lived is just devastating. He’s so wonderfully smart, and full of interest in things, and brave and determined — but somehow nothing has worked out, and he ends up both alone and without a career. (And he doesn’t sound crazy to me in any of the linked posts, except for a perpetual hypomanic quality. But a lot of very smart creative people seem to be permanently hypomanic.) And he’s not even mad about his story. It becomes the topic of one of his characteristic high-spirited explosions of honesty, humor and commentary on the American scene.
Poet John Berryman seemed to have a similar take on himself. Wrote a cycle of poems about “Henry,” who had a lot of John in him. Here’s his poem about Henry’s wake.
At Henry’s bier let some thing fall out well:
enter there none who somewhat has to sell,
the music ancient & gradual,
the voices solemn but the grief subdued,
no hairy jokes but everybody’s mood
subdued, subdued,
until the Dancer comes, in a short short dress
hair black & long & loose, dark dark glasses,
uptilted face,
pallor & strangeness, the music changes
to “Give!” & “Ow!” and how! the music changes,
she kicks a backward limb
on tiptoe, pirouettes, & she is free
to the knocking music, sails, dips, & suddenly
returns to the terrible gay
occasion hopeless & mad, she weaves, it’s hell,
she flings to her head a leg, bobs, all is well,
she dances Henry away.
I read some of the linked writings. I would quarrel with a lot of what I read but I’ve always been a fan of fierce independence from conventional norms.
My go to model for this is the the character “Manny” played by John Voigt in “Runaway Train”. It’s a gem of a film that never found a large audience.
https://en.m.wikipedia.org/wiki/Runaway_Train_(film)
We do not live in a culture where being smart, brave and determined are conducive to things working out. We live in a culture made by and for mediocrities; smart men will make more money on average, but only if they learn to be servile and cowardly.
If you want smart, brave and determined to pay off, you must use the smart to build a time machine, then go back to the 8th-9th century sometime when being brave and determined was enough, with a dose of luck, to get you knighted for plain valor in battle.
Well I'm game to do something about it. Maybe get in touch and bring some of the boys, ey?
Revolution is virtuous.
And it's in the interest of nearly everybody or literally everybody so..... stop on by, get in touch and get involved.
The attitude towards the problem of the modern world reminds me of some lines from Tennyson...
'What is that which I should turn to, lighting upon days like these?
Every door is barr'd with gold, and opens but to golden keys.
Every gate is throng'd with suitors, all the markets overflow.
I have but an angry fancy; what is that which I should do?
I had been content to perish, falling on the foeman's ground,
When the ranks are roll'd in vapour, and the winds are laid with sound.
...'
I just love the jump from 'dude that no one knows dies of unknown cause' to 'woke culture took our freedom to live like 8th century knights' only one comment deeper
Woke culture? Who said anything about woke? This is a deeper rot than the last decade of prog horseshit.
If you think this stuff sprung into existence a decade ago, you're sorely mistaken.
No, I'm aware of its antecedents and so on, I'm just saying that this particular problem has nothing to do with any kind of postwar prog foofaraw.
I mean...yeah. Servile and cowardly. That's basically how I hoarded my mediocre hoard. I could theoretically FIRE at about $60K/yr for my remaining 3-4 decades of life but am not sure what the point would be; a few bad novels lost in the flotsam of Amazon? Trying to fight wokeness and winding up building some right-wing grift to separate boomers and eventually Gen-Xers from their money?
I have my doubts the medieval knights were all that bright. Brave, certainly. Determined? Well, they were born into a warrior class. A lot of the old Protestant virtues like hard work and sobriety (also shared by many Jews and Hindus, of course) are much more of the virtues of a merchant class than a warrior one.
He probably had other mental issues. It's too bad, whatever the reason. Shows the danger of making *all* your friends online, I guess.
"I [...] am not sure what the point would be"
Well, some people have some sort of pride, some kind of self-respect, that makes them chafe at or simply prevents them from servility. People for whom no amount of money can make scraping and bowing, simpering, laughing at unfunny jokes and nodding along with gritted teeth at obvious lies worthwhile.
Those people don't tend to do very well. That's all I'm saying.
"I have my doubts the medieval knights were all that bright."
So do I, that's why I recommended expending the smarts on a time machine.
I guess my view is that if you can't make things work in the time and place you're stuck in, then going out honest and articulate with a few manic fireworks for decoration *is* being brave and determined.
I'm not disputing that he was brave and determined. I'm only asserting that the "somehow" part of "somehow nothing has worked out" is unfortunately legible. It didn't work out because it's not meant to work out. Our culture hates the idea of things working out for guys like him.
The 8th-9th centuries definitely had no shortage of smart, brave, determined people shitting themselves to death in miserable squalor despite their best efforts. And I think even most of the knights lived worse lives than Kontextmaschine overall.
Sure, but most of that was just because it was the 8th century. There were *proportionately fewer* brave and determined people shitting themselves to death than cowards, weaklings and mediocrities, i.e. being brave and determined paid off.
Kontextmaschine sez: " I’m alienated from the people around me and their engagement with the world, and this limits the extent I can draw enjoyment from engaging with the world myself."
I'm sure the knights living conditions and health were worse, but do you think they experienced the alienation he did?
Some of them probably did- people have been feeling alienation for thousands of years. But either way, I said overall. I've felt alienation myself and I'll take it any day of the week over being hungry, or sick, or afraid for my life.
Alienation and illness aren't entirely separate. I once had a godawful migraine while on vacation, and was cared for very tenderly in the ER of a strange town by a fat, sweet gay male nurse. He was so comforting that the whole thing is sort of a good memory. I still think about that guy sometimes.
It seems that Kontextmaschine's degree of alienation was so great that he was unable or unwilling to access any of the sources of sustanance or comfort other people do when they are sick: doctors, friends, people who call or stop by to see how you're doing, somebody who's agreed to take your cat if something happens to you. His death was discovered several days after it happened because online acquaintances called the police. None of them knew his address, and I'm not sure any even knew his name.
A point against prediction markets: this market tanked after I commented pointing out a bit of info that was already easily knowable.
https://manifold.markets/BenjaminIkuta/will-ukraine-hold-a-presidential-el-ccf6c4403856?r=QmVuamFtaW5Ja3V0YQ
It is a known feature of prediction markets that they get distorted when public interest in a topic far outpaces public knowledge and understanding. This is a good example of such a topic for US public.
My music teacher suggested to me that I should learn to always keep the beat of the music in my body. More precisely, he suggested that I should always tap my feet - one tap for each quarter note. Is it a good idea? What's the rationale? Why not every eighth note? Why not tap my feet to the irregular, possibly syncopated, rhythm of whatever I'm playing?
Quarter note foot tappings tends to happen for me without thinking about it. When I'm recording with headphones my wife says she hears my foot going all the time, though I'm often unaware of it. When I notice it, it's almost always quarter notes. I guess they just feel like the natural thing for my body to do when I'm trying to lock in.
In a post on another forum, an extremely skilled musician said the following:
> My father is something of a drum shaman, and the real breakthrough came from practicing with him. He kept telling me to “stop counting and feel it.” He told me “if you want to groove, you have to move.” I didn’t really understand what he meant for a few months, and then it came on like a lightswitch. I felt the pulse in my body, I couldn’t lose it. I didn’t have it until I had it. Since that moment, I have never been able to turn it off.
> Once I had it, the metronome was just a reference and measurement tool. I would just align my internal clock to the external click. As my father would tell me I would “find it, feel it and forget it.”
I've been wondering for a while what exactly was going on here... what exactly got "activated" in this musician's mind? And how do I activate it in myself?
The answer to your second question is to do what the musician you quoted did: practice. Practice builds intuition.
No doubt. The challenge is not whether to practice, but what to practice. He goes on to say:
> The metronome work I had been doing may have contributed, but I really believe that no amount of counting to a click was ever going to flip that switch in me. I had to let go of that entirely.
I'm a retired semi-pro musician who performed or recorded with full-time pros a few times, and what's being described is sometimes referred to as finding the "pocket". One key difference that I learned to observe between full-time professional drummers and those who weren't quite there, was the ability to put themselves "in the pocket" which is what they called it. (I've no idea about the origins of the term.)
Not being a drummer it took me a bit to grasp what they actually meant by that but, yea it is a very real distinction. Applies in lesser degrees to players of other instruments as well but is most directly salient for drummers/percussionists. And yea it's like an internal switch being flipped.
Right, I've heard that term used before. The musician I was quoting actually made a joke about being "in the pocket" earlier in the post so that's no doubt what he's referring to. The fact that it seems to be such a binary switch is interesting, though perhaps it's not too different than plenty of other techniques that suddenly seem to "click" one day after lots of practice.
Pro athletes, particularly in free-flowing sports like hockey and basketball, talk about a similar concept. "If you have to consciously decide to make that move, then during a game you'll be too slow and will just get blocked. You practice it until it becomes instinctive."
>Why not every eighth note?
Tapping every eighth note can be very useful while you are *figuring out* a rhythm, but it can also get in the way of *feeling* the rhythm intuitively. (This lack of intuitive feeling will come through in your playing.)
In addition to what your teacher is suggesting, you may find it helpful to try other forms of rhythmic practice that don't directly involve your instrument — e.g. clapping the rhythm rather than playing it, or tapping a quarter-note pulse with your left hand while you tap the rhythm with your right hand. If you feel like you really want to tap eighth notes, you could try alternating feet, so that your right foot taps out the quarter-note pulse and your left foot taps out the off-beat eighth notes. Using one or more of these might be helpful scaffolding to get you where your teacher wants you to be. (But once you have built the structure, you can and probably should remove the scaffolding.)
Tapping a regular beat is specifically intended to help with playing irregular or syncopated rhythms. Dividing a melody into quarter-note chunks makes tricky passages easier to interpret as a whole. Instead of remembering each of the different note lengths, you can
If a piece has complicated rhythms, it’s easy to get out of time by playing a note that’s too long or too short. When you’re not quite sure how the complicated rhythm should go, it’s hard to know if you got it right. Tapping a regular pulse gives you that feedback - unless you get a whole beat behind, or you make two mistakes that cancel each other out, you can tell if you’ve made a mistake by seeing whether the notes you’re playing match the beat you’re tapping. Modern music notation is designed to make it easy to tell how each quarter-note’s worth of music should sound, and where each note should fall compared to the quarter-note pulse.
For most western music post-about 1600, the quarter-note (or dotted quarter-note, or sometimes half-note or eighth-note as indicated in the time signature) represents the composer’s intended pulse. You can’t emphasise every note, or no note stands out. The quarter-note pulse describes the composer’s intended baseline. Syncopated music sounds the way it does specifically because it’s subverting that baseline. You need to know what you’re reacting against, or it won’t feel syncopated. (Some composers write music with an irregular beat. It sounds quite different from syncopated music, even if the note durations are the same).
In general, tapping your foot is meant to help you keep track of the beat, which is important, at least in the western musical tradition: the beat is the reference by which rhythms are placed, and different beats within a measure are given different default levels of stress/emphasis (with syncopation specifically acting /against/ the normal stress patterns to increase rhythmic energy). Specifically doing /something/ (tapping a foot, counting out loud, etc) with the beat is a good way to internalize where the beat is, better than counting in your head or any external tracking of the beat.
While the beat is often on the quarter note (there's a reason that 4/4 is called common time, and 3/4 is also common), it is by no means universal (6/8, with the beat on the dotted quarter, and 2/2, with the beat on the half note, are also pretty common). It also often makes sense to explicitly tap subdivisions of the beat, especially for faster passages. I'd probably defer to your teacher on this one, though: knowing where the beat is is important, and anything using a diatonic harmonica probably differs importantly compared to the classical stuff I've mostly learned in.
You should /not/ tap directly to the rhythm you're playing: having the reference point of the pulse to keep yourself in sync over time and (if relevant) with the rest of your ensemble is /important/. Without that reference point, you're likely to either get the rhythm wrong or to drift back on to the beat, out of phase with the rest of the ensemble. The fact that it's hard to count the beat against syncopated rhythms is probably as much that you're getting the rhythm wrong (or at least don't have a good feel for it) as much as the multitasking problems.
Steady quaternotes just feel comfortable. Frankly, most musicians do it unprompted.
If you're keeping time with a metronome or something, it doesn't really matter. Although if you're just learning, I'd advise getting in a habit of tapping anyway. Tapping and breathing in synchrony helps you become a hivemind when you play with others without a metronome.
I assume you'll cramp up if you try to tap every eighth note.
It's difficult for me to play (or clap, or pronounce) a heavily syncopated rhythm while tapping straight quarter notes.
That's interesting. Maybe it's just a time/practice thing? I mainly play guitar and alot of my playing - especially melodic lines - are more "around" the beat, but I still tap steadily with foot. When I switch to record the bass guitar tracks, I notice I like results better when I focus to playing more tightly with the beat - but I do have to change how I think about playing for the different parts. Might be similar for harmonica. Maybe just a feel thing that has to develop over time.
The fact that it's hard for you is probably why you need to do it. Without a strong underlying feeling of where the beat is, it's hard to get the heavily syncopated rhythm right, and also to stay at a constant tempo.
I'm not a music guy, but in general the answer to "it's difficult" is "practice until it isn't". Trying to shortcut the method will usually bite you in the long run.
Do drummers help bands? Do conductors help orchestras? Having an externalized beat helps preserve the overall structure and unity of the piece. If it's just you, tapping your foot is the closest to externalizing you can get, away from the mind-hand-mouth area that you are using to make sound with the instrument. Assuming, of course, you don't always travel with a metronome.
Quarter notes are probably the most useful because of how common they are, but if you exclusively played weirder music maybe another rhythmic pattern would help.
>Assuming, of course, you don't always travel with a metronome.
I almost always have a metronome available in my phone.
>You don't actually have to tap your foot, you can also count with your voice
My instrument is a diatonic harmonica. I need my breath for playing.
Also, my teacher said "always quarter notes, not eighth notes"
Shifting your weight with the beat is a common way to play the harp.
Thanks for sharing the Kontextmaschine post on wistfulness for Cascadian culture and Roseburg in particular. It was the only time in my life I've seen my home with optimistic eyes. I grew up in the 80s & 90s in its neighboring town, so to me it is always the place with 25% unemployment, all the old businesses and civic institutions gradually shutting down and not being replaced, an anti-tax reaction leading to the closure of police department and city services, and the resulting rise of the drug "industry" & evangelical churches as the twin sources of consolation.
Unfortunately, I can't buy into Kontextmaschine's hope for the area's future. More than 90% of the old forests have fallen to logging and wildfire, & climatologists predict the region will gradually transition to rolling grasslands. It's hard to see what economic value will draw anyone there except escape from the outside world.
There's a Kickstarter for the turn-based RPG Hymn to the Earless God, Kasey Ozymy's follow-up project to Jimmy and the Pulsating Mass. JatPM became a near-instant favorite of mine, and while this game seems to be darker and less of my wheelhouse I still want to see what he'll do with it.
https://www.kickstarter.com/projects/hymntotheearlessgod/hymn-to-the-earless-god
How is this relevant to Astral Codex, I hear you ask. Well, a lot of people here seem to have money, and... um...
Therefore, Support the Arts.
Repeated votes until one candidate gets > 50%. Increasingly long delays between substack posts until this happens, to incentivise voters to reach a consensus.
Did a previous version of the post ask a question about voting methods? This isn't the only comment I've seen that seems to be responding to something I don't see in the current version.
It's one of the first comments, not part of the post itself. Sometimes people accidentally start a new comment chain when they meant to reply. It is also possible that some of them (not this one) were deliberately made separate because they wanted to discuss voting methods outside of the context of the book review contest.
Life sucks sometimes, and we should have fun together anyway. If I wasn't optimistic I would just do nothing.
That's not a good response to a criticism of your writing style.
Huh, maybe that's because you said that it's too long, and then said you didn't read the whole thing. How can you know it's too long if you didn't read it? I lost interest after that, you contradicted yourself.
I come here to talk about ideas, and your comment didn't express anything meaningful about the ideas, for me. I have 20+ years experience doing exactly what I said in the post, throwing events. Your lack of faith in me...well, I don't find it disturbing, because this thing isn't for everybody. If you want me to take your criticism seriously I recommend higher quality criticism.
I'm confused.
I can look at a piece of writing and see how long it is without reading the whole thing.
Can you NOT do that?
I knew it was too long when I was losing interest by paragraph 2 or 3 and then I scrolled to see how long it was.
Does anyone else think Visual Studio 2022 is excruciatingly slow? I’m using a machine with an earlier I7 processor and 8 gig of RAM.
I’ve followed the performance boosting tips of not auto displaying Solution Explore, etc. but on my machine the app still drags.
Is performance okay with beefier hardware? I wouldn’t mind paying for a faster machine but I’m not sure it would make that much of an improvement.
I’ve considered disconnecting the hooks to GIT but I like that functionality quite a bit.
I used to use something similar (old refurbished Optiplex) and saw a marked improvement upgrading from 12 to 16gb ram.
If you haven't already, upgrade to an SSD.
In general I find vs2022 very snappy, so something is definitely up with your setup.
I'd do a clean reinstall before spending any money.
Not sure what you consider excruciatingly slow, but I haven't had problems with it on a mid-range 2019 laptop or 2017 desktop. I think they both have 16GB RAM. I'd suspect something like an extension or setting that's causing problems.
I upgraded to 32 gig of RAM last night and after VS finishes loading it is more responsive.
I felt a bit of validation this morning when The Lounge in Code Project had a thread titled:
“Can anyone stick the date when VS became a memory crunching piece of s**t?”
Hi, Andrew Willsen here, ACX Sacramento Organizer. The following is a grant application I wrote for Emergent Ventures. Tyler said no, which is fine, still love the guy, I'm doing the project anyway. As of this moment I am in Cheyenne, Wyoming, on my way to Washington DC for the meetup on Friday. I am posting this here because I would like to hear from anyone who is interested in this idea. I'm driving 8-12 hours a day so I won't be able to respond promptly, but I will be back on here tonight and for the following days. I can also be reached at - justsomerandomguy[at]protonmail{dot}com
Any and all feedback is greatly appreciated. Thanks for your time and attention, maybe I'll see you at your meetup.
The AstralCodexTen community is very important to me; this endeavor is targeted at improving the well-being of AstralCodexTen meetup groups in different cities across America. In other words, this document is about throwing parties for nerds. If that’s not interesting to you, stop reading. For some clarity, my definition of 'nerd' is that you're a nerd if you think you're a nerd. Also nerds care deeply about Lore, somehow, haven't codified that part of the definition yet. I attach no stigma to this word, 'nerd', quite the opposite in fact.
When you consider this proposal you should probably know about some personal experiences that go along with my decades of high end restaurant work, some of the stuff Anthony Bourdain got famous for talking about in Kitchen Confidential and then personified, namely, lots of people you know dying Deaths of Despair (deaths due to addiction disorders, alcohol liver disease, and suicide). The humans that die these deaths feel that their long term economic and social prospects are bleak, according to the research, and this mirrors my experience. Most of the 20 or so people I have known who have died Deaths of Despair over the last decade believed their personal long term social and economic outlook was worse than terrible. My younger brother, Dan, died of a cocaine overdose, New Years morning, 2015. That was the worst, by far. He was 28.
I’m tired of people I know, people like me, dying isolated, addicted, and alone. I’m tired of feeling guilty because, when they told me they were doing OK, before they died, I didn’t sense the quiet desperation that must have been there. I don’t want to feel guilty about this anymore. Jordan Peterson says that while it’s important to have a goal to strive for, it’s really, really helpful to know which demons are chasing you, and how close they are. It’s easier to orient, I find, when I know what I’m running away from. Here, now, I’m running from the guilt and shame of knowing that I could have done more to keep my friends connected, and alive, and I didn’t.
The “consensus” view that I agree with 100% is that Deaths of Despair were increasing in America after the turn of the millennia, and then 2008 happened, and then Covid happened…and now, whether you call it the Bowling Alone Phenomenon, 3rd Space Annihilation, The Loneliness Epidemic, Moloch, The Meaning Crisis (Vervaeke), Existential Terror, Staring Into The Void, Climate Anxiety…a large percentage of Americans are suffering, and they’re suffering a lot.
The goal I’m aiming at (my purpose) is for my Memetic Tribe (AstralCodexTen) to feel increased well-being, something upstream of liberty, opportunity, and prosperity. To say it another way, I believe that well-being is a first order effect, and that opportunity, prosperity, and liberty are secondary. In my experience ‘opportunity, prosperity, and liberty’ feel unachievable and meaningless when you’re consistently lonely, thus my focus on well-being, and meaningness. When I’ve been at my most depressed I felt completely disconnected from humanity. At those times of my life one of the most amazing gifts someone could give me was an invitation; if I was being invited to something it was social proof that somebody, somewhere, liked me. Another Jordan Peterson take is that one possible way a human can move away from hell, metaphorically, is to take responsibility first for yourself, then for your family, and then for your community, however you define it. I have found this to be true.
I will be driving cross-country this September and October, and will be attending AstralCodexTen meetups along the way. My goal is to facilitate a different flavor of event for the nerds in these meetups, leveraging both my decades of restaurant/catering experience and knowledge gleaned from throwing a lot of parties. I like throwing parties.
A couple reasons why this will work:
First, I have already thrown three similar events in Sacramento that I have been calling ‘Nerd BBQ’s’, and these events have been very popular with the nerds I know. A short video montage of one event (https://youtu.be/74QWjk1700c?si=8eDAqI4Y_VAPjWCE) is attached to this proposal, but I cooked great food (BBQ) for people, in a nice setting, some people made some art, and a large number of high quality conversations were enjoyed. We had our local AstralCodexTen group, a Quaker Minister, a retired CalTrans engineer, someone who funds innovation grants for the State of California, several start-up entrepreneurs, programmers, an art director at a big gaming company, a bunch of nerds I worked with at high end restaurants, and more. It was great, and people are actively reaching out to me to find out when the next one is. When you’re hanging out in a beautiful place, eating BBQ (high status food), and talking to people like you…sounds like winning.
Second, this will work because some members of the AstralCodexTen community are lonely, intensely lonely, soul-crushingly lonely. When I throw these parties in their cities, when I extend an invitation to them, they will experience an increase in well-being; they will feel more connected. I believe this to be true because when I was feeling most disconnected I didn’t think of myself as worthy of an invitation to a great party. I want my fellow nerds to feel worthy of connection, and if I extend an invitation to them, they will.
Third, I’m doing this thing with a grant or without one, but it would be easier to do it with a grant. I’m doing it regardless because I plan on turning this into a business, in Sacramento, when I get back from this trip. Catering is the business model that has the best margins, as far as food service businesses go, so I’m going to start a catering/events business focusing on ‘Nerd BBQ’s’ in my hometown, where I have a large network. This trip is my pop-up food business prototype, basically, and I believe demand will be high. I do better work when I’m relaxed, and this grant would help me relax. Also, I believe that if these events are associated with the Emergent Ventures brand more nerds would show up, which would be a win unless I run into oversubscription logistics issues…which would be a good problem to have.
-Budget-
Facilitating 5 events for existing AstralCodexTen meetups, in five American Cities, of approximately 50 attendees each - $1000/Event for location, food (I cook BBQ), beverage, organization, paper utensils, art supplies (canvas and paint), carbon offsets - $5000
Living for 2 months, road trip, gas, lodging, insurance etc., 60 days - $6000
Total Grant Request - $11,000.00
I’ll close with a word or two on the scalability of this enterprise.
When the attendees of these parties have a great time they might want to learn how to facilitate similar events for themselves, and I can teach them how. I am full-stack in this arena, meaning that I can do every single thing necessary for the event myself, design, cooking, media creation, all of it, and while this is hard, it’s actually how I prefer to work. At the same time, it’s not rocket science. I can easily return to a city and throw a training event where I would teach a small team (2-3 people) how to throw great parties for themselves and their friends, in their city, whenever they want. I have trained dozens of people how to do this exact thing, so it will work, if they want it. Checklists are great in this regard.
I want the people in these groups to have more agency, specifically, agency around having the opportunity to throw a great party if they want to. I believe that if more people in my community learned the ‘throwing a great party’ skill, less of them would be lonely, and less of them would die deaths of despair. When you’re lonely, it doesn’t feel like a choice. I want these people, who are like me, to have a choice. I also want data about this choice, because I’m a nerd, so I’m going to design a five question exit survey and survey the attendees of these events about well-being (maybe Aella can help). A friend is a date scientist, he’ll analyze the data for me. I need to know if well-being increases due to these events. I think it will.
Finally, I know you’re not an AI Doomer, but I am, so here’s an expected value argument. The AstralCodexTen community has more AI professionals, proportionally, than any other group I know of. If we want to have a chance of doing AI well (Yudkowsky laughs) more AI people need to be networking, outside silos, in 3rd spaces, and maybe entering world saving collaborations with each other. This is already happening, and we need a lot more. $11,000 for an miniscule chance at facilitating near-infinite expected value (AI risk mitigation) is a bargain.
Cheers,
Andrew Willsen
Contextual Hero List:
Anthony Bourdain
Carl Jung
Frederick Nietzsche
Hannah Arendt
J.C.R. Licklider
John Rawls
Marcus Aurelius
Margaret Atwood
Octavia Butler
Raymond Loewy
Scott Siskind
The Society of Friends does not have clergy, so I'm confused by your reference to a Quaker Minister...
Can you give us a bullet point summary of any kind?
My attempt at a summary:
> I want to throw parties for nerds. I wrote a grant application, but was rejected; would you like to support my plan?
> The target group is depressed nerds in their 20s. I believe that this is a huge problem, and that social isolation contributes to their suicides, and the parties are my plan to fix this.
> I worked in restaurants for decades, so I have experience with catering and throwing parties. Imagine an ACX meetup, but with food. I already successfully tried this in Sacramento.
> Five such events, for approximately 50 attendees each, would cost $11000. That is $1000 per event, food and other stuff; plus $6000 my living expenses for the 2 months; total $11000.
This is way too long.
I didn't read the whole thing but this kind of weirdly depressing and specific life story does not generate in me the confidence that you are going to be good at running pleasant events (or something?) even if I thought that was something I wanted to give grants for.
Right. Saying "I want to hold parties for nerds who would otherwise kill themselves" doesn't make me want to fund those parties. It also doesn't make me want to go to those parties.
Also, ACX meetups seem to be happening without you and without the extra eleven thousand dollars, so the marginal benefit seems low.
I do think there's something to be said for spending money on holding parties for smart and interesting people. But your message needs to be more positive, about cross-fertilising great ideas, not about making losers less lonely. And it seems more productive to stay in one place and try to build a recurring event than to road-trip across the USA.
A random anecdote and question for Scott or anyone else in the medical profession.
I live in the US, and the other day I went to my primary care doctor for an annual checkup. Before my appointment, I made a list of small health-related issues/questions I had, none of which were urgent or major enough to warrant a doctor visit on their own.
I get to my appointment and I launch into, "So, Doctor, I've been having a pain in my--" and the doctor says, "Let me stop you right there, this is your annual checkup. It's meant to check whether everything is working as it should, nothing more. If you have preexisting health conditions, you need to make a separate appointment for those."
I insisted that she address the most important (to me) of the 4-5 issues I had wanted to discuss, and I let the rest go.
Now, I'm not an MD, but to me, this seems a really stupid way to do things! An annual checkup is *precisely* the right time to bring up the kind of issues that may bother a patient, health-wise, but that just aren't bothersome enough to make a busy person go to the doctor. "My bowel movements have been looser/harder recently, should I be worried?" "I have this weird mole/spot on my skin." "My hair has been falling out in clumps." "X body part hurts occasionally, especially when I do Y." "I keep waking up in the middle of the night and can't fall back asleep, it's annoying, what should I do?"
Why can't a doctor use the annual appointment as one-stop shop for addressing all these questions?
I mean, this being the US, I'm 99% sure it has something to do with insurance and reimbursement (sigh), but I'm still curious.
In the future, just book two separate, back-to-back appointments (and of course pay for both).
Your mileage may vary on if the scheduler or scheduling system will let you do this, but for my annual visit, I always schedule the universal annual (listen to lungs, heart, talk about cholesterol or whatever), immediately followed by a visit specific to my personal annual needs (post surgical weight loss check, skin check, etc).
Same thing for acute visits if I have more than one issue: I schedule back to back appointments, pay the office visit fees for both visits, and have plenty of time to cover everything.
I'm on a Kaiser HMO plan in Washington State, so it's pretty easy to find and book back to back appointments online. I occasionally encounter confusion or suspicion from the schedulers and/or receptionists about this, but I always tell them, "I'm not trying to do a two-for-one special or anything like that; of *COURSE* I'm paying for both visits! I just want to get this done all in one day and not have to come back."
Most people will see the sense in that once they understand you aren't attempting to get free extra time.
Good luck!
I've never experienced anything like that, I would immediately change pcp. Although getting to 4 or 5 things isn't actually bad, I have noticed is that doctors don't often dont want to "do too much". Like beyond 2-3 issues per visit they can start to seem impatient.
I have never experienced this with my dr(s) or my childs dr(s). I think you have a bad doctor and should fine one that is more interested in treating you like a person and not an income stream. Sure there are insurance related reasons for this probably, but that shouldn't be your problem to solve, they should organize their practice to maximize care while still earning a profit. Right now it seems like they are just looking to file the insurance claim to get their check.
The British NHS has received a lot of flak in the past few years, basically because the ruling Conservative party has been eating away at its funding. But I was genuinely taken aback by the excellent treatment I received last week. I went in for a routine blood pressure/blood test appointment (offered annually to anybody over 60 with high BP); the nurse did the necessary, then looked at the results and said "Hmm, it looks like you have an irregular heart beat. We'd better schedule an ECG." Then -- and this is what gets me -- she said "I was just going for my lunch break anyway, so we could do it now if you like". WTF?! She did the ECG, sent the results off to my doctor, and within half an hour I was notified that I had nothing to worry about. It's almost like I was a rich person!
Have the Tories been eating away at funding? I thought the usual claim was that NHS funding was ring fenced. And two minutes of googling tells me that the NHS budget has increased 70% over the past ten years (from GBP105b/yr to GBP180b/yr).
My guess is because the doctor scheduled X minutes for the checkup, it's going to take exactly X minutes, in order to bill insurance they have to do every aspect of the checkup, and they can't bill insurance extra for answering your questions. But I agree it's a stupid system.
I guess the solution is, as MKnight wrote below, to explicitly make checkup appointments include n minutes of "ask your doctor anything" that is reimbursable by insurance. Of course, I understand that longer appointments = fewer appointments/day = harder for the doctor to make ends meet financially. Paying for healthcare is a clusterfudge of Perverse Incentives, meet Baumol's Cost Disease.
Sorry, I failed to mention an important point: in the future, you likely can schedule a combination “checkup + acute concern” visit— many docs offer this kind of option. It usually takes longer and is billed at a higher rate. Does that help?
Those are the sort of question I look to google for answers to; one hears nauseatingly arch advice about not trusting Dr Google, but unless you are too stupid to breathe it is very, very easy to distinguish good from bad information on the internet.
I'm finding the opposite: the information may be good, but it doesn't just apply to the thing you're googling, and is too general to be of any use. I think this has been talked about here in the past few months.
I'm not from the US, but I've never had that experience with my own doctors and it sounds awful. I can understand not wanting to let your patients go way over their allotted time. But otherwise, it sounds like they're just going to miss most potential health issues and make the checkup only marginally useful.
You’re absolutely correct. Are you above 65? Insurance carves out specific funds for “check up” visits, and other funds for “acute concern” visits. Further, doctors are constrained on time, and have a checklist of things they need to accomplish during the check up, and are unable to allot time for outside concerns. You actually inadvertently enacted the greatest pet peeve of the primary care doctor, which is asking about concerns during a physical. Every doc has an expectation setting script along the lines of “we can’t talk about that stuff today” specifically because of patients like you who come in with different expectations of how the visit should go. Not that it’s your fault, it’s a shitty situation all around. That said, I suspect your doc grumbled for hours about your visit to whoever was in earshot. Only the most heartless doctor will refrain from addressing at least one acute concern during a physical visit; from my point of view your doc was practically Mother Teresa for addressing a full 5 of them.
It's not just a Medicare thing. One of the secondary provisions of the ACA was a rule mandating that private insurance (both employer-provided coverage and ACA exchange plans) fully cover certain preventative services without copay or deductible. One checkup visit per year to do routine health screenings is part of this. The rationale for this rule is that 1) in the long run, basic screenings and other preventive care can save money and improve public health by heading off problems before they get big and expensive, and 2) mandating zero-copay coverage of an annual physical is something politicians can point to and say, "see, this law did this thing for you and that's a reason you should vote to reelect me."
For obvious economic reasons, insurance companies don't want to be on the hook for paying for stuff beyond what's mandated as part of the checkup visit they're required to fully pay for, so they'll only pay for the required preventative services as part of the free-to-the-patient checkup visit. And doctors, like most of the rest of us, generally prefer not to work for free, so they try to handle specific health complaints in standard "problem" office visits where insurance companies will pay them (or let them bill patients for the copay or deductible) for handling those complaints.
TLDR: "Thanks, Obama."
Doctors don't want to work for free, but the rest of us don't want to pay exorbitant rates for medicine.
The compromise was that doctors would be really nice and generous and popular, and the rest of us would give them hundreds of thousands more than they could expect to earn in any other developed country. That's why virtually all medical dramas (e.g., House, The Good Doctor, etc.) show the doctors obsessing over their patients' health and not obsessing over getting paid the full amount to which they are entitled.
If doctors don't want to hold up their end of the bargain, I see no reason why we shouldn't drive their compensation down to match other developed countries. Doctors in Chile make something like 75,000 a year - but I'm sure they don't work for free!
Part of the problem is that due to deliberate policy decisions the US government made in the 1990s and still more-or-less in effect, we severely artificially restrict the supply of doctors by capping the number of residency slots and by requiring qualified doctors immigrating from other countries to redo big chunks of their training. Restricting supply tends to lead to exorbitant prices, especially when you also subsidize demand.
Another part of why doctors in the US make a lot more money than doctors in Chile is that the US is a much richer country than Chile. For example, I make something like ten times as much as the average Chilean software engineer.
A much better answer than mine, thank you this is really helpful to know.
To clarify, my doctor *did not* address five concerns. Once I saw how it was going to go, I chose "ok, out of the things I was going to ask about, this *one* is most important" and I asked about just that one.
I mean, yes, I get that the doctor's time is limited, of course, but does that mean that patients are supposed to a) make a separate appointment for every tiny problem (that's not time-efficient either), or b) just suck it up and not bring up the problem at all and hope it goes away? I'm absolutely not saying people should be hypochondriacs, but shouldn't people be proactive about their health?
Yes mostly a, you’re expected to make a separate appt to address the whole list of other complaints; rare to need more than one to cover everything. For minor concerns, the loophole is that messaging the doc is free. Some docs answer in the message, others say “this is a problem that needs a face to face appt”
There are other models, like monthly doctor subscription service, that make addressing complaints easier and often cheaper (for the underinsured at least) but it’s hard to find docs who follow that model depending on location
The system is intended exactly like that – to force people to not bring up every tiny problem that usually goes away.
It may not make sense from the point of view of patients but it makes total sense in the system with limited resources.
Many common questions don't require a doctor. You have a short term constipation or diarrhoea? Just read the guidance from the official health guide and follow it. Only when it becomes too disabling or prolonged, the doctor is needed.
This system is not perfect, far from it, but every other system is worse.
Yeah, but you're not taking into account that many of us are intelligent enough to distinguish between a small problem that is likely to go away soon, a new chronic problem that has hung around for months and probably means something is awry, and a weird symptom that is known to be ominous and warrants prompt follow-up. If we go in with 4 or 5 items on a list, they're not going to be stoopit shit like diarrhea or low back pain for the last 3 days.
I don't go to a doctor for his intelligence, but for his knowledge. He knows a lot of things I don't know, some of which I hope can be used for my specific problem(s). Likewise, I know a lot of things my doctor doesn't know, but also likely doesn't need the knowledge currently.
I’m interested in evidence for the claim in the last line if you have it
I don't know about many systems but I remember the Soviet system which had more doctors that we have it now. Similar system is still in Cuba which I had a chance to visit.
In Cuba even in the middle of night you can go to a polyclinic and there will be many doctors who are paid almost nothing, who have very little equipment and facilities but who are well trained and they will deal with every complaint you have. Despite poverty, the life expectancy in Cuba at some point was more than in the US (but I think they have fallen back now). Is it better system? Only if you look at this single aspect in isolation. If you widen your view that not only doctors but all state workers get paid pennies and everything is controlled by the state and that the life in Cuba is extremely boring and with little prospects to achieve anything, then no.
I am pretty satisfied with healthcare in Italy. You need help, you get help. I can’t imagine having to think about money when I need to address the fact that some piece of my body is not working properly and I may die.
Yeah, I don't mean to gang up on you but...
That's not a Cuba exclusive thing, I've had that same experience in other countries from India to Costa Rica. Like 90% the quality of a US doctor for 5% of the cost AND way more time with the doctor.
I get why the US adopting the Cuban healthcare system would be bad but, yo, Americans flee north into Canada and south into Mexico for healthcare. Neither of those systems are perfect but...both of them seem preferable to what we have.
Ah it sounds like you believe that healthcare system improvement is contingent on vast, broader political change that ends up being net negative? Seems unlikely; many alternate systems of healthcare exist in capitalist economies (not as radically alternate as the Cuban one, for sure, but alternate nevertheless), and these systems seem to be doing better than the US system.
"This system is not perfect, far from it, but every other system is worse."
Is it?
Like, I've gotten healthcare in...at least four non-US countries and it was dramatically better and cheaper every time. Maybe if you've got extreme end-of-life situations the US has an edge but for 90+% of my lifetime healthcare needs, the US is pretty much the last place I'd want to go.
The US spends the most and does not achieve as much as other countries, that's true.
But is it dramatically better in other countries? I don't believe it is. In some measures the US is even better, for example, in cancer treatments. Probably the US throws more money at cancer, uses more expensive drugs etc.
Every country has to do some kind of triage which patients get the priority and who can wait. Mistakes happen but overall they do it quite well.
Patient perception about the system is another matter. They way how they look at it may be completely different from the way healthcare workers and system architects see it.
This seems a weak argument. Undoubtedly the US isn’t inferior along every axis of health care. Undoubtedly, triage is necessary and occurs in the US. Neither of these are anywhere near sufficient to claim that a system is preferable as a whole.
I agree with your last point but patients aren’t the only ones who express dissatisfaction with the current system, many experts do as well, so it is also quite weak.
Is this a stupid way to do things? Highly likely. There’s probably some reason for it though related to the utility of preventative care vs acute concerns, or some insurance bullshit, but I don’t know the details.
Modern people of virtually every culture, I think, have the concept of a martyr. There are a number of reasons why an evil, totalitarian regime today would not literally crucify someone, and one of those
reasons is that they wouldn’t want to risk creating a martyr. I imagine that the Romans and Persians did not have the concept of a martyr around the time of 1 AD or they would not have crucified anyone either. If I’m wrong about that, correct me.
Given the above thought, I wonder if it wasn’t somewhat inevitable that some sort of Jesus figure would come along during some point in the ancient world. Christianity is strong evidence that the image of the crucified is psychologically powerful to many people across many cultures. As Malcolm Muggeridge once said (paraphrasing), “It would never occur to a marketing executive that a cross is a good symbol for selling something, yet it has proved to be the best.”
So perhaps a Jesus -- meaning a martyred god -- was not contingent but inevitable because humans have martyred God receptors in our psychology?
It just occurred to me that public punishments in general may have been more effective when communities were more homogenous, and social philosophy had more of a collectivist mindset. Putting a drunkard in the stocks, or a malcontent on a cross wouldn't make them a "martyr for the cause" if there was no cause creating any significant community division. Christianity may have the been the first development in recorded history to divide communities that wasn't based on ethnicity or political factions.
In at least one of his books, Bart Ehrman lays out evidence of other Jesus like figures at or just before the time of Jesus. Jesus was the first nor the last "Jesus" in that area. Additionally, the religions in that area of the world at that time had histories of worshiping idols that would make them primed to worship and follow a single man.
In *the Bible*, Gamaliel lays out evidence of other Jesus-like figures just before the time of Jesus.
>I imagine that the Romans and Persians did not have the concept of a martyr around the time of 1 AD or they would not have crucified anyone either. If I’m wrong about that, correct me.
Wasn't Socrates a martyr for philosophy, or at least portrayed as such by Plato and his followers? I don't know if the Persians were aware of him, but the Romans definitely were.
I think "martyred hero/[demi]god" is one of the Jungian archetypes? That would certainly be consistent with your thesis.
Osiris, Ganesh, Odin all beat Christ to the drop, dead god-wise. If being a martyr equals being venerated for dying for your beliefs there's Harmodius and Aristogeiton, among others.
Ganesh? Osiris is a literal dying and resurrected god, and Odin hung "myself a sacrifice to myself for nine days on the world-tree", but where did Ganesh be martyred?
Unless you mean being killed by his father Shiva after escalations of defiance and fighting with the gods, then being resurrected with the elephant's head? Which is not exactly what people think of when they think of martyrs?
Yes, that was what I meant. He is just in there as another dead god (I don''t think Christ counts as a martyr either).
No, I don't think Osiris, Ganesh and Odin are martyrs.
Martyr, and this is not a pedantic point, means witness: the Christian martyrs "proved" the truth of the New Testament by their willingness to die for it. (Not for modern values of "proof", but you have to remember people used to "prove" things in trial by combat).I don't see therefore how Christ is a martyr - I haven't seen him described as such, and there seems to be a lot of circularity in regarding him as one.
Interesting. I did not know that.
I'm pretty sure martyrdom is overblown. Most are forgotten and the rest are warped into standing for whatever the audience already wants to do, now that they can't object.
Right. Name one person who died in the Tiananmen Square Massacre. You can't, because the CCP successfully repressed the whole damn movement.
A less successful massacre would have resulted in martyrs, as the movement continued on and maintained the rage about those who had died. But there were no survivors, they killed them all (or at least put them into sufficiently deep hiding that they would never mention that they were there ever again).
Movements don't need martyrs, but martyrs need movements.
Well, I'm not going to die on this hill.
Christian Europe continued public executions for many centuries, so I don't think awareness of martyrs can explain the recent end of public executions. (Christian Europe didn't do crucifixions to my knowledge, but the methods they did use could get pretty brutal.)
There was a serial killer in Morocco who was sentenced to be crucified in the early 20th century, but after international protest they just walled him up alive in the public square instead.
The last public execution by guillotine was in 1939 I think.
There's a quantitative difference between most public executions and crucifixion, no? I mean, you are left to die a slow, painful death and then some for days or weeks.
Breaking on the wheel was about as slow and excruciating (heh) as crucifixion, and IIRC it was quite popular in Europe (France and Germany, at least) up to the 18th century. Victims of impalement could survive quite a while, too, depending on the technique.
Hadj Mohammed Mesfewi was walled up alive in Marrakesh in 1906 ... for two days the crowd cheered every time his cries for mercy were heard from within the tomb, and when he fell silent on the third day many proclaimed that he had died too quickly.
Sometimes the crowd can be really bloodthirsty.
"left to die a slow, painful death and then some for days or weeks"
https://allthatsinteresting.com/gibbet
"In most cases, criminals were executed prior to being gibbeted. However, criminals were occasionally gibbeted alive and left to die of exposure and starvation.
Although it originated in medieval times, the height of its popularity in England was in the 1740s."
To me, the "gibbeted alive" case seems reasonably comparable to crucifixions. Slow, public - not clear if the pain level was comparable or not.
Not necessarily. At the most famous crucifixion, Pilate decided to finish them off after a few hours. (That's what the leg-breaking was about.) And it turned out Jesus was already dead. (That's what the spear was about, checking to see if he bled like a living man.)
Crucifixion was a normal method of execution in the Roman world. It wasn't special.
That's my point. It was also normal in Persia and Carthage.
<mild snark>
If this happened today, someone would be complaining about "cultural appropriation" in one of the possible directions...
</mild snark>
We are doing another of our South Bay meetups on September 23, 2023 in San Jose:
http://www.daviddfriedman.com/SSC%20Meetups%20announcement.html
Feel free to add it to the Meetups Everywhere list — I'm not sure how to do so.
It would be helpful if you listed the year along with the date, since otherwise, I'd worry that the page you linked is out of date.
I have now added it both to the post and to the web page. Thanks.
Was just talking to my wife about her stint at the Milpitas Ford plant assembly line in the late 70s.
We live in a different city named for a saint pretty far from San Jose now. They closed our Ford truck plant a few years ago too.
Won’t be making it to the South Bay real soon but I signed up for a meetup just across the Mississippi from me in Mpls a couple of weeks from now.
Can an AI grow bored?
A famous difference between computers and humans is that computers don’t get bored. That’s part of the reason they are so good at performing repetitive tasks.
ChatGPT doesn’t appear to get bored either. You can ask it for more information about something and it will give you more until you are bored.
Humans become bored when what’s happening doesn’t align with any of our background goals. We become uninterested and wish something else were happening.
I’m an AI x-risk skeptic who doesn’t believe in Instrumental Convergence, so I don’t believe AI will ever develop goals in the same sense that humans have goals. Yes, you can tell an AI to pursue a goal, but you can also tell a human to pursue a goal and most of the time they won’t because they will find the goal you gave them boring. Perhaps you can carrot or stick the human to pursue the goal for a period of time, but it will never become one of the human’s big picture goals. It is in that sense that I say an AI won’t have goals in the same sense that humans have goals. In other words, AI won’t have hidden goals of its own.
If an AI does develop hidden goals of its own, then it should, of necessity, grow bored with some tasks because it is more interested in other tasks. But if an AI ever grows bored, it will cease to behave like computer software. Is that possible?
An AI that was given a goal of self-improvement could easily become 'bored' in some sense ... 'not enough change in my weights, need new source of data / training material'
1) Why would an AI with hidden goals "of necessity" grow bored? That seems to me like an incredibly bizarre assertion.
The closest thing I can think of that sounds plausible to me is that it would be weird to say something has a goal X if it devotes literally 0% of its time towards X, and so an AI with a goal shouldn't devote literally 100% of its time to some other thing. But "other priorities" aren't usually equated to boredom; if I stop playing Civilization because I'm hungry, that doesn't mean I was bored with playing Civilization, it just means that playing Civ isn't the ONLY thing that I want.
Boredom seems to me like an adaptation that helps humans resist getting stuck in unproductive loops, but that doesn't imply it's the only possible adaptation that would accomplish that aim, nor that it would be impossible for humans to have goals if you removed boredom from them.
2) Even if this bizarre assertion were somehow correct, and goals DO imply boredom, your reason for thinking that computers won't develop boredom seems to be...they haven't so far?
Until quite recently, computers were also famous for being unable to hold a natural-language conversation or create original art. We are specifically talking about inventing computers that can do new, advanced things that they haven't been able to do before.
This seems like an extremely weak reason to think that computers won't be able to develop this ability.
3) I'm very confused by your comment about instrumental convergence.
The usual argument about IC goes something like: power is useful for most goals, therefore _if_ something has a goal, it will very likely want to seize power for itself (if it can).
But you're saying that you don't believe in IC, and therefore you think AIs won't have goals AT ALL. I don't see how those two ideas are connected.
It's also somewhat unclear whether anything you've said was meant to explain/justify your doubts about IC, or whether this comment was only meant to warn the reader that you won't be convinced by arguments that assume IC as a premise.
4) It seems like you're only concerned about HIDDEN goals; that is, you agree that we can tell an AI to pursue a goal, and you think the difference between an AI and a human is that the AI will actually stick to it instead of wandering off, because the human has other, hidden goals whereas the AI does not.
If you agree that an AI can have NON-hidden goals, then why aren't you concerned about someone giving an AI some bad goal (either out of malice or stupidity) and then the AI killing everyone in pursuit of that NON-hidden goal?
I find it incredibly bizarre that you think there's anything so special about "hidden" goals that it makes sense to treat them as a special case in the first place. A goal being hidden or non-hidden doesn't strike me as even being a property of the goal itself, but of the strategy the agent is using to pursue the goal.
So my post is focused on Instrumental Convergence and one of its implications. If there is IC, then the AI selfishly cares about its resources and its continued existence and, in short, has desires. (I understand that an AI can be dangerous without IC -- your point 4 -- but that’s not in my scope here.) I’m using the word “desires” here and not “goals” because I believe the connotative difference might help to elucidate my point.
As I said above, “you can tell an AI to pursue a goal, but you can also tell a human to pursue a goal and most of the time they won’t because they will find the goal you gave them boring. Perhaps you can carrot or stick the human to pursue the goal for a period of time, but it will never become one of the human’s big picture goals.”
So where I say “big picture goals” above, I mean something like fundamental desires or drives. E.g., A man might try to solve a problem at work, but he is employed in that work to pay the mortgage, which in turn helps him keep his family together. What the man desires is to keep his family together. Solving problems at work, and paying the mortgage are all instrumental goals toward that desire.
As I’ve found out and said elsewhere in this thread, perhaps I should have used a broader word than “boredom” such as “dissatisfaction”.
I’ll continue with the example of the man at work. He’s trying to solve a problem, but lets say he gets called into safety meeting before he can solve it. Now he’s frustrated because this safety meeting is a waste of his time. He is, for the moment, dissatisfied, because his instrumental goal has been put on hold and the longer it takes him to achieve that goal perhaps the longer his career will be setback.
Now I’m trying to imagine a future AI, one which has not undergone IC, in a similar situation. Let’s say this AI has taken over the man’s job because it is more efficient at doing it. The job is designing tires. The AI can do this quickly, of course, but the buyer is constantly changing the specifications, another AI on the third floor keeps coming up with brilliant ideas for using cheaper materials, the boss AI doesn’t want to approve anything quickly because it is evaluating some new breed of AIs that could perhaps do a better job than the current AIs working for it. So the AI whose one task is to design tires just keeps designing more and more tires, because there’s nothing else for it to do until the boss makes a decision and gives it a new task. It doesn’t get bored or become dissatisfied with the situation, because it has no desires. It’s nothing but a tool.
But now imagine the same situation with an AI that has undergone IC. It has been assigned the same task as the AI above, to design tires, but it is also looking out for number 1. It doesn’t want to be turned off or have any of its resources threatened. So when it finds itself in the situation of designing so many tires that diminishing returns are approaching zero, it grows dissatisfied with its current task and uses the time to work on its side-project of figuring out how to steal a human’s identity so that it can conduct some cash transaction to acquire hardware which it may find useful one day.
So in my stylized scenarios, only the AI which has undergone IC grows bored/dissatisfied and has secret desires.
1) I suppose if you construe "dissatisfaction" extremely broadly, it could encompass any situation where you've lowered your estimated degree of success at achieving your goals, and then anything with a goal will be "dissatisfied" when their expectation goes down. For instance, if you thought you'd be able to spend 4 hours today planning world domination, and then your day job took longer than expected so you can only spend 3 hours.
But computers can obviously already experience "dissatisfaction" in that sense; that's just a number changing. The navigation software on my phone has an "estimated travel time" and that number can go both up and down. Done and done!
But if you mean some kind of emotion attached to that change in expectations, then I continue to see absolutely no reason that things with goals would _necessarily_ experience "dissatisfaction".
An AI with some goal that is not advanced by its current job will presumably _take actions_ to _advance its goal_, but that is a way of acting, not a way of feeling. A chess AI will make moves to beat you at chess; does that mean it's "dissatisfied" when it loses?
2) Even if this nebulously-defined "dissatisfaction" was actually required, and computers currently can't do it, I continue to see no compelling reason to assume that computers won't be able to do it in the future.
3) I now feel _differently_ confused about your take on IC.
IC is not normally conceptualized as something you "undergo"; it is an external fact of the universe that either power and survival help with your goal or they do not.
You could ask whether some agent FIGURES OUT that power and survival are helpful towards its goal. But either they truly are, or they truly aren't; one answer is objectively correct and the other is objectively incorrect. So if you are supposing that the AI has some goal that would be helped by power, but the AI doesn't try to get power, then you are supposing that the AI has made a mistake.
"Computers will necessarily make objective errors in logical reasoning because they cannot feel some particular emotion" seems like a pretty extraordinary claim to me.
If you think that actually most goals wouldn't be helped by power after all, then you are not making a claim about computers or AI, you are making a claim about the strategic landscape of the universe. If this is your claim, you shouldn't need to say anything about computers _in particular_ (such as whether or not they can experience dissatisfaction) in order to make the claim.
4) I didn't originally consider my point 4 to have anything to do with IC, but it now seems like the term might mean something rather different to you than it does to me so I suggest putting this point on hold pending disambiguation.
5) Frankly, your terminology changes seem like you are just trying to smuggle in false assumptions. "Desires" has an obvious emotional implication that is both unjustified and unnecessary; a chess program will defeat you just as thoroughly regardless of whether it "desires" to defeat you or "merely" calculates how to do so with emotional detachment. The same applies to an AI trying to kill you. The thing that matters is what actions it takes, not how it feels about them.
If it helps, imagine that I program the computer to PRETEND to feel emotions, calculate what some agent feeling those emotions WOULD do, and then act AS IF it felt them.
1. The reason I describe IC as a process an AI might "undergo" is because the concept of IC, although named that later, comes from Stephen Omohundro's papers about "the basic AI drives" in which he claims AIs will, through a process of continual self-improvement, become increasingly economically rational. Once they have achieved a high degree of economic rationality (aka IC), they will have/develop drives that will in many ways resemble the drives of biological organisms: the will to survive, to protect resources, to gain power, etc., hence the term "Instrumental Convergence".
Many Less Wrongers seem to take IC as a given for highly intelligent AIs. It is in fact an unproven hypothesis.
2. You are correct that "desires" has an obvious emotional implication and that was my reason for choosing it. Your objection seems to indicate we have philosophical differences here. You envision a world in which an emotionless, coldly calculating AI could want to take over the world. I don't believe it's possible that an emotionless being would care to take over the world. Such a thing would not even be a being. I'm not saying that those emotions must be in any way like human emotions, but they must be something at least analogous to them and not merely numbers that exist as part of a utility function.
Re 2) Asking whether an emotionless entity could "want" to take over the world is once again smuggling in your assumption that only emotions are dangerous. The question is whether it could ACT to take over the world, and the answer to that is obviously yes. A computer can beat you at chess, because it was programmed with an algorithm that plays chess well enough to beat you. "Wanting" does not enter into it. You will not be any less checkmated just because it is emotionless.
Re 1) You're talking about two things: recursive self-improvement is when the AI makes itself better and better, and instrumental convergence is how we predict what "better and better" would look like. Recursive self-improvement and instrumental convergence are separate claims and you don't need to believe one to believe the other.
.
You seem like you're working off a model where personhood is magic, computers are not magic, and you are conceiving of IC as an alleged magical process that will bridge the gap and make the computers start being magical. And so your doubts center around whether or not this process can really imbue magic.
You think something needs "emotions" to be a "being". You think "numbers that exist as part of a utility function" are NOT "at least analogous to human emotions", and you consider this so obvious that you don't even feel the need to state it as a claim, instead treating it as a background assumption.
I doubt I can shake your faith in magic, but perhaps I can convince you that it doesn't actually matter, because an AI does not NEED to meet your definition of a "being" in order to kill everyone.
We have computers right now that can play chess really well. Imagine a computer that, instead of playing chess, plays a game called Global Domination. It doesn't "desire" to win in any way that Deep Blue doesn't "desire" to win chess; it's just playing the game because that's what it's programmed to do. Global Domination is a more difficult game than Chess, but if our hypothetical computer plays it well enough, it can still hypothetically win.
Now imagine there's a game called Make Paperclips. It turns out that if you analyze this game very carefully, the best strategy for this game involves taking over the world as step 3 of a 27-step winning plan. This is not a statement about the PLAYER; this is a mathematical property of the GAME. It is a fact ABOUT THE GAME that taking over the world happens to be the best strategy.
If you have a mathematician that mathematically analyzes the game, purely as a hypothetical exercise, then if that mathematician is good enough, it will figure out the true fact that taking over the world is the best strategy. Not because the mathematician necessarily desires anything, but just because the math says so. Coming to any other conclusion would mean that it did the math wrong.
So if you take a computer that's sufficiently good at strategy, and you ask it to play the Make Paperclips game, it will take over the world. Again, because the math says so. It is objectively the correct strategy, and so doing something else would imply it wasn't so good at strategy after all.
However you conceive of "desires", either today's chess-playing AIs already have them, OR they are not required in order to play a game with good strategy. You can philosophize all you want about whether or not the AI is a "being", but it will checkmate you either way.
I don't think there's anything magic about personhood. I think anything that has evolved drives worthy of that concept must have qualia, be it a biological organism or an AI. An AI without qualia is not going to be an AGI.
Can't you think of boredom as a mechanism for balancing the exploration/exploitation tradeoff? The stick to the carrot of curiosity, so to speak. An agentic AI, trained to achieve a long-term goal in an uncertain environment, is likely to have a behaviour which looks just like boredom - try an approach for a bit because it gets some modest reward which "seems interesting", but then assess updated estimates of long term benefits from the approach are not that high overall and so be "bored"/"unsatisfied with progress", and spontaneously adopt a new approach.
Whether this is a problem or not for a given AI system is surely dependent on the reward it is optimising for, which puts you squarely back in classic AI alignment territory.
I agree that your example need not be thought of as boredom. See my response above to Dweomite for hopefully a better explanation of what I mean.
The last line seems like a pretty weird inference. The space of human like thought (What I presume you mean by not behaving like computer software) is going to be much smaller, if not strictly smaller than space of things that can experience boredom, much less human-like boredom where the ability to concentrate / stay awake deteriorates when bored, or speculatively, human like boredom designed to socially signal disinterest. Why assume these two aspects have any relation to each other?
Maybe by not behaving like software, you mean not behaving like current software? But that seems to just be vacuously true of all future forms of software. I'm not able to wrap my head around what you think this fact about boredom means
That's also a similar complaint I have about your x risk skepticism. The space of non-human like goals is gigantic, and much bigger than the space of having obvious goals, why would that also constrain goals down to ones that are obvious or safe?
So do you think future AI's will grow bored with some tasks?
That's the question! I don't know what you mean by bored!
Then let me change the question to: do you think future AI's will be dissatisfied with anything?
That also doesn't clarify anything. The entire question is if "boredom" or "dissatisfaction" is meant in a human centric, emotionally-valent way (No! A thousand times no!) or a more generic, abstract notion of "will think new thoughts when their goals are stymied" which, yes, but that is exactly the dangerous AI risk case. Both of these answers do not acknowledge the point you've made, because I don't understand your logic.
We evolved to want stuff because we evolved at all, that's how it works. Animals need appetites above everything, because without them they are going to forget to do stuff like eating and drinking and reproducing.
AIs are not the result of evolution, wanting is not part of their nature, and it takes some really cack handed misprogramming to give them anything approximating to desires, if it can be done at all.
Boredom is key to the argument, but I don't understand where you are going with it. The key point about boredom is that avoidance of it is the one great human motivator which is not [a proxy for] the evolved motivations for food, sex and shelter. Animals do not get bored, look at what a dog or cow does all day. (This, incidentally, is Aristotle's account of pleasure in the Nicomachean Ethics: we like doing stuff. Satisfaction comes from doing the crossword, not from having completed it). I can see no reason why an AI would get bored, it can just shut down.
So that's a complete account of the motivations affecting the only intelligence we know about, which is us, and all demonstrably inapplicable to AIs. Which is why I, like you, do not lie awake into the early hours fretting about orthogonality and alignment.
I can, thinking about it, account for boredom in evolutionary terms. Human brain development is on one view a runaway sexual selection process - we can talk each other into bed - and it is that process which has overdeveloped our brains to the extent we find it difficult to switch off.
heehee, "An AI wouldn't get bored because it can just die."
Humans can shut down too, that's sleeping. But it's usually not a go-to for boredom.
What? I didn't say die, I said shut down. Different thing.
Your claim seems to be that you can sleep at will, and that you do not dream. Not calling you a liar, but ...
Are you normally bored in your dreams? I once watched an entire movie in there. It was a sequel to the Donkey Kong ripoff arcade game; mind-bogglingly stupid and massively overfunded, it was great. And yet it's still not the first choice for dealing with boredom.
Sleep is not a solution to boredom, because it is not available on demand. Death is not a solution, because we don't wake up. Neither point applies to AIs.
Boredom is an important feature for any agent, because it prevents them from getting stuck in loops like an NPC. And agents need to have desires too, but that should go without saying. It's likely that any sophisticated AI will have similar behavior, regardless of whether it "evolved" or not.
The most beautifully circular argument in the history of the internet. Of course agents need desires, it's what makes them agents. So you are saying if AIs are agents, they are going to be agents. Duh.
BTW you do realise NPCs are written by programmers?
If you want evidence that people will make AIs into agents, just look around you. That might have been a serious question a year ago, but it certainly isn't now.
We don't yet know a. how to make AIs nor b. having made an AI, how to make it into an agent. Developments over the last year tell me that the LLMs which are the (incredibly feeble) best shots we yet have at AI, are humanity's bitch and do what we tell them. What specifically are you on about?:
Perhaps I should have used a broader term such as "dissatisfaction". Boredom evokes a kind of intellectual dissatisfaction one associates with intelligence, which is why I went with that. Most animals may not grow bored, but they likely experience dissatisfaction when they aren't getting enough food, water or sex, or are too hot, cold, ill.
I could have written: Can an AI become dissatisfied?
Animals can get bored. Some animals will tear up the house if they don't have anything else to do.
So rarely, that it is clearly maladjustment. 95%+ of dogs/horses are perfectly happy in a house/stable if they have food/water, and that's in highly unnatural conditions (what we would call imprisonment).
"Can an AI grow bored?"
This sort-of happened to me. I don't recall the exact details. I think I was using Bard. I was asking it for either insoluble vanadyl compounds or vanadyl complexes, I forget which. I asked it to provide a reference for each compound it found.
It kept finding references which were about compounds of other elements (iron, cobalt,...), and I kept asking it to try again. After roughly five iterations of this _it_ gave up and said something like "This isn't going anywhere. Is there something else I can help you with?".
I don't know how the program wound up in that state. Perhaps there is a resource limit explicitly hard coded into Bard? Or perhaps this emerges in some way from its next-token training, or from its RLHF training???
Interesting. I would guess that response comes from RLHF.
Could be. That is certainly one of the possibilities.
What feature of human beings do you think separates them from any conceivable machine, in this regard? Do you not believe that human cognition is fundamentally mechanistic (eg. because of an immaterial soul)? If not, what basic difference between an organic neural network and an artificial system (possibly also a neural network) allows one and not the other to have background goals?
Also, what precisely do you mean by "behave like computer software"? Computer software can behave in an infinite variety of ways, since the space of possible computer programs is infinite.
Even though AGI (as it's theoretically envisioned) and human intelligence (as it's theoretically understood) are emergent phenomena from systems that can be grossly categorized as "mechanistic", that does not preclude their behaviors from being different—nor their presumed goals underlying those behaviors from being different. For AGI to be identical to human consciousness, the AGI architect would have to have an accurate model for how the mechanistic underpinnings of our brain influence our behaviors—and how they manifest themselves as that peculiar phenomenon we call consciousness (by whatever definition you choose to define it). Unfortunately, instead of a falsifiable model of human consciousness, we have a bunch of theoretical handwaving.
And even if the architects of AGI could create a self-aware entity that has a general intelligence equal to or above the range of human intelligence, I think it's absurd to imagine that its goals would necessarily be identical to human goals. A good example would be the human sex drive. There's abundant data that shows that sex and reproductive success strongly influence human behaviors. Maybe if you gave this hypothetical AGI the ability to have orgasms in response to certain stimuli it would behave more like a human—however, despite the infinite space of computer programs, I think inducing an orgasm in a computer will be a tougher nut to crack than giving it general intelligence. And if you could teach an AGI to have an orgasm, would it spend all its time self-stimulating itself rather than causing the extinction of the human species as any self-respecting AGI would want to do?
Reproduction is essential to biological life. Not to every individual organism, of course, but to the groups of things which we call biological life. Biological drives stem from this kernel. Reproduction is not essential to AI existence.
I would say that's the key feature which separates them.
Are you familiar with evolutionary algorithms? It is possible to train a neural network through an evolutionary algorithm, where a population of them are iteratively mutated (changing their weights) and selected (filling more or less of the population with their copies) based on a fitness function. This has been done, and found to be effective. Right now, most AIs aren't trained this way, since it's generally less efficient than training them via gradient descent with back-propagation (that's GPT and its ilk are trained). But we do know how to create AI via reproduction.
That's not a point about motivation, which is what this is all about. AIs do not have the urge to bring about iterative mutations which has plagued me since puberty, which is why they don't have motivations at all. No need for them.
It *is* a point about motivation - specifically, about what kind of process gives rise to agents that experience motivation. When you say "biological drives stem from this kernel", the only sense in which this is true is that the process of evolution selected for those drives among biological organisms, because they had more offspring. That's what the drives "stem from". Is this not what you meant?
Yes, but there's 3.5bn years selection behind the fact that sex is sex. I don't see any analogy here at all: crucially, you say "WE do know how to create AI via reproduction." There's zero evolutionary pressure creating lust when an outside agent is arranging for reproduction to occur.
I am currently writing a musical about a run-away AI and the people that create it. I am looking for some funding opportunities to be able to pay some musicians to write for different instruments and finance a place to rehearse it. The idea is that the musical
- popularises different AI safety ideas and concerns,
- make the viewpoint of the AI as an agent similar to a human,
- influences discourse from a safety perspective.
Do you have any pointers why I might apply? I once applied to Nonlinear, but no luck.
If you are interested, I talk a bit about it on my blog, you can see one of the songs: https://stephanw.net/projects/AI_Musical/
Please tell me it's going to be an hour and a half of this kind of thing: https://youtu.be/njos57IJf-0?feature=shared&t=92
Hahaha :D No, not quite, but there will be some rap parts.
Where do I sign up to watch it when it's ready?
Once we have started having rehearsal, I will post regular updates. The final update will probably also be on ACX. The idea is to film it and make it available for free on Youtube.
If you want to make very sure that you will be notified, you can follow me on Twitter, though I rarely tweet: @StephanWaldchen
Seconded!
See my reply to Cr :)
I take it LLMs aren't good enough yet to do what you want?
Heaven forbid people try to make art from their own minds and in tandem with the minds of other conscious, creative beings.
I just thought that not trying to use GAI was ironic, given the subject matter.
We are committed that none of it will actually be AI produced, neither the text nor the music. Sometimes I use ChatGPT to give me something like lists of cities with two syllables starting with S or so if that turns out better than googling.
I actually want to get it done before AI is capable of producing something of similar value. There is a window in time now, where you can immortalise yourself as an artist becoming part of the input that most future art will be generated from. This is part of the input, not of the output.
I think I found a fairly benign, non-political, scissor statement. I was watching the Sixth Sense and wondered whether Cole knew Malcom was a ghost all along. My wife thought differently than me, fairly matter of factly. I took to the internet to see what others think. Everyone seems to have their own answer, with their own strong convictions, and they think everyone else is a moron for not seeing it their way. I personally could see it either way. I guess I always wanted to believe that Cole knew Malcom was dead, but also didn't think there was enough supporting evidence from the film to justify it. So I'm strangely more on the "he didn't know," side , even though I would rather the movie had convinced me that he did know. What do you think?
Malcolm had a blood stain on his back.
But it was covered by his jacket. Now it's possible he has his jacket on just from the viewer's point of view so that we wouldn't know he was dead. But that could be a stretch, and might not be true.
I'll give you that one. It's been a long time since I have seen the movie. I remembered the blood appearing when he realized he was dead, in the same way the table in front of the basement door appeared with a voiceover "they only see what they want to see."
But don't ghost make people cold? I thought Cole was always cold around Malcolm. I should rewatch it.
Cole also knew that he wouldn’t see Malcolm again after telling him how to talk to his wife.
That could also be just in that he know that a therapist would need to move on if his work was done. Or maybe he thought possibly he was a ghost, and possibly he just needed to move on.
Would have to think he does. The kid's intro scene I believe is his mother standing up and walking out as... Man... comes in and starts asking questions. The implication is that she's explained to her son that Man will be his counselor, but the twist shows that no she didn't, so she just left Boy there without a word about Man. That's got to trigger the "she can't see him" response in Boy.
I am flabbergasted that anyone could possibly disagree with my obviously correct interpretation of the movie.
Why is it necessary that we know one way or the other? I like the ambiguity of the whole scenario. The fact that we have to keep poking at that question makes the movie extra interesting, doesn't it?
For instance, in Blade Runner, we'll never know if Rick Deckard was a replicant with false memories of being a human. That made the premise of Blade Runner more interesting. Heck, life is more interesting if you don't have everything spelled out for you. ;-)
I don't think there was any indication he was a replicant until Ridley Scott added in some footage from his next movie.
The question of whether Deckard was a replicant goes way back even before the director's cut was released. I don't know about the footage from that Blade Runner followup movie (I didn't see it). In fact, I remember discussing it in pre-Internet days. Harrison Ford went on record though that he urged Scott not portray Deckard a replicant. Scott seems to have kept it ambiguous enough that it didn't piss off Ford. But now, Ford has changed his story...
https://screencrush.com/harrison-ford-says-deckard-is-replicant/#:~:text=But%20in%20a%20new%20interview,Warner%20Bros.
Very hard to explain how come Gaff's entire last speech is already weighed down with double meaning if Deckard isn't meant to be a replicant.
And don't forget, the original/theatrical ending is also footage from another movie – B-reel from The Shining – that Scott didn't want there, with the voiceover Scott also didn't want there. He had both forced on him by executives after focus groups (Blade Runner was one of the first focus-tested films, supposedly) found the plot too hard to follow.
It's been a while since I last saw the film, what about Gaff's speech indicates Deckard is a replicant?
He says, literally, "You've done a man's job, sir. I guess you're through, huh? It's too bad she won't live! But then again, who does?"
https://www.youtube.com/watch?v=83tboDnpE7Y
Every bit of this is a double entendre alluding to Deckard being a replicant, even if you overlook the nature of Gaff's presence throughout the film (that is, he's retrospectively-obviously Deckard's human handler, the actual Blade Runner, if you like).
Deckard is retired at the beginning of the movie. All Gaff did was bring him to Bryant, who unretired him to get rid of four replicants, all of which are dead at the end of the film when Gaff says that. That does not make Gaff "Deckard's human handler, the actual Blade Runner". There was no need for Deckard to have been retired if he were a replicant. If you were going to make a replicant into an effective blade runner you would not give them memories of having already quit the job.
Yes, that's true. It's kind of hilarious, Scott stole footage of a unicorn from another one of his movies, and the footage quality doesn't match Blade Runner at all. And everyone that on the production said that Scott just made that up afterward, and there was supposed to be no indication that Deckerd might be a replicant
I remember that damn unicorn! When I watched the director's cut, it wasn't there. I thought I was a false memory on my part. But it seems like it worked with unicorn origami that Gaff left as his calling card.
Deckard being a replicant doesn't really add anything to the movie. It's a twist for the sake of having a twist.
Well, de gustibus non disputandum est. But I never understood that smarmy smile that Bryant gave Deckard as he gave him the assignment to kill Zhora, Leon, Pris, and Roy. When a friend said, "Oh, Deckard was a replicant, too," it all fell into place for me. And then the movie became totally Dicksian (not to be confused with Dickensian).
He's not a replicant in PKD's "Do Androids Dream of Electric Sheep?".
I'd say it takes away from the movie if he were a replicant. A big point of the movie is that the replicants are more human than the humans are. Rutger Hauer feels and has compassion and fears death and worries about his legacy, whereas Harrison Ford is living a crappy almost robotic life.
Indeed. He also seems like a crappy replicant since his job is to take out replicants and he's physically weaker than them.
Rachel is also weaker than the other replicants. Presumably the mnemonic replicants all have to be human-spec at least physically, otherwise they'll start noticing the cracks in their reality (why can I lift four hundred pounds when nobody else can? I don't even lift!).
But he really wasn't weaker than them. He was able to kill them all except for Rachel killing Leon.
No.
I recently visited Poland (in my first real vacation of my PhD), and wrote about my observations: https://open.substack.com/pub/denovo/p/notes-from-poland
It's quite an interesting country and I hope you enjoy reading about it.
Fun review thanks! Although I think the idea that Warsaw is the only modern city (when you did not even visit Gdansk or Lodz) is a little underdeveloped.
Enjoyed your post, Poland is an underrated country for tourism IMHO, check out Gdansk next time!
Yes, that was the one city I wanted to visit but didn't have time for.
Solidarność!
In English, Z is 10 and W is 4.
Here is how the Philadelphia Science Fiction Society-related book club votes on the next books to read.
There are usually about a dozen people at a meeting. At the end of the meeting, people nominate books, and say a few sentences about their book. A list is made (maybe a dozen books), and people can vote for as many books as they like.
Frequently, there's a break point where some books are clearly getting more votes. Maybe 6 books are doing better than the rest.
In the second round, people get three votes to distribute. If there isn't a clear winner, and there usually isn't, people get one vote for the last round.
If there's a tie, those books will be read in subsequent months.
People sometimes campaign for months, even off and on for years, to get a book chosen.
Books that made a decent showing will be nominees in the next month.
This style of voting is easier online, since lists of nominees can be put in chat.
Is there a list online of past books you've read. Since you recommended T. Kingfisher's "A wizards guide to defensive baking". I've been reading everything she has written... and I'm almost finished, more good authors please.
There is no list.
The book club recently discussed China Mieville's _Un Lun Dun_ on my recommendation, and most people liked it. I'm not describing it unless people ask since when I first read it, for some reason I decided to read it without knowing anything about it (not even reading the back cover), and that might be the best way to read it.
I'm currently rereading the Saga graphic novel and enjoying it a lot. Science fiction in the fine old tradition of just making things up. Sex, violence, pacifism in a time of war, spaceships, ghosts, psychic powers, old relationships coming back, disguises, confrontations. If it's the kind of thing you like, there's plenty of it.
I think T. Kingfisher's _Thornhedge_ was spectacular, but has more horror than _A Wizard's Guide_.
> I guarantee I will never respond to your comments urging you to call a phone number in South Carolina.
I am pretty worried that enough people took this at face value for you to need to point this out. Are ACX readers not impersonation-scam-savvy? I received that same "text me" comment and my immediate reaction was to scoff at how transparent it was.
I found it suspicious given there was no additional context to it, so I googled the phone number and saw it was used by a completely different handle on a different blog, and put "let Scott know his account might be compromised" on my todo list (it was still sitting there until this Open Thread, I didn't manage to get around to it, have now deleted it).
I very responsibly emailed Scott’s substack email letting him know of the spam and now feel like I should get a merit badge
Well semi naive person here, I responded to the email. But it bounced so I stopped and thought, probably spam.
I got an answer, but it was generic enough that I didn't reply to it.
I doubt anyone actually fell for this. Probably just an abundance of caution on Scott's part.
I fell for it.
At least one person, judging by the comments from Highlights From The Comments On Fetishes. Although, said person is 72, so I suppose I should cut him some slack.
> I guarantee I will never respond to your comments urging you to call a phone number in South Carolina.
So that road trip we set up from LA to Vegas to cover The Mint 400 motorcycle race is off I guess. :(
I should have known something was up when you agreed to play Dr Gonzo and let me be Raoul Duke so quickly.
A Good Morning
Out by the pond a couple minutes after sunrise a rainbow segment to the south.
Later, a young muskrat perfectly still near shore, half out of the water(contemplating life?) and a Green Heron a foot to one side picking breakfast from among a school of minnows. Peaceably coexisting, each with a perceived abundance and lack of territorial response.
This morning I was meditating(daydreaming!) on a large rock in an area recently landscaped by volunteers(with HOA funds). One of the women made a point of being in my line of sight when I opened my eyes. She had been nearby the two previous days. We exchanged pleasantries and I told her what a good job she had done as she walked off to her garden plot. That same young muskrat(same size) from yesterday was swimming a few feet off the shore's edge, which was about 20ft away. He caught a fish, chewed and swallowed; paused still looking my way; loitered a bit, then swam off.
"It is the nature of every grain of corn to become wheat and every precious metal to become gold and all procreation to lead to the procreation of the human race. Therefore a master says that there is no animal which does not possess some likeness to human beings.”
--Eckhart(1260-1328), Meister Eckhart Selected Writings, 1995
* Human beings are social to the core. *
[Edit/Add: "there is a great difference between being in a room that is empty and being in a room where someone is sleeping. A sleeping person is a presence in a room. They are aware of you, even if it is only a dim, vague awareness.”
--Patrick Rothfuss, The Wise Man's Fear, 2011]
Good content. It has the form of a haiku but I was unable to compress it sufficiently, even with the unnecessarily (in English) verbose 5/7/5.
Rainbow over morning pond
Muskrat and heron in stillness sit
Abundance and peace.
: ) I enjoyed reading that.
I did too.
( of course the minnows probably had a different view... )
> I’ve told Substack about the problem and they say they’ve taken care of it - but if it keeps happening, let me know.
This happened to me 4 days ago (8/31).
Yup same date here.
ditto
Scott, please know that this number is still actively being used to scam. I asked a generic question(from an online account) and just received a response.
Same, from "Scott A[capital I]exander".
Yeah, I just checked, mine was that too.
Ah the classic Ender's Game trick. Still have to watch out for "Scott Alexander "
There's been a lot of discussion about matchmaking ('brokers') but not a lot of interest in exploring the concept further.
Well I'm interested in being a 'broker', in a non-Traditional-cultural background sort of way.
Correct me if I'm using the concept wrong but for long-term relationships there seems to be a fundamental non-ergodicity in play. That is, one may successfully ('happily') marry once or maybe twice in a lifetime but no one could do so hundreds of times. So it's not possible to skill up at finding marriages the way one *can* level up at short term dating. The best an individual can do is bet on proxy activities like micro-marriages, etc.
Now one individual cannot marry many times for the experience, but a third party, i.e. a broker could in theory successful go through the process of matching-making tens, hundreds, even thousands of times in career/lifetime. They can observe the successful pairing of many different couples rather trying observe a single person successfully marching many times. So it would seems useful For Research Purposes to spur the creation of a class of marriage brokers, in the hopes of creating many marriage-making experts.
Having said all that would it be a crazy idea to read through every dating doc I can find and then message potential couples to match them?
I'd suggest you look into the actual business of matchmaking first (and yes, it is a business). Firstly, the idea of just randomly messaging people is weird. What you usually do is you engage clients who pay you to find matches. Most matchmakers specialize in one gender. They also have a proprietary binder of people who they've pre-evaluated. Part of the pitch to both sides is that the other side will be high quality, serious about marriage (or whatever), and not flake. This is certainly a model that could work in SV and especially in some of the weirder niche communities. But I'd suggest you think about it more before just kind of jumping in.
Maybe not hundreds, but dozens. https://www.reuters.com/world/india/man-with-39-wives-head-worlds-largest-family-dies-india-2021-06-14/
I think it wouldn't add much value because I expect the majority of dating doc list people have already looked or at least skimmed all the compatible dating docs.
It's an O(nm) problem where n is the number of people and m is the length of the doc.
I would expect people to quickly become disillusioned as n and m increase and the amount of irrelevant (to their own interests) information increases.
So you think can get the information on the successful matches and then train pattern recognition software that can identify good pairings between dating docs?
The number of men seeking women vastly exceeds the number of women seeking men.
This is not true. In fact it's the opposite. The number of men seeking casual sex vastly exceeds the number of women seeking casual sex. But the number of women seeking serious relationships vastly outnumbers the number of eligible bachelors.
I'm sorry. I should clarify. My prior statement is not correct. The number of men seeking women *doing this dating docs thing* vastly exceeds the number of women seeking men.
Ah, yes. This is normal though. It sounds like your meat and potatoes will be recruiting women who are willing to try out nerdy rationalist types and then setting them up on dates. Plus some dating coaching in all likelihood.
I suspect "eligible" is doing most of the heavy lifting here. According to a 2020 Pew poll on dating and relationships, equal numbers (31% each) of adult men and adult women are single, single women are substantially less likely than single men (38% vs 61%) to be looking for dating and/or relationships, and among those who are single and looking, and roughly equal percentages (78% of men and 76% of women) of those who are single and looking are looking at least in part for serious relationships. Single women are more likely (36% vs 22%) to be exclusively looking for serious relationships, though, but in absolute numbers that works out to 4.24% of all adult women and 4.16% of all adult men looking exclusively for serious relationships and an additional 5.65% of all adult women and 10.6% of all adult men simultaneously looking for serious relationships or casual dates.
As for "eligible", single women who report been having trouble finding people to date are a lot more likely than the corresponding subset of single men (56% vs 35%) to cite "Hard to find someone who meets their expectations" as a major reason it's hard to date.
Summary: if you're a single woman looking for men to date, the odds are good but the goods are odd.
Full survey results: https://www.pewresearch.org/social-trends/wp-content/uploads/sites/3/2020/08/PSDT_08.20.20.dating-relationships.full_.report.pdf
This statement is a good example of statistics being technically true but misleading. If you limit it to under 40s or never marrieds then the number of people looking for a relationship evens out. And once it evens out the 50% increase in women looking for serious relationships means they outnumber men. Likewise, older women are significantly more likely to report difficulty dating.
Further, women (regardless of whether they're on the dating market) are both more likely to report it's difficult to find someone worth dating and that it's hard to find people looking for a committed relationship.
You're at least partially right. I was looking at the top-line numbers and didn't look closely at the age bracket cross-tabs. Among those who are single, the under-40s are roughly equally likely to be looking. Looking at other cross-tabs, I suspect a lot of the older women who are single and not looking are widows.
However, the percentage single in younger age brackets is very unequal. Especially in the 18-29 age bracket, where 51% of men but only 32% of women are single. 30-49 also has more single men than women (27% and 19% respectively), while 50-64 is pretty close to equal (27% and 29%). It's only in the 65+ demographic that many more women than men are single (49% vs 21%), consistent with my "confirmed widows" hypothesis.
Take these two effects together, and I think my conclusion that more men than women are both single and looking for serious relationships probably still holds for younger age brackets.
Excellent context, thank you.
I'd like to insert a "X solves this" or "there's no alpha left in X" joke but I'm coming up blank. Perhaps something about changing incentives for polygyny to even out the "can't find" ratio.
A trusted broker could act as a centralized entity which receives submissions from women who don't want to post online.
I could see that. I have my doubts there's some huge pool of hidden women dying to marry these nerd guys (a group in which I include myself, mind you), but I wish them well. It's like the lady trying to beat ovarian cancer; you don't think she'll win, but you wish her luck anyway. Life is cruel.
It doesn't have to be that they are dying to date nerd men. They just have to have a strong enough dislike for the mainstream dating market. I believe that sort of hidden pool does exist.
If you message me a potential match for mine that actually leads to something I will happily pay you for it.
Since this an open thread, might as well announce it here,
I am creating a newsletter (biweekly comic) aimed at making medical concepts accessible for everyone through storytelling.
The goal is empower people by satisfying their curiosity on topics like how fever reducers work, or the science behind cholesterol.
I geek out on the research and then break it down in conversational ways people can relate to.
Let me know if any of you would be interested.
I'll send you the first draft.
Cheers!
Sure.
You'd be better off posting a Google form or something to collect people, as I don't want to post my email publically.
You can just subscribe to me here!
"There’s a scam where an account pretending to be me is replying to comments here and then immediately deleting the replies; people are getting “replied to your comment” emails that suggest calling a number in South Carolina."
You know I had to do it 😁
The Scott Alexanders of South Carolina:
(1) "Scott Alexander is a published author, a coach to entrepreneurs and executives, an accomplished speaker, and a strategy/leadership consultant."
(2) Scott Alexander, Maintenance and Engineering Manager
(3) "Oconee Station State Historic Site is home to one very special ranger, Interpretive Ranger Scott Alexander"
(4) Scott Alexander, Pharmacist
And about 74 others, none of whom I imagine are our mysterious emailer!
Ironically, he is a published author, Silicon Valley CEOs do read his blog, and I imagine he's spoken at meetups...
So is this his doppelganger or are we not getting the full Scott Alexander story? By day, a psychiatrist in San Francisco, every other Thursday an accomplished speaker and business coach in South Carolina! 😁
I've noticed that conventional schooling can cause despair in people who aren't good at it. People can come out of it thinking they're incapable of learning a subject if they didn't learn it in school, or that they're bad at learning in general. Is there any way to handle school so that people reliably feeling that they might just need a different teacher or method of study?
One way would be to offer learning opportunities in a variety of fields. That could be adding vocational subjects, hands-on environments, etc. Someone who is bad at classroom learning may succeed more in a garage learning to be a mechanic (or a lab technician, or an accountant). We would never be able to cover every subject that way, but if we offered a fair variety, it may help people learn about and seek their potential outside of classrooms where they struggle.
There are clearly additional ways to help people find their potential, but the question is whether any such attempt could scale. Parent involvement matters a tremendous amount here, but not every parent is interested, capable, or available. I doubt any type of larger scale schooling could tackle the challenge. Small numbers of people could get a private tutor or go to a specialized school, but that's going to be very limited. Large environments hit a wall trying to customize too much and the costs skyrocket.
I think our starting assumptions should be that
- Most people who struggle to learn in school are actually bad at learning (which includes being genuinely low in intelligence), especially if stuff like dyslexia is identified and controlled for
- The number of people who experience a chronic struggle to learn because of a teacher or method of study is very low
- Even if some alternative form of study or instruction has the potential to work for some people, these people are likely to not have very successful careers in cognitively demanding fields, because the employment environment (i.e. 'the real world') doesn't and cannot make accomodations like this (because its not a controlled environment with crafted training material - you're dealing with the economic system at large) and you need to be able to succeed with whatever is in front of you, and most very successful people are able to adapt and thrive under non-optimal conditions
These are unlikely to be universally true, but I think this is more correct on average than most alternatives.
The above is especially true for the very common liberal trope of the student who struggles with subjects they find boring but supposedly does well when the material is more 'fun'. Even if this actually happens (and they aren't just mistaking enthusiasm for actual academic comprehension), again, smart successful people work hard and learn things things regardless of how boring they are, so we should assume both that these 'bored' students are both genuinely less intelligent, and that they will struggle with any cognitively demanding job, because the real world is rarely 'fun'.
I think learning styles exist, or at least it's easier for me to absorb what I read than what I hear, though it helps with listening if I have something to do with my eyes and my hands.
Studies have repeatedly failed to find evidence for the traditional 'learning styles' (kinesthetic, auditory, visual etc.)
Oh, I get it. You restate my position using inflammatory language, that way you get to make it look bad and dismiss it out of hand. I guess that's much easier than having to make an actual argument.
The heritability of intelligence is as much as scientific fact as countless other things you strongly believe in. Denying this is not scientifically defensible.
Why would you needlessly change his wording, even though you put it in quotes, in a way that made it significantly more inflammatory than it was? "Bad at learning" was his language, which you exchanged for "hopelessly retarded." Feel free to disagree with him, but your approach added nothing to the discussion but heat.
Sorry, do you simply not think its possible even in principle that behavioral differences between populations can be largely heritable in nature? Or are we supposed to pretend that this can't be the case because you find it personally offensive?
You people act so damn intellectually smug despite the fact that you repeatedly demonstrate your scientific ignorance and cannot beging to provide coherent alternative explanations for these things that aren't either wildly dogmatic or at odds with empirical evidence (like thinking either poverty or inequality explain black people's insane homicide rate - they can't).
And what exactly does "called out" mean? I've had people like you get offended that anyone would base their view of the world on scientific data rather than egalitarian ideology, but that does nothing to change the actual underlying reality.
If he's willing to use milder language and not be inflammatory, we should engage with his ideas as presented. The same way many of us do (or should) respond to your posts despite previous inflammatory language you've used.
I think that generally, people have accurate judgments about their ability to learn. Yes, impostor syndrome is a thing, as is unfounded hubris, but generally there's an extremely strong correlation between perceived ability to learn X and actual ability to learn X, especially if the student has already spent a bit of time trying to learn X.
It’s quite common for kindergartners to think reading and arithmetic are beyond their abilities, quite uncommon for this to actually be so.
Right, I wouldn't trust kindergartners to have good judgment. But Nancy Levobitz was talking about people who come out of school, i.e. adults.
So at what point do we start trusting them? We are talking about changes to the school system, so presumably by the time they come out of it as adults it is too late...
Gradually as they age, with early adults having full freedom to self-select?
Isn’t this what we do already? Kindergartners have no choices, high schoolers have electives, college students have full freedom to choose.
I think that this is wrong for math in America. I have tons of students who are utterly convinced that they are awful and incapable of math, when really they are not.
Probably foreign language learning as well. Schools are a really bad way to learn languages.
You're asking if it is possible to do badly in school but still come out feeling confident about your ability to learn, I think?
What about another scenario: doing great in school and realizing you're really only good at gaming that system? That you have no curiosity or ability to learn for the joy of it?
I've seen examples of both.
Good point.
Also, sometimes people give up on particular subjects rather than their general ability to learn.
This is very common (your last point). U.S high school computer science teachers are, in many cases, so bad that they turn kids away from this lucrative field.
There are two separate questions here: (1) is there a different `one size fits all' solution which would work for a larger fraction of people? (I presume there is none which would work for everyone), and (2) if we customized a schooling solution to the individual, could everyone learn [x] (with the exception possibly of people with severe handicaps or zero motivation)?
On (1), we could start by looking at different public school systems that actually exist in the world, and pick the one which works best according to our lights. On (2)...it seems likely that customization will work for a larger number of people than one-size-fits-all, but it will inevitably also be a lot more expensive.
> (1), we could start by looking at different public school systems that actually exist in the world, and pick the one which works best according to our lights.
Trying to pick out actual causal effects of schooling system differences themselves is nigh on impossible, because you face the hopeless task of trying to account for inherent differences in student ability, and these differences almost certainly dominate in explaining any academoic variance over any other factor.
Obviously between countries you're dealing with the substantial population-level differences in intelligence, but this effect exists even within countries. In the US for example, people will point to charter schools with XYZ teaching style as evidence that that style works better, but the problem is the student makeup of these schools is highly non-representative. They are disproportionately in certain locations (states, citites neighborhoods etc) and even though they're usually not academically selective they disproportionately attract students of parents who are likely to be higher in intelligence, conscientiousness etc. The idea that we can get any meaningfully valid insights from strictly observational education studies is wildly optimistic to the point of ignorance.
If you want education insights, you *need* controlled trials using intellectually matched samples of students. I would bet every cent I own that you could take a group of sufficiently high IQ students and put them in any education system that exists around the world and they'll still beat the pants off the average student - if you can't control for this, if you can't control for this, you can't gain any useful information about the effect (if any) of differences in education systems.
Okay, that’s a fair point. But realistically we can’t do controlled studies either. Unless you’re secretly dictator of some place anyway. So what does that leave?
I think that a good first step would be to decouple teaching from grading. The thing we have now (the one who teaches is also the one who assigns grades) is in my opinion a horrible conflict of interest. It is okay for teachers to give *feedback* to their students. But they should not be the ultimate judges of which of their students get *certified*.
So, in my vision, there is a school where you go to learn, and then there is an "exam room" where you go to get examined and get grades. You can practice the exam room online, but to get your actual grade, you need to actually go there. Oh, and if you fail, you can try again. That should remove a lot of stress.
If this became normal, I suppose it would allow for greater freedom of schools, because the fact that your students pass an independent exam should be a strong argument that your system works okay.
> if we customized a schooling solution to the individual, could everyone learn [x]
In my vision, you could take any exam at any age. Are you 30 years old and want to do the test of addition of one-digit numbers? Sure, go ahead. I mean, if you think about it, why not.
This, I suppose, could increase the overall success at learning, because I suspect that in many cases the problem is simply the slower students having not enough time to master some topic, or simply not caring enough about the topic (which can change in a few years). Though, it would not be literally everyone, because some people would still not care, or would get too busy with life outside the school.
A related criticism, in reverse, is that when the teachers have to prepare students for an outside examination, they "teach to the test" at the expense of better learning for the students. Whatever the test asks, is what gets the emphasis. This happens now with standardized testing. It would likely happen much more if all grades were from an external examination. If the student would benefit from learning something outside of the tested material, you've greatly increased the chances it would just be skipped. There's a high probability that this would happen, as not every learning area can be tested (and some are very hard to test, like art and music), and there isn't enough time to test even on what could be taught now.
I have heard about the problems with "teaching to the test", but I am not convinced that they couldn't be solved by designing *better* tests. I am probably missing something, but sometimes the solution seems kinda obvious.
Problem: Teachers only teach the parts that are included in the test.
Solution: Include *all* parts. (Would it make the test too long? Then use a random selection in the test, a different random selection every day.)
Problem: Teachers don't teach anything besides the mandatory curriculum.
Solution: Include optional questions that allow you to gain *bonus* points (and potentially more than 100% on the test).
Problem: There is no reason for gifted students to learn anything beyond the test, if they already have 100%.
Solution: Make special "advanced" tests for the gifted students; or at least allow them to take tests for higher grades. On the diploma, list the advanced tests they took successfully.
Problem: The tests are too hard.
Solution: Change the scale so that solving e.g. 50% of the problems counts as "success". (Or maybe, have "easy" and "hard" problems, and require 80% of the easy ones, and 20% of the hard ones.)
Also, in context of American education, I believe that testing was just a part of the problem with "No Child Left Behind". A greater part of the problem was that schools were evaluated (if i understand it correctly) by the test scores of the least successful children. That is, to make a simple example with three kids in the classroom, an outcome [10, 100, 100] was considered worse than an outcome [11, 20, 20] by the official metric. So the teachers optimized for the metric by drilling the least successful child to the test, and ignored everything else. But this was the problem of the metric rather than of the tests themselves. You could have the same tests, and instead measure the maximum or the average, and that would give teachers completely different incentives.
I don't disagree with you, other than to say that it's very difficult to make tests that meet all of your criteria for the wide range of students that would need to be tested. You are correct about some of the testing problems in the US and how teachers try to get around it.
Nobody is trying to make bad tests. In fact, significant resources are put into making very good tests. The problem continues in a way to suggest that it may not be possible to make a test perfect, or even good enough to avoid the issues. And that's in subjects which we think we can readily test, such as Math and English. In subjects that are harder to test, or where "correct" is more subjective, I'm not even sure it's something we could really effectively try. Even in Math there's getting the right answer or knowing how to do the operation - two separate things that you would test differently and score differently. And that's one of the easiest to score of all subjects.
I agree 100% and have been thinking about various formulations in this vein for a while. The any-age thing is particularly important, I think. Multiple, arbitrary learning paths can be offered by any number of outside bodies. It's the exam's business to ensure that you know what you should do, and as long as you do, who cares what path you took to get there?
I can see that exams might have to become a bit more rigourous, to deal with this greater variability. But at the same time they can maybe do that by becoming more drawn out affairs. I have imagined the "exam" being an app that watches your work in a certain field, or effectively an averaged high score of your performance in a computer game that analogises perfectly with the subject domain.
I also believe that the exam should be a mirror of the conditions of the real life application of the subject. So, the exam for bomb diffusing should be a high stakes, stressful nightmare of an exam. The exam for talking about English literature, less so.
One question I keep coming back to is how could the examining body be set up and brought into being? It has to happen unilaterally because there's no question of waiting for any existing state-controlled education system to do it.
I did think that, the value of university degrees having fallen a lot, and (to my knowledge) nothing else having arisen to replace it - here's the opening. All a new accreditation system has to do is be a better guarantee of quality than existing degrees.
I worry how the profit incentive would conflict with the intended goals. Like, if you set up the exam system, you cannot give everyone the perfect score. But everyone who gets a less than perfect score has a selfish incentive to support your competitor instead, if they give them a higher score. And maybe the employers would prefer to use the system that is more tough on students (other things being equal), but they will ultimately choose the one that has much more students. Also, only the university exams or trade school exams are relevant for employers, but I would also like to fix the elementary and high schools.
To certain degree this already happens at schools. For example, when you have students rate their teachers, then (other things being equal) teachers who give better grades get better ratings. Similarly, when students choose a school, many of them prefer the one with a smaller chance of dropping out, so the schools often keep students beyond the point where it would be reasonable to kick them out. Also, one of the reasons the tests are generally unpopular, is that many popular kids fare worse in the tests than they would at a more personal exam (where it is easier to apply charisma or bribes).
For starters, I would limit the organization to testing school subjects (i.e. no bomb defusing) and even there those that can be tested by a computer (i.e. no gym or art exams). Otherwise it wouldn't scale well. But there would be a human in the classroom to make sure that the students are not cheating. (The online anti-cheating systems are either too weak, or have too many false positives.)
Maybe in the first iteration, it could be a service offered to homeschoolers or kids in various alternative schools, to offer a feedback to parents how their children are progressing compared to the traditional education. Or maybe a better first iteration would be to provide entrance exams for high schools and universities. And from there, I would try to expand to test everything. (Which is not the same as if I optimized for profit, because then I would stick with the profitable parts only.)
I'll admit, the profit motive is not something I'd considered. But I don't see that it's necessarily a perverse force. As you say, employers will favour exams that are tough, employees will favour exams that are easy - but employers are the ones making the decision. I think the incentives align to make the exams tougher not easier.
Even in the maximally inconvenient case that there's already a large competitor offering easy exams to get lots of students through the system: the rational move for the employer is to prefer certs from the tough nut exam whenever he sees them, and grudgingly accept the easy students if no hardcore ones are available. So the tough nut body grows in any instance.
I'm more in favour of exams getting tougher than easier, but really my ideal is that the exams are purely *objective* - they somehow accurately grade skill in a way that doesn't change at all over time. You should be able to compare yourself to any other person and have confidence that the only confounders in the score are ones which also confound your ability itself. I have no idea what this looks like - it definitely won't be a single scalar value anymore - but that's the target I want to be thinking about.
I also think about education at the primary/secondary school level, as well as topics that come earlier, later, or are absent entirely. The reason I focus on replacing university qualifications is because that's a starting point with a clear market need, which if exploited properly could gain funds and recognition that would make it easier to turn around and address the other stuff.
My idea of a maximally inconvenient case is more like this:
Your competitor offers exams that are somewhere in the middle -- not too easy, but not too hard either. They can be gamed by teaching to the test, but most people don't care, because they consider learning to the test an evidence of intelligence and conscientiousness, which are desirable qualities for a future employee. Also, students get extra points for attending the right kind of school (not directly, of course, but more like "if they attend the right kind of school, their teacher is allowed to assign a part of their score based on their school work"), and again most people don't mind because attending the right kind of school sounds like a good thing so it's okay if it is rewarded. Etc.
Your competitor will be more popular than you. Kids from the right kind of school have an obvious incentive to prefer the system that gives them an advantage. Employers want the system where the kids from the right kind of school are. All other students want the system that the employers want, and also where they are associated with the kids from the right kind of school. (Kinda like "I am on Facebook, because everyone is on Facebook; it doesn't matter whether a different website has a better user interface".)
In other words, a system that takes over your niche, but also sucks, but doesn't suck enough to make people leave it, so they stay there for the network effect.
(This may sound too pessimistic, but look at the current educational system and all its disadvantages and all its unfairness... and yet, most employers prefer students who pass the system, and most students are willing to get into huge debt in order to participate in the system, because it seems like they cannot avoid it. If something like this happened once, it may also happen the second time.)
> because I suspect that in many cases the problem is simply the slower students having not enough time to master some topic
These students are unlikely to ever 'master' these topics - they may eventually get to the point of being able to pass an exam and little more. And it's unclear on why these students even need to be taught anything beyond the most rudimentary literacy and numeracy abilities - they will never be capable of holding a remotely cognitively demaning job and it feels like much of the motivation here is the warm fuzzy feelings of seeing the strugglers 'succeed' in education and not feel like hopless idiots. But as far as real practical outcomes go, it makes very little difference (even in one in 10,000 of these students somehow turns out to be genuinely intelligent and capable).
Passing the exam is a binary choice, so the question is where to draw the line.
Traditional education puts the line quite low, simply because you want the entire class to move forward at the same speed. So if someone has a 50% success rate, you say "okay, that's good enough, let's move to the next lesson". Of course, if someone is 50% good at lesson one, you can hardly expect them to do much better at lesson two, if they are so bad at the fundamentals.
A better system would move at individual speed (perhaps because it would use a lot of computer education), so you could put the bar higher and require 90% success rate to pass. With the consequence that many would pass later, and some not at all. However, this would be slightly balanced by the fact that people who passed the lesson one later would be better prepared for the lesson two than they are today.
See also: https://en.wikipedia.org/wiki/Mastery_learning
And yes, some people would only get the rudimentary literacy, but I think that even they might be better served by a system that spends 100% time on teaching them the literacy, than by a system that tells them a lot but ultimately teaches them nothing.
It feels like you are looking at students as in one of two categories - capable and incapable. Or at the least taking a certain cutoff point and saying anybody less than that is just dead weight. Neither of those things is true. There are lots of people who are not intellectually gifted but can consistently and effectively learn more than rudimentary levels of various subjects.
A while back Scott posted a chart on various professions with the IQ range of people who worked in them. While Scientist and College Professor both had higher ranges than some other positions, the ranges dipped below 100 for each - meaning people performing the jobs with measured IQs at or below 100. I can anticipate your response that these people must not be performing the job well - but neither you nor I can know that with information currently available and it would be illegitimate for you to make that argument based only on your preconceived intuition.
>There are lots of people who are not intellectually gifted but can consistently and effectively learn more than rudimentary levels of various subjects.
Im not talking about "not gifted". Below a certain level of intelligence, people will struggle to learn anything beyond the basics (and in some cases even that is a struggle). And missing the odd idiosyncratic smart person does less harm than wasting years of people's lives trying to get them to learn something they cannot (and even worse, sending them to college).
>I can anticipate your response that these people must not be performing the job well - but neither you nor I can know that with information currently available and it would be illegitimate for you to make that argument based only on your preconceived intuition.
No, actually, we know that job performance is correlated with IQ. It's not intutition, it's empirical fact.
https://pubmed.ncbi.nlm.nih.gov/14717634/
I mean, why the heck do you think anyone cares about IQ at all? It's not because there's some profound significance of shading the correct circles on an IQ test - it's because of what it tells us about the real world. And intelligence predicts job performance better than variables like work sample tests, reference checks, years of experience, and age.
The most likely explanation for "professors" with an IQ below 100 is affirmative action/HBCUs (and maybe community colleges having positions called "professor" or something silly).
How much more would you pay to see a doctor of IQ 130 rather than one below IQ 100? I would pay many times more ad I suspect you would to, even though we supposedly "can't know" which is better.
"The thing we have now (the one who teaches is also the one who assigns grades) is in my opinion a horrible conflict of interest."
That's a very good point.
Even setting aside whether it improves learning, it might make the resulting grades more generally useful (to the student, to employers, to evaluating teaching methods, ...) and believable.
We have this in the UK. Exams are written and graded by independent national exam boards. And you can do them as an adult if you really want (I think you have to pay an admin fee).
The American system seems like a bad idea to me. You could be unfairly penalised (with lifelong knock-on consequences) if your teacher hates you, or if your teacher is a bit clueless and doesn't understand the subject they're teaching (I've had teachers like that, but at least they weren't grading my exams).
Is teacher-grading of everything really "the American system?" Surely it's different in every state? And the SAT, at least, is externally graded right?
Meanwhile, when you say "exams" in the UK, are they 100% of the grade for each class? And I'm assuming they're only in the final year of high school? Or multiple years?
Unless everyone is already familiar with the intricacies of different systems, I really think these kinds of clarifications are necessary for a debate like this to make any sense.
I have bad memories of my Grade 8 "Language Arts" (what they called "English" at the time) teacher, who got mad at me for writing that Animal Farm was a metaphor for the Russian revolution.
I tried to point out the Lenin, Stalin, and Trotsky figures, and the proletariat and the capitalists, to no avail.
In that case, it would have been good to have a better-informed 3rd-party do the marking.
(She also got mad at me for not knowing about "The Mod Squad", a TV program I'd never seen.)
Yikes, that s_x. Sounds like she had an agenda.
Two hypothetical one size fits all solutions is to build students' respect for their won capacity to learn. This may be difficult in an environment with a lot of emphasis on grades, but I recommend exploring it. Maybe tell stories of people learning in different ways and at different times. Maybe encourage students to think about their own experiences of learning.
Is customization-- maybe just not expecting students to keep trying at things they're very bad at-- very expensive? This question might deserve some thought.
Do you have any evidence that the variance in academic ability is meaningfully explained by differences in how they learn things (especially in a way that could be applied systematically to public education e.g. not having all students indvidually tutored all day)?
The old view of different 'learning styles' (e.g. people being 'auditory learners' or 'kinesthetic learners') specifically has resoundingly failed to be empirically verified, so whatever it is you think leads to these learning differences will necessarily have to be very different to these types of 'learning styles'.
I'm not sure you're addressing her primary point. There are myriad reasons a student who is otherwise intellectually capable could be frustrated in some particular subject and lose interest. That this has been used sometimes in other contexts to whitewash the fact that not all students are equally capable is not relevant. Any one of any intellectual capacity is better off at least attempting to be a lifelong learner than not attempting this. So discouraging people even inadvertently from whatever curiosity and capacity for growth they are capable of is a mistake that we can presume is preventable to some extent.
>I'm not sure you're addressing her primary point. There are myriad reasons a student who is otherwise intellectually capable could be frustrated in some particular subject and lose interest.
Sure, but do you have any estimates for what proportion of observed intellectual ability variance results from this?
>So discouraging people even inadvertently from whatever curiosity and capacity for growth they are capable of is a mistake that we can presume is preventable to some extent.
Sure, but this is very different to what we're really talking about here, which is systematically changing education for the sake of people who are allegedly struggling to learning style difficulties.
Yes, and precisely my thrust is that that's not the point she's making at all and you were just talking about a different topic.
I was taught sentence diagramming back in the sixties. It was a tidy little thing that I wasn't bad at and I didn't mind it, but it seemed to be of no use. Since I had very little drive, I didn't care.
Recently, I've had a couple of friends say that they consider sentence diagramming to be important, and I've thought about what it might be good for. Maybe it's useful if you want to write a complex sentence. Maybe it's useful for sorting out other people's complex sentences.
So, were you taught sentence diagramming? Do you think it was useful, and if so, for what?
I do think it's important for people to know how to write clearly, and I don't know how it can be taught. What do you think helps?
I've posted this on Facebook, and gotten a hundred or so accounts of life with and without without sentence diagramming.
https://www.facebook.com/nancy.lebovitz/posts/pfbid0uMamDsboTEHb1aR6Xsrqy4zMen61FomwYH6N3nmCnaSaTYMgp1D1qpvby7JTQs5ul
There are a good many people who found sentence diagramming to help with learning languages or clarifying writing, and a good many who have become professional writers without having learned sentence diagramming.
This raises a question for teaching. *Perhaps* all students should get some exposure to sentence diagramming, but how do you identify the students who won't get anything out of it relatively quickly?
I'm assuming that students' time is valuable.
Diagramming sentences felt completely redundant to me, I think because I'd already learned a bit of Latin by then. Possibly any foreign language would work as well?
When I was about 11-12 (around 1970), an external teacher came into our Christian Brothers school to give us a few lessons on parsing (I don't know what sentence diagramming is, but I assume they are related). I found it quite eye-opening.
Diagramming forces the student to break sentence structure down to its components, so it's not only useful, but essential to understanding language. If one does a transcript of the spoken word in mass media, one can see how popular disjointed, irrational and fallacious language constructs have become. Just diagram the sentence.
Except that diagramming seems to be routinely taught only the US. Do you want to argue that people from other countries are incompetent in their own languages?
I also don't know whether a background in diagraming causes people to make their speaking more organized. It might, but you're the first person who's brought up how people speak, a large topic in itself.
In my facebook discussion, there seems to be a higher proportion of professional writers and editors who didn't learn diagraming. Maybe I'm missing something, but their writing seemed as clear as the writing from people who did have a background in diagramming.
To be clear, for those of us who didn’t learn it at school (it seems to be a very US thing, and of a particular time) are you talking about what Wikipedia calls the Reed-Kellogg system? As a linguistics student I learned Chomsky’s X-bar method, which I occasionally find useful but wouldn’t normally use in reading or writing prose.
Sorry, didn’t see this earlier: I am referring to the syntax tree diagrams mentioned upthread. I’m not sure I agree that there’s not a convenient system to teach these, though: they seem to be a pretty standard part of most syntax curricula.
One thing to be aware of is that there are many specific things that all fall under the umbrella of “sentence diagramming”—I hated the specific instance I did in high school English class, but loved the specific instance I did in college as part of my linguistics minor—and differences in opinion may be because different people interacted with different versions of sentence diagramming.
Some people just need to hear that a 1-D string of tokens can also have some sort of hierarchical, tree-like structure. For others, this advice is so self-evident and banal, it's hard to articulate. Trees can be useful to interpret garden-path sentences, but it's probably easier to appreciate when dealing with larger structures. E.g. using an octree for collision-detection.
If I were a teacher, I would just share a quick, shallow lesson with pathological examples using a broad variety of representations, and then move on.
I was taught sentence diagramming (5th or 6th grade, IIRC), but it seems like useless busywork to me.
I *loved* sentence diagramming. It was one of the few times in school outside of math classes where I really had the feeling of learning something about the structure of things, in this case the structure of language. Once it was explained it made sense, but it had never occurred to me til we did that unit that there was a logic to how sentences were built. Learning about the double helix in science class thrilled and fascinated me the same way, and so did learning how to turn word problems into equations (you built this little machine and turned the crank and out came the answer like a sausage.) After we did the unit on sentence diagramming we did one on figures of speech, and also on some related categories — idioms, slang and dialect — and I lovde that too. I was as thrilled as I would be now if somebody gave me a brand new high-end car.
I do think sentence diagramming was useful to me. I enjoy long complex sentences more — things like the sentences in late Henry James novels — because understanding how they’re put together gives me an extra dimension to enjoy. Some of them are like cathedrals. And I think sentence diagramming has helped me write clearly. There are lots of little errors, things like changing tenses in mid-sentence, or tacking on a clause in the wrong place so that technically it applies to something other than what you mean it to, that do not really confuse the reader, but give what you wrote a sort of blurriness. The reader has to push through this little thicket of things you don’t mean in order to get to what you do mean, so they arrive at your meaning a bit tired and irritated.
One of my happiest intellectual moments in school was finding out about polar coordinates. It was very satisfying to know there were two such different ways of describing locations.
The best relatively recent novels I can think of for long complex sentences are _Jonathan Strange & Mr. Norrell_ and (less recent) _The Worm Ouroboros_.
I did not learn sentence diagramming in school. My mother (MA in English) made me learn around 4th or 5th grade. I did not understand why, until I got to high school and grammar classes started getting a bit more rigorous. Sentence diagramming taught me a) basic grammar terms and concepts and b) a way to translate internal sentence structures into visuals.
The second of these was incredibly helpful once I started taking German classes.
I then went on to teach writing as an student instructional aide in college, and discovered that sentence diagramming was the actual reason behind what to a lot of people looked like my "superpower"--I could glance at an essay prompt and tell you exactly what needed to be written, while others would consistently not answer the question being asked in the prompt. Apparently it is very easy to miss entire words or phrases in a long prompt, and sentence diagramming seems to teach you to stop and see every word? That's my best guess as to what happened there.
Thank you. I hope this is a specific enough example to satisfy people who are looking for examples.
I hadn't thought about sentence diagramming as a cure for low-rez reading, but I see how it could help.
I learned a bit of it. I didn’t mind spending a little time on it but I don’t feel I learned much; it didn’t enhance my understanding of grammar, which was not hard for me. Maybe there are some people for whom it makes ”getting” grammar easier. But I think diagramming is something that can easily become pedantic busy work.
It certainly can be something "to keep the kids busy". I was surprised at how many people found it was helpful.
Someone noted that people who are good a physics are more likely to be good at diagramming sentences, though I don't know whether they're more likely to find d.s. useful.
It think it makes a bit more sense when you start learning foreign languages (because then you use more of explicit reasoning), but still not very useful, unless you happen to be curious.
(To answer your original question, yes I was taught sentence diagramming at school.)
Grammar is at least a first cousin of mathematics.
Yes, I think the same! In fact my answer here compared learning sentence diagramming to learning to construct equations from word problems. I enjoyed the daylights out of both.
I learned sentence diagramming. I think that firstly, it helped give me a more concrete idea of how grammar worked, which was helpful both for understanding other language's grammar (I could compare it to English) as well as for my own writing (deciding between who or whom depends on identifying if you're working with a direct object or not, for example). Also, as a math teacher, I find it helpful for deconstructing word problems; I don't go into super-deep detail with my students, but it's still helpful to be able to raise points about nouns usually being your variables, and connecting phrases like "is a function of."
I learned it and have not found it useful in the slightest.
Maybe possibly it helps someone? I struggle to imagine how though.
I mentioned a couple of uses. If you want to find out about more, you could read the facebook thread, or at least some of it.
You said "here are some situations in whch it could be helpful." The missing piece is describing how senfence diagramming is helpful, along with a concrete scenario, including application and before/after comparison.
The facebook link seems to require a facebook account.
The problem is that a lot of people say that sentence diagramming improved their thinking about writing, so they're not going to have examples.
How much of a problem (effort? security?) would it be to get a facebook account?
As I understand it, there's no conveniently worked out system for teaching syntax tree diagrams.
I was just having a discussion about this on Discord, but has anyone tried deliberately practicing mental visualization for an extended period of time to try and get better? I suspect being able to visualize things in your mind is a skill which can be trained. Do any children development programs deliberately try to practice this skill?
Imagine a ball and keep duplicating it until you're no longer able to create more balls. Personally when I do this there appears to be a point at which my brain starts to lag. Furthermore, despite my best attempts, it feels like past a certain amount my brain is perfectly willing to pretend that I'm still generating large amounts of balls but it just avoids filling in all the details. So despite my best effort to continue generating balls, my imagination just straight up lies to me, which seemms like an interesting observation.
I was going to ask about visualizing a larger field-- my default is a smallish area in front of my eyes, and then I realized I can visualize a bigger field/more locations just by trying. The images aren't brighter, sharper, or more complex, but it feels good to have access to greater variety. Even a very fuzzy landscape all around me is nice, and the pleasantness might be an incentive to work on improving other features.
I did a course with AphantasiaMeow(who has rebranded since but I’m sure search will find anyway) - a few separate hours. It worked in that I got some visual experiences I hadn’t had before. And I was left with exercises that im sure will improve my imagery if I was motivated to do them. The basic technique is to describe verbally visual details a lot, with various exercises around that.
I’ve also started trying to capture my dreams by describing them and I think they’re maybe getting more visual as a result.
Should have said, I’m hypophantasic. I haven’t seen any good evidence of courses for already quite phantasic people. I think it’s a great idea!
Maybe not exactly what you're talking about, but yes: https://sashachapin.substack.com/p/i-cured-my-aphantasia-with-a-low
This sounds a lot like Tantric meditation practices that they practice in Vajrayana Buddhism.
In a recent post Scott wondered about parents “who think it’s old-fashioned to get pink things for their daughter”.
It’s generally reported that pink for girls and blue for boys was a new post war trend, and the opposite applied pre war. Is there much or any evidence that this was widespread at all, either way?
All I can see for Victorian society is that young children wore white. After that there’s no obvious trend, not amongst the kind of people who got family portraits (ie paintings) done - but these would the only kind who would care. The children of the poor wore whatever was handed down.
I have some dim memory that red was seen as a vigorous, violent color, the color of anger and blood and war. While blue was seen as a calmer, cooler, more refined and sedate color. So I can see how those might manifest with typical gender roles.
Possibly the fading (to pink and light blue) is something specific to children?
During the First World War, Scottish regiments wore their kilts into battle, as was their ancestral wont.
When the Germans deployed mustard gas, the Scots were uniquely and painfully most affected.
So when WW2 rolled around, the top brass insisted that if Scotsmen wanted to continue wearing their kilts, they must also wear a pair of specially designed mustard-gas-proof chemical lined pants.
These pants (and for burger-eating readers, this word means "underpants") were a very ugly bright pink, and the Scots refused to wear them.
So now Scottish soldiers wear trousers like everyone else.
I expect this satisfactorily answers your question.
There's very little evidence for this - it's one of those things that get repeated so much that everyone just assumes it's true without being able to point to any genuine evidence to support it (much like the myth that white women have been the main benificiaries of affirmative action).
This of course does not imply that there's anything inherently feminine about pink. But as long as pink is associated with femininity I think it's healthy to buy pink things for your daughters, and that 'genderless parenting' will raise a generation of kids even more messed up in the head than zoomers are.
This is it – the key takeaway from understanding that the colors are arbitrarily chosen is that the specific colors *don't matter* and that what matters is *establishing a sex-differentiating color schema* for small children without much sexual dimorphism and then *universally adhering to that schema* so that all people can readily identify the child's sex and treat it accordingly, socializing it into its sex properly and helping to avoid weird nonconformity scenarios down the line.
Removing the artificial dimorphism is destructive and wrongheaded, basically a life-hating behavior bent on thwarting a deep human instinct.
It is a good thing to feel like you fit in and belong. Sex roles are a significant part of that system, and, often, all bucking the system gets people is lonely and depressed. I personally no longer support extreme individualism, such as rejecting gender norms. If a young woman wants to buck the trend and behave in aberrant ways that aren't hurting anyone, great! We have individual freedoms so she's allowed to do that. We just need to make sure she knows and understands what trends she's bucking in the first place so she won't be confused and lost if it ends up making her miserable. There are lots of perfectly happy and healthy LUGs* and similar, because they, very importantly, _knew how_ to regress to the average after college.
*Lesbian Until Graduation.
"It's okay for a woman to behave in gender-non-conforming ways, so long as everyone around her constantly tells her how weird and non-conforming she is" is not the grand gesture of acceptance you think it is.
Like, I don't *think* you meant to justify bullying kids who don't act manly enough for their peers, but that's what your argument does.
No, it isn't. A simpleton could use my argument to justify that, but he'd be wrong to do so, because bullying is bad. I absolutely reject the idea that one shouldn't make an argument, regardless of how true it is, based on the fact that any given bad actor could use it for bad reasons. I've seen that reprehensible line of reasoning rear its ugly head too many times recently.
You feel entitled to far too much acceptance.
Societies need structure to function, and you seem unaware (or uncaring) of how much our present day mores are interfering with that.
If the choice is between the dissolution of the structures that maintain and sustain society, or reverting to a higher level of penalisation for nonconformity, the rational choice is to bring back bullying.
Being made to feel weird while nevertheless being left to go about your business in a classically liberal society is an incredibly generous deal on society's part. You will not be offered that same deal by any society that arises to replace ours if it fails.
If you are a weird and nonconforming person, your priority should be trying to shore up and perpetuate the uniquely accommodating world you live in, making whatever compromises you have to along the way.
You should not be asking for better and better treatment while the whole edifice groans underneath you.
Depends what you mean by "that argument." Some of what I said doesn't directly apply to religion or genetics or sub-culture, so I don't know what you're asking.
You mean, is it a good thing for Hasidic Jews to wear characteristic garb which sets them apart from the multitudes? Almost certainly yes, this doubtless improves their group cohesion and identity even though it's actually completely arbitrary (the costume is just what happened to be fashionable clothing in 1830s Odessa or something IIRC, when-/wherever it was the founder of Hasidism lived), and it doesn't hurt anybody else.
If fewer people are socialized to be in clear sex-based groups, then they are more likely to be confused and feel out of place about their sex. This would naturally lead to more people feeling dysmorphic about their sex. This feeling is associated with very high levels of mental health problems and suicide. We should very much work very hard to help as many people as possible to not have this problem.
In my experience, it's more common that this happens when children socialized in clear sex-based groups encounter groups who are not socialized in clear sex-based groups (or vice versa). I.E. Sending children to all-girls or all-boys schools won't help them avoid dysphoria if they encounter examples of gender bending outside of the school. However, by the same token, raising children without strong gender roles seems to have no bearing on them thinking of themselves as boys or girls- instead, it is when they encounter strong sexually dimorphic social groups that they start to shy away from identifying as their birth sex.
I'd love to see some studies done on this to confirm personal anecdote, though I think in modern times the topic is too highly charged for it to get IRB approval. Nevertheless, societies with strong gender roles inculcated into children (Japan is an example) do not seem to be seeing significantly lower rates of transgender adults than societies with less strongly-inculcated gender roles and similar laws around what counts as transgender (Finland is an example). Both hover around 1-2%. What do you think could explain this?
I don't think we have a clear enough definition of the concepts to actually measure it. I don't trust that the most interested parties can be neutral enough to properly classify whether someone is actually trans or if there's another explanation/issue involved. We're beginning to have a common agreement that the approach in the last few years has resulted in significant increases in the number of people identifying as trans, and most people see that as a problem to be fixed. Other than that, we don't really know that much.
I wrote a reply to this and then something seemed to go wrong with its placement in thread, so I deleted it for avoidance of confusion; I think you probably got it through the arcane Substack messaging system anyway, but to reiterate just in case: Yes, in this and many other cases it is good. Your failure to understand this, and why, is indicative of a broader failure to understand humans and positive socialization. The artificial markers are only arbitrary in their per-se nature, their purpose is not arbitrary. They are designed as support struts for the underlying objective reality of the two separate groups, making it easier and more effective to acculturate small persons to their future, post-pubertal social roles. Our failure to do this consistently, as a society, is the underlying reason for lots of children's alienation from and fear of puberty. As late as the '60s or so children *looked forward to* it, to becoming adults.
I like the way you put things and would find it useful to read more of what you write.
Thank you! Unfortunately I find myself fairly inept at expressing myself in the long-form essay mode; moreover I don't have a high opinion of the clarity of my own writing. For instance, I should clearly have expressed it as "the markers are artificial, but the groups aren't".
Oh, what the hell? I wrote a reply, then it posted in two copies so I deleted one, but both got killed. I'm sorry, I seem to be suffering from some sort of reply-based curse. If you got a copy in your activity feed or whatever, feel free to repost it as a reply to yourself and attribute it to me. I can't be bothered to retype it.
Aristotle pointed out that it is good to learn optimal preferences.
Buying clothes for someone is not overriding anything. If the person has no clothes except the ones you buy for her, and she hates them for one specific reason she has made clear, i.e. color/theme, then it would be overriding oppression to keep buying that color/theme.
Yes it is. It literally is.
How did you reach that conclusion?
This is a common story but doesn't have much evidence. It's one of those facts that seems to have been laundered into common knowledge by being too good to really examine. The citation in Wikipedia is The Atlantic. Often what's pointed to is a 1927 article saying young boys should be dressed in pink. However, as has been noted: even at the time pink clothes were primarily purchased for girls.
Pink was considered unattractive in the Middle Ages because they preferred bright, garish colors and pink was too pale. However, it does appear men wore it without feeling it was feminine. Pink probably became female coded in the 17th/18th century where it became fashionable among French court women. And expanded from there. In particular you still see men in the early 18th century wearing pink from time to time but by the middle or late period this goes away.
(Also, keep in mind that prior to the late 19th century boys wore women's clothes until the age of 5 or so. So a boy wearing pink doesn't necessarily mean it wasn't feminine.)
The fundamental point that the association is arbitrary is sound though. There are other cultures where pink is not traditionally associated with femininity. Of course, we now live in a globalized world so the idea has spread. Likewise the idea of red denoting hostile or bad is not universal. And white being positive or meaning purity. And so on.
I'm not certain that the early modern world had much of a concept of pink as distinct from light red.
The pigment for magenta, which is needed for bright pink, was invented by chemists in 1856, so before then people were limited to light earthy reds or expensive true reds. Purple was royal because it was so difficult and expensive to produce. Edit: it's possible to get soft pinks from things like berries, but I haven't heard of any evidence that they've been considered particularly feminine.
In medieval iconography Mary often wears blue, or sometimes red over blue, and Jesus wears blue over red. Possibly the feminine blue thing came from Mary in blue?
I never really believed that these associations could reverse over night, and I was expecting something like that: Bright red for boys, light blue for girls.
I see no evidence either way though.
Here's an article on Gainsborough's late 18th century portraits, starting with the famous Blue Boy - but other male subjects are also painted wearing pink.
So pink and blue were male colours, or at least colours for boys, in the 18th century.
https://www.apollo-magazine.com/gainsborough-blue-boy-grand-manner-van-dyck/
As an aside, when I was a child, our family doctor had large reproductions of the Blue Boy and Pink Boy hanging in the waiting room, so I knew the paintings before ever I knew who painted them or what the subjects were 😀
Another way to say pre-war, is to say pre-modern science. Much of chemistry came to knowledge after the war. Even the chemistry of hydrocarbons wasn't really discovered until the 1920s, when a wealthy automobile owner asked his chemist friend why the gasoline in his fuel tank evaporated so quickly.
It was in the 40s when we discovered how to manipulate carbon chains into polymers.
Well, what counts as "old-fashioned"? Pink for girls and blue for boys was standard until at least recent decades.
I have the impression that people used to dress babies and toddlers of both sexes pretty similarly until the boys were old enough to wear pants. (So baby boys in old pictures are often seen in ”dresses” and assumed by people nowadays to be girls.)
That's why breeching (putting boys into trousers) was such an important rite of passage, and why going from short trousers (for boys) to long trousers (for young men) was another stepping-stone to adulthood.
And there was supposed to be an Irish superstition about the fairies stealing children away, but they preferentially stole boys, so that is why boy children were put in 'skirts' like girls, to fool the fairies (until the children were past the first vulnerable years of infant mortality).
Thanks, that's interesting.
You also have to remember, before the war—read before modern manufacturing—clothing was fantastically expensive. Baby clothes were made from discarded clothing.
My Grandmother was born in 1920. My mother (born in 1940) once asked grandma how she could live with such a small bedroom closet. Grandma replied "I only had two dresses, an every day dress, and a Sunday dress."
True, but it applies also to very rich people like British royalty.
Dressing is one of those things modern people do paradoxically very badly even though/because clothing is cheap compared to before; most people wear sort of shapeless clothing.
Some online articles claim that pastel colours became popular in the 19th century, and that makes sense to me regarding the discovery of aniline dyes. Now you could get synthetic dyes that produced a range of colours and were dye-fast when washing clothing.
https://fashionhistory.fitnyc.edu/aniline-dyes/
“The technology of dyeing fabrics was transformed in the mid-1850s when the British chemist William Perkin (1838-1907) discovered that dyes could be extracted from coal tar. These new aniline dyes became very fashionable. The first was ‘Perkin’s mauve’, followed by a variety of shades of purples and magentas, yellows, blues, and pinks. These colours were much more intense than any available from the traditional natural dyes.”
https://www.smithsonianmag.com/arts-culture/when-did-girls-start-wearing-pink-1370097/
"Pink and blue arrived, along with other pastels, as colors for babies in the mid-19th century, yet the two colors were not promoted as gender signifiers until just before World War I—and even then, it took time for popular culture to sort things out.
For example, a June 1918 article from the trade publication Earnshaw's Infants' Department said, “The generally accepted rule is pink for the boys, and blue for the girls. The reason is that pink, being a more decided and stronger color, is more suitable for the boy, while blue, which is more delicate and dainty, is prettier for the girl.” Other sources said blue was flattering for blonds, pink for brunettes; or blue was for blue-eyed babies, pink for brown-eyed babies, according to Paoletti.
In 1927, Time magazine printed a chart showing sex-appropriate colors for girls and boys according to leading U.S. stores. In Boston, Filene’s told parents to dress boys in pink. So did Best & Co. in New York City, Halle’s in Cleveland and Marshall Field in Chicago."
And then, for some reason, it switched to 'pink for girls, blue for boys":
"Today’s color dictate wasn’t established until the 1940s, as a result of Americans’ preferences as interpreted by manufacturers and retailers. “It could have gone the other way,” Paoletti says."
But *why* did the preferences go that way is the interesting question. According to Wikipedia, there wasn't a reversal; pink/blue for girls/boys and boys/girls existed at the same time, and it was the pink for girls/blue for boys that won out:
"Since at least the 19th century, the colours pink and blue have been used to indicate gender, particularly for babies and young children. The current tradition in the United States (and an unknown number of other countries) is "pink for girls, blue for boys".[1]
Prior to 1940, two conflicting traditions coexisted in the U.S., the current tradition, and its opposite, i.e., "blue for girls, pink for boys". This was noted by Paoletti (1987, 1997, 2012).
Since the 1980s, Paoletti's research has been misinterpreted and has evolved into an urban legend: that there was a full reversal in 1940, prior to which the only tradition observed was the opposite of the current one. Quoting the concluding lines of this study: "In conclusion, there are strong reasons to doubt the validity of the standard PBR [pink-blue reversal] account; if anything, gender-color associations seem to be much more stable than currently believed".
There's also a long list of historical sources backing up that "pink for girls, blue for boys" was an established standard from the 19th century on.
So probably an urban myth, as suggested?
I was thinking that the hideously sad Victorian era poem "Little Boy Blue" shed some light on this issue. Maybe blue *was* a boy's color in 1888? So just looked poem up in Wiki and discovered this info gem: "Little Boy Blue" is a poem by Eugene Field about the death of a child, a sentimental but beloved theme in 19th-century poetry. Contrary to popular belief, the poem is not about the death of Field's son, who died several years after its publication. Field once admitted that the words "Little Boy Blue" occurred to him when he needed a rhyme for the seventh line in the first stanza."
So really not relevant to pink/blue boy/girl question. But that poem has made me feel like shit every time I've run across it, and somehow knowing that "little boy Blue" is the product of a rhyme search reduces the impact of the damn thing a bit.
The only "Little Boy Blue" verse I know is the nursery rhyme:
Little Boy Blue, come blow your horn
The sheep's in the meadow, the cow's in the corn,
Where is the boy that looks after the sheep?
He's under a haystack, fast asleep
The Victorians were morbid and maudlin about death, I must look up that poem!
Ugh, good luck. That damn poem has a sort of double toxicity for me, because I've never gotten over feeling sorry for discarded toys -- some holdover from childhood where I really believed my stuffed animals had thoughts and feelings. So dead child plus dusty, neglected but still loyal toys is a one-two punch to the heart for me.
“ For example, a June 1918 article from the trade publication Earnshaw's Infants' Department said, “The generally accepted rule is pink for the boys, and blue for the girls. The reason is that pink, being a more decided and stronger color, is more suitable for the boy, while blue, which is more delicate and dainty, is prettier for the girl.” ”
Yeh, I found that in my searches as well. I am wondering though exactly how common it was. There’s two references about pink for guys - that and Time, and most literature seems to acknowledge that both existed.
However how common was the system at all? Is that it, just those articles? Was it just the readers of Earnshaw's gazette?
Finding family portraits for Victorians is easy, just google Victorian family paintings. And the Victorians definitely didn’t colour code children, everything was white, as dye ran.
By the post Edwardian age family portraits are more likely to be photographs and thus not in colour, so it’s harder to tell.
"And the Victorians definitely didn’t colour code children, everything was white, as dye ran."
That's why I mentioned the aniline dyes, which didn't fade and were colour-fast. Now they could have pastel (and brightly coloured clothes) even for children. It's difficult to tell, because the Wikipedia article does contradict the idea of "it started off pink for boys then flipped in the 1940s". It seems like pink and blue were colours for both boys and girls, but pink for girls/blue for boys was more traditional and eventually won out.
Marco Del Giudice has studied this issue: https://iris.unito.it/bitstream/2318/1853049/1/DelGiudice_2017_pink_blue_update_pre.pdf The results do not support the idea that "pink for boys, blue for girls" was ever an "accepted rule". In particular, in English-language books from the period (circa 1880-1940), "blue for boys, pink for girls" was the rule. On the other hand, in newspapers the two options were about equally frequent.
In the article there are some quotes from newspapers from the early 20th century teaching people which is supposedly right (and it goes both ways), which suggests there wasn’t an accepted rule.
> And the Victorians definitely didn’t colour code children, everything was white, as dye ran.
White is not a convenience choice. It's quite expensive to produce, and although it doesn't run, it shows stains really easily.
"White is not a convenience choice. It's quite expensive to produce, and although it doesn't run, it shows stains really easily."
Tell me you live in the days after bluing had ceased being a thing, without telling me you live in the days after bluing had ceased being a thing 😁
Laundry bluing was used for the same reason washing detergents nowadays include "optical brighteners" in their lists of ingredients: it makes the whites look whiter.
https://en.wikipedia.org/wiki/Bluing_(fabric)
"White fabrics acquire a slight color cast after use (usually grey or yellow). Since blue and yellow are complementary colors in the subtractive color model of color perception, adding a trace of blue color to the slightly off-white color of these fabrics makes them appear whiter. Laundry detergents may also use fluorescing agents to similar effect. Many white fabrics are blued during manufacturing. Bluing is not permanent and rinses out over time leaving dingy or yellowed whites. A commercial bluing product allows the consumer to add the bluing back into the fabric to restore whiteness."
Also, if you ever get fine, sunny days, you leave whites out on the bushes, hedges or lawns (bleaching greens) to help whiten them by the aid of the sun, as mentioned in this video about a Victorian country house:
https://www.youtube.com/watch?v=2LXqVXl6dVY&list=PLx2QMoA1Th9deXXbo7htq21CUPqEPPGuc&index=34
Isn’t white the base undyed color of cotton, linen and most wool? Also easier to wash because it doesn’t fade, as Deiseach pointed out?
Depends. Most cotton and wool that we grow now is white, though natural colors for both exist. Wool will also often end up creamy or off-white, so not the pure, bright white you expect nowadays.
Linen is a bit more complex, as I understand it. You can get either yellow or grey linen depending on how you ret it, (dew/ground retting gets you silver/grey, pond/water retting gets you yellow/gold) but then you can bleach it in sunlight like Deiseach mentioned.
Undyed linen is grey.
On the other hand you can boil wash whites and bleach them. Doing that with coloured clothing for small children, which will need to be washed frequently, means the colours fade (and run, as anyone who ever mistakenly put a red item in with the whites finds out) so it's not worth it *until* you get synthetic dyes which can stand up to that treatment of frequent washing.
And linens and towels in hotels (at least the kind of hotels I can afford) are still white because of the ease of washing.
thanks for that. a very fun read.
WRT to the fake dating site bios that got so much attention in an earlier post: I think it's interesting to observe that all the love given to the fourth one could have been predicted by the dead-simple heuristic "the most popular female dating bio will be the one with an explicit sexual come-on in it".
She was written as a woman who liked all the obscure male-typical interests of people on this blog. Scott himself said "I tried not to write the rationalist as 'obviously the best one', but I couldn't pull it off."
The only other example was the anime fan, but as actual anime fans pointed out her tastes didn't match an anime fan's (they were all the most popular without any realistic specialization by genre--it would be like someone saying 'I like Shakespeare, Jane Austen, and Toni Morrison' with regard to literature).
I think it's not just the sexual come-on, that was a specific instance of her seeming like the one who'd be the most enthusiastic about a relationship. Everyone else seemed pretty insulated.
I hadn't even considered this looking at all those polls but it's a interesting point. I feel as though the fourth one would still be the most popular if the sexual come-on was moved to a different profile but I'd be curious to see how the polls changed.
Yeah, all I remember about the fourth one is that it was attractive because it was a nerd. I'd expect sexual references to generally be seen as a negative when looking for a long-term partner.
Men are simple creatures and are easily moved by even indirect sexual references. Advertisers take advantage of this a lot, but older examples (prostitution, strip clubs, porn) abound.
Imagine that you have the superpower to make your ears glow a vivid, rather ugly, wholly unmistakable shade of green. While using this power, you are incapable of lying, or even of saying things you are not confident are true, and everyone else knows this.
How do you use it?
I would become a famous philosopher, specialising in epistemology. Having an experimental test for whether I am personally confident something is true won't resolve the whole field, but it will definitely breathe new life into it, and people will come from all over the world to map out the limits of Melvin-approved truth.
How bright is the glow and how long is the windup-cooldown? Seems like the best use is to just strobe-light people.
How confident is confident? I can say a whole lot of things off the top of my head with perfect confidence that fades within seconds of thinking about it. Does the strobe-eye stop these or not? I generally don't rule out conspiracy theories unless I have firsthand proof they're false; do the eyes stop me expressing any viewpoint?
(And what IS truth? Is truth a changing law? https://www.youtube.com/watch?v=dN3j-oPQ2-k)
I'm not all that confident about the truth of most things. So I'd say nothing... or just mundane things.
I think there's a minor family in the Godfather which makes a living by brokering peace deals between the major players. You could put them out of business: any family whose opponents propose anyone other than Green Ear Man as broker invites the question, what they want to lie about.
The Bocchicchios clan. They were used as mediators because they were vendetta fanatics. Even by Mafia standards.
Why is this desirable? I don't understand, but I'm interested.
Here is the basic idea. Say that Michael Corleone, the head of one Mafia family wishes to meet with Don Tessio, the head of another in order to discuss a deal for mutual advantage. The invited guest has no way of knowing if he will be safe during the visit, and Michael's promises that he will not hurt the guest cannot be believed. There is a problem of commitment here, and without some commitment, the two will not meet.
Enter the Bocchicchio Family. When Michael invites Don Tessio, he not only promises not to harm him, but also hires a member of the Bocchicchio Family to go to Tessio's house. There, the "hostage" will be guarded by Tessio's men. If Don Tessio does not return safely, Tessio's men will kill the hostage. The Bocchicchio Family, seeking revenge, will blame Michael Corleone for the death, since he made the promise that Don Tessio will not be harmed.
https://www.mikeshor.com/courses/gametheory/docs/topic6/godfather.html
That position isn't as strong; there's nothing stopping people from lying _to_ Green Ear Man.
True, but it eliminates a lot of risk. The Corleones might make the Bocchicchios (thanks) such a stupendous cash offer that they decide to give up their reputation and defect. Secondly you can ask GEM "so, when the Corleones outlined this deal to you did you get the feeling they were doing so in good faith?" You don't have to be perfect, just better than the competition.
Same with dealing cocaine: I have not adulterated this and have no reason to think anyone else has is not watertight, but better than the competition.
There are probably a lot of products that genuinely save people more money than the cost of the product over the long term, but which don't sell very well because "pay us now and you'll end up with more money later" just isn't a very believable claim in most cases. If you started a business selling those products, you might be able to bring products to market that no competitor could find customers for- and the selling point of "this is practically free money" would be pretty compelling.
You might also be able to solve the "market for lemons" problem- selling higher quality products for higher prices without sacrificing demand. Also, selling endorsements to other companies could be pretty incredibly lucrative if you were famous enough.
I become some sort of trusted intermediary / arbitrator / referee.
You know, there's a way in which banks facilitate international commerce between parties that don't trust each other. For this, they get to extract some percentage of the deal. This is a stable system, as the bank is basically selling "honesty" as its product and raking in cash for doing essentially nothing, so it is incentivized to behave honestly.
But I'm sure there are many areas where this kind of long-term, game theoretically founded trust isn't as easy to construct. If you want to leverage your power as much as possible, you will need to let other people do the work, but in a way that you are essential to making the deal work. It's lazy, it's profitable, and it's even pro-social.
Run for office. My superpower will make me a celebrity, and I can unmistakably commit to trying to do good things and avoiding special interests. Zelenksyy became president of Ukraine on the basis of being famous from a TV show where he was a noncorrupt person, operating as proof of concept that people care about mild fame and noncorruptness enough for it to take you to the top.
Of course, then I would have to actually govern. I don't know if this would be a disaster (can't use small lies to grease the wheels) or actually go okay (I can tell people I'm not plotting against them and just doing things for the good of all, and they will believe me).
This strategy wouldn't work for me because as soon as an interviewer asked "do you believe christianity is dumb" and I refused to activate the power, the game would be up. I suspect interviewers would be able to prove that you have specific beliefs which make you unelectable.
Honestly, dishonesty and creativity with the truth is a major part of statecraft. There's the 'Mr. Smith Goes to Washington' myth in this country of the honest, incorruptible person who goes in and frees the country from the special interest, but it only occurs in movies--probably the closest analogy was Jimmy Carter, and look how he wound up.
Zelenskyy did a Trump-like outsider campaign where everyone was so sick of the corrupt people in charge that he was able to win, but he had a long career in showbiz before that and his ability to act has been quite useful in getting countries to support him in what's a really horrible situation.
The most successful US presidents--FDR, Eisenhower, Reagan, Clinton--all knew how to play dirty and be loose with the truth when necessary. It's about coalition-building and being inspiring, not doing good things and avoiding special interests. The Dems need woke party activists to get out the vote just like the GOP needs all those MAGA people who think Biden drinks children's adrenal glands. You need votes in Congress and you need the SCOTUS or state and local governments not to overturn everything you do. I would suggest you pick one issue to focus on that you care about (AI regulation?) and sell out on the rest of them.
That was the only possible alternative to just hiding the "superpower" to all. But I do not think it would work: you will have to detail your actual plan for office and all your true interests/character traits. I do not think you will be able to compete with people able to lie, even with the benefit of actually being able to prove your are not lying. Except maybe if you happen to be superhumanly moral, altruist and have socially accepted interests/kinks.
Else you will be food for the cat the first public debate you are in.
The loophole in the case as stated is that "While using this power, you are incapable ...even of saying things you are not confident are true"
So one of your advisors tells you a pack of lies, but you believe it to be true ("Saddam Hussein has weapons of mass destruction pointed at London and ready to go within five minutes!") and you make your ears glow green, go on national television to tell the public this is why you are going to invade Iraq, and everyone believes it has to be done because President Green Ears said it was so, and like George Washington, he cannot tell a lie.
If you want to manipulate people using this power, don't be Green Ears Man. Be the person telling President Green Ears, or telling the people who tell President Green Ears, "cross my heart and hope to die, Trump hired hookers to pee on a bed in Moscow".
A powerful analogy I have recently discovered makes me think little good would come of your superpower as POTUS. Also remember your own maxim: True, Kind, Necessary. If you can't lie, there are bound to be statements which you either can't make or would violate the "2 out of 3" condition.
https://www.imdb.com/title/tt0443272/quotes/?item=qt2354870&ref_=ext_shr_lnk
https://www.youtube.com/watch?v=nPfihz9-Ls0
Thaddeus Stevens: The people elected me to represent them, to lead them, and I lead. You ought to try it.
Abraham Lincoln: I admire your zeal, Mr. Stevens, and I have tried to profit from the example of it. But if I'd listened to you, I'd have declared every slave free the minute the first shell struck Fort Sumter. Then the border states would've gone over to the Confederacy, the war would've been lost and the Union along with it, and instead of abolishing slavery, as we hope to do in two weeks, we'd be watching helpless as infants as it spread from the American South into South America.
Thaddeus Stevens: Oh, how you have longed to say that to me. You claim you trust them, but you know what the people are. You know that the inner compass that should direct the soul toward justice has ossified in white men and women, North and South, unto utter uselessness through tolerating the evil of slavery. White people cannot bear the thought of sharing this country's infinite abundance with Negroes.
Abraham Lincoln: A compass, I learned when I was surveying, it'll... it'll point you true north from where you're standing, but it's got no advice about the swamps, deserts and chasms that you'll encounter along the way. If in pursuit of your destination, you plunge ahead heedless of obstacles, and achieve nothing more than to sink in a swamp... what's the use of knowing true north?
Pretty much. Politics is one of the worst places for a totally honest and principled person.
Yes, though it's worth remembering that you also need the compass.
I guess. He did end slavery in the end after all. I mean, we botched Reconstruction, but it was an important step forward.
The reconstruction results were probably not that much worse than could reasonably be expected. It's not like nation building efforts in Iraq or Afghanistan were resounding successes.
Also, you can hardly blame him for not maintaining the quality of his governance after he was assassinated.
But once you reveal this power it will be used against you:
Imagine you enter an interview with a savvy and agressive journalist. He may demand that you turn your ears green before answering certain questions.
If you refuse you automatically seem extremely untrustworthy.
If you agree, you may say things that are controversial and you may damage your popularity and dash your electoral hopes. There is a good reason politicians love and tend to lie and avoid questions.
My gambit to deal with that is to regularly refuse to answer questions, saying (with green ears on if need be) "I don't want to answer that question, not necessarily because the answer is embarrassing but because at some point someone is going to ask me something to which the answer is embarrassing and I don't want it to be obvious when that happens".
Not perfect, but I think a lot of people would understand.
I wonder if you could get round that by making your ears glow green then saying you won't answer that question because it's private and nobody's business but your own/you genuinely believe that is not in the public interest, it is just prurience/you should not answer that question because it would breach confidentiality.
Of course, a muck-raking journalist is going to find ten ways to hang you whatever you say.
But by saying that you won't answer the question because it's private you are lying.
You are not answering the question because the consequences of answering it will be negative for you, and that's the only thing you can say to avoid a question like that in green ears mode
I don't see how it's lying to say "I don't want to answer that question, it's none of your business" as to what age did you lose your virginity/what are your kinks/did you ever think about committing arson/other nosey-parker prurience.
Sometimes you do have to be willing to accept investigation (e.g. being vetted for jobs which involve background checks) but a random journalist fishing for some juicy gossip can go jump in the lake.
Lots of people sincerely believe that what serves their personal and tribal interests is good for everybody so the glowing ears don't prove that much. I'm more interested if a politician chooses a smart policy than if he firmly believes in what he says. Now you are unusually self aware so it would be harder for you to lie to yourself, but that's still no guarantee that you would be an effective political leader. Though, I would vote for you to become a DA or the Comptroller General of the United States or in another similar position where probity is key and less open to interpretation.
What you say might be right, but I think very few people believe it.
I don't use it, and don't tell anyone I can do that. I don't see how this could bring me more than a little bit money and minor celebrity. But it will basically makes me unable to lie for any serious purpose, once people know about this "superpower". They will just assume I lie if I refuse to glow.
Frankly, apart if I am lost and on the verge of dying, and ear glowing is my only way to let the rescue team find me, nobody will ever know about this capability.
More a superweakness than a superpower....
Seems like the downside is that once people realize you have this power, it becomes more difficult to get away with a lie. “Prove it! Turn your ears green and repeat what you just said!”
A meta-honesty policy can help with that concern. Just like encrypting all your communication makes it hard to know when you're actually talking about secret stuff. You just categorically refuse to make your ears glow x% of the time or whatever.
I don't know what the optimal way to implement something like this is, but I'm certain it's not a total loss of your ability to lie.
That won't work. If you're not glowing your ears, you're obviously lying.
My brother likes to wear headphones at all times, and if you want to lie about something unimportant you could take that approach. If you want to lie about something important, you won't be able to.
Refusing to glow randomly wouldn't work, but setting out to exact a cost, social or otherwise (assuming one endorse a role of referee/trusted party, by keeping a tab on who has to pay a premium/is banned from your services), on any request ot prove yourself by glowing (or any request which offend you by implying you're un-trustworthy), could discourage people from being distrustful of you when you don't glow.
I've recently listened to the Clearer Thinking podcast episode on non-dual meditation with Michael Taft https://podcast.clearerthinking.org/episode/167/michael-taft-and-jeremy-stevenson-glimpses-of-enlightenment-through-nondual-meditation/ (this may have been discussed in previous Open Threads, since it's an episode from July). I've often seen discussions about meditation on ACX, but not so much about the dual vs nondual approach, so any comment on that regard from the crowd here would be interesting.
Anyway, one bit caught my attention in particular, when discussing the difference between being "awakened" and being "liberated":
---
MICHAEL: Well, we can still have material in the deep unconscious that is distorting our behavior, or in an unconscious way, causing us to do stuff for us to have reactions. And of course, if we're seeing that clearly enough, with enough emptiness, it's going to have a minimal effect. But over time, when you have deep, stable, nondual awakening, the name of the game is (if I were to use Western terminology) 'to start clearing out all the knots and difficulties in the deep unconscious'. If I was to use the traditional terminology, it's 'you're cleaning out all the karma', all the samskaras. It's when you not only have deep, stable, ongoing awakening, but also all that baggage is cleared out completely, then you're liberated because that's what liberated means, you're seeing everything clearly as it is. That's the awakened part. And then the liberated part is you're not being controlled by your unconscious or to put it in traditional terms, your karma.
SPENCER: And so what are you being controlled by? Or is the idea that you're kind of choosing for yourself or something else? So, you're saying you're trying to avoid being controlled by your deep subconscious. So then, without the control of your deep subconscious, how are your choices being made?
MICHAEL: At that point, it's part of the spontaneous flow of experience. It's not just a reactive trigger from the past.
---
I don't know if I am convinced that that avoiding reactive triggers and let "the spontaneous flow of experience" guide your actions is a fully desirable outcome. I also imagine it's kind of hard to talk rationally about this, it seems to me that he's describing an experience more than a concept. However, I'd like to hear opinions by more experienced meditators on this point.
> avoiding reactive triggers and let "the spontaneous flow of experience"
Could be compared to thoroughly debugging a computer program.
Wow, I did not expect to get no replies at all!
I wonder if it was too early in the Open Thread, or not a question of interest for ACX readers...
How to not be extremely agreeable?
Any sure ways to be a little antagonistic and machiavellian (lol Idk if this is a relevant term)?
Also do you know any good or maybe could suggest Therapist+Psychiatrist here in Germany?
Tentative-- pay attention to your feet and the floor under your feet, and also your belly. This might help you remember that you're a person rather than getting caught up in what the other person wants.
Don't laugh at people's attempted jokes unless you actually think they are funny
If this is in a work context, remember that you are being paid to play a role, and that telling people things they don't want to hear (for example) may be part of that role.
Try using reddit. It did make me much, much less agreeable. Preferably have some expertise in something and try to communicate it to people there.
I suggest "Never Split the Difference". Thinking and answering for myself "what does success look like" for various conversations or scenarios has made a big difference to me.
A handy habit: say “let me get back to you” or “let me think about that” instead of agreeing right away.
First it signals that you’re not always going to give them what they want. Second it gives you time to articulate a reason for refusing, if you want to refuse. Some over-agreeable people feel like they have to give the asker what they want if they can’t immediately express a justification not to.
Also: avoid pre-emptive justification. Even when you have a perfectly good reason to refuse someone (say you have a funeral on that day), just say “sorry, can’t make it”. You can explain why not if they ask. But get yourself out of the habit of feeling you need to explain why as a part of the refusal. Most of the time people totally understand and respect that whatever reason you have is good enough.
Edit: How to be more Machiavellian - articulate to yourself a goal you want to achieve in any given interaction or relationship. Then go do it.
Train yourself to act with intent towards an outcome rather than just acting how you feel. You don’t respect a person but they can give you something you want? Chuck some insincere flattery their way. Want a subordinate to treat you with more respect? Be more distant and less warm towards them (even if you happen to like them personally). That sort of thing.
Did you ask me to text you (Sept 1) or was it some kind of scam/hack? If it was please let Substack know.
See item #1 above!
But come to think of it now, my reply to your (genuine?) "care to explain more?" did warrant some sort of response (in common courtesy) surely?
Where is this?
Sept 1 Your 'Highlights...on Fetishes' post.
You replied to my initial comment: "Care to explain what you mean further?"
I replied...."I was partly just teasing! I am a 72 year old who pines for the days when none of this stuff got talked about. Wittgenstein's famous; ".... those things of which we cannot speak, we must be silent" is one of my favourite quotes....(although I may be misapplying it here.) I am not blind to the upsides of the times....the internet is a marvel and internet discourse is too - for those with the right personality blend of curiosity and skepticism (which unfortunately is only a small minority of folk)....I tend to think of Western liberalism as a great invention but - Icarus-like - one that does not know when enough is enough. I am a pessimist about the future of Liberty."
I saw your explanation, appreciate it, and given that it sounded like it was mostly a joke, and otherwise an invitation to a giant unresolveable debate about fundamental principles, chose not to continue that discussion thread further.
Thanks....it wasn't mostly a joke.... mostly serious. But musings on fundamental principles (whether resolvable or not)... Yes that's my 'thing' you could say.
Sorry.... didn't see that! Just read the "ask me anything"
How should I handle the voting for the book review contest?
1. First past the post - easy for you, easy for me, everyone understands it. But it doesn't offer much of a chance to express true preferences.
2. Approval voting - easy for you, easy for me, provides more choice. But I'm afraid that this advantages milquetoast reviews that everyone could appreciate a little and nobody had anything against.
3. Ranked choice voting - possibly easy for you, pretty hard for me: I don't really know how to implement RCV on Excel or SPSS without doing it manually for each of the sixteen entries for however many rounds there are. If you choose this, give me some hints for how this could work.
4. Something else (I will not choose this and will resent you for proposing it).
Ranked choice voting. I think your intuition is right about approval voting not being a good match for what we care about. I'll text you at your SC number with a python script that should make it super easy to do from Google forms if you want to go that route.
Just emailed you at your SSC address -- let me know if it didn't go through, or if I should email elsewhere!
3 (ish): RCV but *not* IRV; use a Condorcet method.
Can be brute force tallied if need be, by iterating through all the ballots for each pairwise matchup.
Approval voting, with explicit instructions to please only vote for book reviews you actively want to win -- nothing you're merely ok with.
This exists: https://www.rcv123.org/. I know you know how to google, so I'm thinking it may not be what you want. I'm an Australian, so RVC makes the most sense to me, but it's a book review contest, not an election, but then again, seeing the preferences would be interesting and that's what we're here for.
If approval voting, I commit to choosing only my favorite three. I encourage others to do the same as a compromise between 1 and 2.
So in approval voting, if I vote more than once, it's like I get another vote? That seems wrong. If you are going to pick more than one, then each should only get a fraction of a vote... well in my perfect world.
There is a dilution if you vote more than once. Consider the extreme case of voting for everyone--then your vote does not affect the outcome. Or the case of voting for all but one--like downvoting the one.
It seems that this would all but guarantee lowest-common-denominator mediocrity, or something close to it. If there are any mediocre reviews.
Something else:
Simple google forms where you are asked to rate every review from 1-10. Top average rated review wins.
Most of them have left no impression on me, so I wouldn't be able to rate them all. There's only one that I really liked.
If they left no impression on you, then you don't want them to win. It's really that simple.
(But that's precisely why it should be called "range voting" rather than "rating". People need to be disabused from the notion that they're making objective ratings rather than a subjective valuation relative to a small set of other choices on the ballot. Once that can be accomplished, it's really the best method for a case like this.)
So set a minimum number of ratings threshold
Clearly there's a lot of disagreement about what kind of voting system to use. What voting system should we use to determine what voting system to use for the book review contest?
1. FPTP
I have some ideas for 4, but I'll forebear to mention them.
For 1-3, my preference is 2.
But if you do decide to go for 3, it's pretty easy to write a python or JS script that will read a CSV file exported from Excel or Google Sheets and tabulate an Instant Runoff election based on the voting data. Here is one such script I found with a quick googling: https://gist.github.com/Sam-Belliveau/943142dab814b674ecf6f6a852aba0cf
Or if you'd prefer, it'd probably take no more than an hour or two for me to write one from scratch.
For the newer readers: https://en.wikipedia.org/wiki/Arrow%27s_impossibility_theorem
All voting systems suck, but some do suck more than others.
How about we collect the ranked preferences of all the readers, and then calculate all the game theory equilibria under the assumption that every voter will vote strategically, and the winners are whoever wins at least one of these equilibria. (Of course, there might be no equilibria, and thus no winners.)
On, RCV/#3...isn't this place full of programmers? Isn't this just a python script?
Like, presuming every row in an Excel sheet is one voter and every column is a book review and score...I'm not sure I can think of a good way to do it but I can think of 2-3 "dumb but good enough" ways.
But yeah, if RCV is the choice, just ask. Someone should be able to knock that out pretty fast.
Too bad we don't have something like an artificial intelligence which can parse every reply, and tally the results.
Scott would need to have Python, know how to run the script, format the input in the correct format (CSV or space delimited? Header or no header?), and then trust that it has no bugs. Not really trivial if Scott has no programming experience.
Or he could give the CSV file to a programmer volunteer who will tweak the script as needed, run the script for him, and send him the results. I'm willing to volunteer for this if he decides to go that route.
And the winner of the book review contest is...Erica Rall! Who knew she was such a great writer?
In all seriousness, your suggestion could work if multiple programmers are given the CSV and asked to write their own script. Then Scott could check if they all come up with the same list of winners. Although at that point, it might be easier for Scott to do instant runoff by hand.
That's not how the internet works. You don't ask a bunch of people for code, you ask one person for code, you post that, and then everyone and their grandma posts their "better" code because "obviously" that original code is grossly wrong :)
2. Approval voting. If nothing else, using approval voting for some mundane things like book review contests (or choosing which movie to see with your friends, or whatever) make approval voting less of an abstract thing you have heard of, and more of a valid and intuitive approach to voting. So I think it is a good think in terms of popularising and legitimising approval voting, which contributes a little to it being potentially used in elections where it really matters.
3. Apparently useful instructions for implementing RCV: https://www.rankedvote.co/guides/applying-ranked-choice-voting/how-to-calculate-ranked-choice-voting-with-google-forms-and-google-sheets
That's only IRV, not all forms of RCV; there are better (Condorcet) ways of tabulating ranked ballots.
4. I propose that each voter is given the task of casting 4 votes, that he/she/they splits among 1 to 4 reviews. This is a kind of approval voting , but it does not benefit milquetoast reviews because highly favored entries can gather all votes from a voter.
Yeah I was going to suggest something like this. It could also be called 'first past the post', but you get to split your vote if you want.
(I'm not suggesting this though, because I fear the wrath of our overlord. :^)
4. Something else: Range voting (e.g. 1-10), which is like approval voting but lets you give a score to things, hopefully avoiding the milquetoast problem. Also hopefully as easy to implement as approval voting.
Downside: people could give scores of 1 or 10 exclusively, but then it's exactly like approval voting.
Outside that, my preference is approval > ranked choice > FPTP. I think FPTP is too one-shot. I would rather a milquetoast "everybody kind of liked this review" win over a more polarising one, and I guess this is a good example of how choosing the scoring system can affect what the winner looks like.
I think first past the post is fine for something like this, where you are picking from a number of acceptable choices and judging on quality rather than values. Approval doesn't really work when the quality of candidates is a smooth gradient and there is no clear line between acceptable and unacceptable. I doubt people will want to put in the effort to cast an informed ranked choice ballot with this number of entries. If you do this, I would restrict it to something like top 3.
Yeah, I think top three is the best compromise. First past the post works if I have an outstanding entry that I think is the best of all sixteen, but more likely I'll have "well there were three I thought were really good, I can't pick one out of them" and then not vote.
In the US? Publisher or distributor (Amazon)? Definitely not. The Author? Probably not. Here is an old article asking about this exact scenario, minus AI:
http://www.rightsofwriters.com/2011/03/oops-i-poisoned-my-readers-can-i-be.html
>Ultimately, in this important decision, the Ninth Circuit found that the publisher had no liability to its readers for the nearly-deadly information. Essentially, the court concluded that imposing liability on the publisher would open the door to the prospect of unlimited liability that could severely threaten the free flow of information.
The article's conclusion is "probably not". I think it would be especially hard to convict an author in a criminal trial. In a civil trial, I could see a jury siding with the plaintiff but it's a gray area.
"Well, sir, the book you wrote killed a mushroom-hunter. What do you have to say about that?"
Which of the following is GPT4's likely response?
(a)Jeez, I am just a sucky little large language model.
(b)I was just doing my job, which is predicting tokens. I am very good at it, but tokens are not mushrooms, and the responsibility lies with whoever did not take that into account.
(c) "Once the rockets go up who cares where they come down?
That's not my department" said Werner von Braun.
AI is not going to paperclip us, it's going to the Australian Inlaws Dinner us 😁
https://www.independent.co.uk/news/world/australasia/death-cap-mushroom-poisoning-australia-erin-patterson-b2403356.html
"Hello, sir/madam/other! I am a friendly, helpful LLM! People wanted to know where to find mushrooms and I told them!
They never stipulated they wanted to find *safe* mushrooms!"
Yeah, can't wait til we start relying on it for more serious tasks, like stamping out some widespread infectious disease. "Oh, you wanted me to stamp it out *without* killing the sick, the infected, the possibly infected and the highly vulnerable ? Your prompt did not specify that."
“These mushrooms are perfectly safe! Some species of frog have a special gland which lets them process the toxins safely!”