Posts
Comments
I'm replying to a post that said they lacked energy despite iron supplementation. Maybe it wasn't iron deficiency, or maybe it could have been solved by raising the supplement dose, I don't know, but if it was iron deficiency and the supplements weren't helping then it's good to know you can get iron in the same form as in meat from Impossible burgers.
Yeah, this is a US-centric perspective of mine but there's no shortage of land here. This sounds to me like classic thriftiness about that which is not scarce, which isn't real thriftiness. I mean, "effective use of both farmland and rooftops"... rooftops? What's scarce here in the US is labor, not land. Why have all these people climbing up on rooftops? An interesting contrast is Texas (mostly utility solar) vs California (lots of rooftop solar). The interesting number, which I don't know off the top of my head, is how many people are employed in the solar sector per unit capacity installed. I seem to remember California employs a lot more people, disproportionately to how much more solar it has.
The heme iron in meat is absorbed better than the non-heme iron in iron supplements, but Impossible Burger has heme iron. It's very frustrating that this biotech advance of in vitro heme production is so far only used for this specific brand of meat substitutes but that's the situation. I'm not sure why iron supplements didn't work for you, as that same paper shows that even non-heme iron is absorbed well when blood iron is low, but maybe it depends on the individual? In any case, and I promise the company isn't paying me to say this, I recommend Impossible burgers to vegans. It has to be this specific brand, they have the patent. There was another company with a heme production patent but they recently shut down.
The natural gas generation capacity that you need to cover for solar when it's cloudy is, of course, less than what is required to make up for loss of solar after sundown.
Sure. There's enough sunlight to run the whole country, so it's physically possible, but it's not at the moment technologically or economically practical, and may not be our best option in the near future. Until this wave of battery installations, though, I thought even California had saturated its solar potential. In the next post I'll write in more detail about what I think is now possible, but briefly, it's now feasible for all western US peak load (the extra power needed while people are awake) to be provided by solar and batteries. Whether we'll also use solar for base load, and whether we'll use it in cloudy areas, is a more difficult question that requires extrapolating prices, and I'll try to address that in the third post.
I finally googled what Elon Musk has said about solar power, and found that he did a similar calculation recently on twitter:
Once you understand Kardashev Scale, it becomes utterly obvious that essentially all energy generation will be solar.
Also, just do the math on solar on Earth and you soon figure out that a relatively small corner of Texas or New Mexico can easily serve all US electricity.
One square mile on the surface receives ~2.5 Gigawatts of solar energy. That’s Gigawatts with a “G”. It’s ~30% higher in space. The Starlink global satellite network is entirely solar/battery powered.
Factoring in solar panel efficiency (25%), packing density (80%) and usable daylight hours (~6), a reasonable rule of thumb is 3GWh of energy per square mile per day. Easy math, but almost no one does these basic calculations.
Sort of. I think the distribution of Θ is the Ap distribution, since it satisfies that formula; Θ=p is Ap. It's just that Jaynes prefers an exposition modeled on propositional logic, whereas a standard probability textbook begins with the definition of "random variables" like Θ, but this seems to me just a notational difference, since an equation like Θ=p is after all a proposition from the perspective of propositional logic. So I would rather say that Bayesian statisticians are in fact using it, and I was just explaining why you don't find any exposition of it under that name. I don't think there's a real conceptual difference. Jaynes of course would object to the word "random" in "random variable" but it's just a word, in my post I call it an "unknown quantity" and mathematically define it the usual way.
In solar-heavy areas before batteries (and without hydro), electricity in the early evening was provided by natural gas peaker plants, which can and do quickly shut off. Consider a scenario with growing demand. Prices in the early evening have to get pretty high before it's worth paying for a whole natural gas plant just to run it for only a few hours.
I seen the argument made by Robert Bryce and Alex Epstein, who don't suggest economic models, but the reason it's at least not obvious to me is that we need to consider supply, demand, price as functions of time. Solar produces a glut of electricity during the day. It makes sense to me that it would increase the price of electricity in the early evening, when solar isn't generating but demand is still high. It would do so by reducing the profitability of building natural gas plants to supply those hours, which results in either fewer natural gas plants (if demand is elastic) or the prices rising until they're profitable (if demand is inelastic). How this affects the average price throughout the whole day I don't know.
Yes, Tesla's role in battery storage is actually odd and niche—their importance seems to be reducing the skill level required to built a battery energy storage facility by packaging batteries into self contained modules that contain all of the required equipment (thermal management, inverter, etc). The battery cells come from Asia.
The question to which Musk is relevant is not "how did we get to a world where battery storage is feasible" but "why would someone be investing in this ten or fifteen years ago when the technology was not there". It seems to me to be a case where a futurist calculation that ignores the engineering details plus seemingly naive reasoning that the rest is "just engineering" would have actually given the right answer.
This is a good example of neglecting magnitudes of effects. I think in this case most people just don't know the magnitude, and wouldn't really defend their answer in this way. It's worth considering why people sometimes do continue to emphasize that an effect is not literally zero, even when it is effectively zero on the relevant scale.
I think it's particularly common with risks. And the reason is that when someone doesn't want you to do something, but doesn't think their real reason will convince you, they often tell you it's risky. Sometimes this gives them a motive to repeat superstitions. But sometimes, they report real but small risks.
For example, consider Matthew Yglesias on the harms of marijuana:
Inhaling smoke into your lungs is, pretty obviously, not a healthy activity. But beyond that, when Ally Memedovich, Laura E. Dowsett, Eldon Spackman, Tom Noseworthy, and Fiona Clement put together a meta-analysis to advise the Canadian government, they found evidence across studies of “increased risk of stroke and testicular cancer, brain changes that could affect learning and memory, and a particularly consistent link between cannabis use and mental illnesses involving psychosis.”
I'll ignore the associations with mental illness, which are known to be the result of confounding, although this is itself an interesting category of fake risks. For example a mother that doesn't want her child to get a tattoo, because poor people get tattoos, could likely find a correlation with poverty, or with any of the bad outcomes associated with poverty.
Let's focus on testicular cancer, and assume for the moment that this one is not some kind of confounding, but is actually caused by smoking marijuana. The magnitude of the association:
The strongest association was found for non-seminoma development – for example, those using cannabis on at least a weekly basis had two and a half times greater odds of developing a non-seminoma TGCT compared those who never used cannabis (OR: 2.59, 95 % CI 1.60–4.19). We found inconclusive evidence regarding the relationship between cannabis use and the development of seminoma tumours.
What we really want is a relative risk (how much more likely is testicular cancer among smokers?) but for a rare outcome like testicular cancer, the odds ratio should approximate that. And testicular cancer is rare:
Testicular cancer is not common: about 1 of every 250 males will develop testicular cancer at some point during their lifetime.
So while doubling your testicular cancer risk sounds bad, doubling a small risk results in a small risk. I have called this a "homeopathic" increase, which is perhaps unfair; I should probably reserve that for probabilities on the order of homeopathic concentrations.
But it does seem to me to be psychologically like homeopathy. All that matters is to establish that a risk is present, it doesn't particularly matter its size.
Although this risk is not nothing... it's small but perhaps not negligible.
It's great to have a LessWrong post that states the relationship between expected quality and a noisy measurement of quality:
(Why 0.5? Remember that performance is a sum of two random variables with standard deviation 1: the quality of the intervention and the noise of the trial. So when you see a performance number like 4, in expectation the quality of the intervention is 2 and the contribution from the noise of the trial (i.e. how lucky you got in the RCT) is also 2.)
We previously had a popular post on this topic, the tails come apart post, but it actually made a subtle mistake when stating this relationship. It says:
For concreteness (and granting normality), an R-square of 0.5 (corresponding to an angle of sixty degrees) means that +4SD (~1/15000) on a factor will be expected to be 'merely' +2SD (~1/40) in the outcome - and an R-square of 0.5 is remarkably strong in the social sciences, implying it accounts for half the variance.
The example under discussion in this quote is the same as the example in this post, where quality and noise have the same variance, and thus R^2=0.5. And superficially it seems to be stating the same thing: the expectation of quality is half the measurement.
But actually, this newer post is correct, and the older post is wrong. The key is that "Quality" and "Performance" in this post are not measured in standard deviations. Their standard deviations are 1 and √2, respectively. Elaborating on that: Quality has a variance, and standard deviation, of 1. The variance of Performance is the sum of the variances of Quality and noise, which is 2, and thus its standard deviation is √2. Now that we know their standard deviations, we can scale them to units of standard deviation, and obtain Quality (unchanged) and Performance/√2. The relationship between them is:
That is equivalent to the relationship stated in this post.
More generally, notating the variables in units of standard deviation as and (since they are "z-scores"),
where is the correlation coefficient. So if your noisy measurement of quality is standard deviations above its mean, then the expectation of quality is standard deviations above its mean. It is that is variance explained, and is thus 1/2 when the signal and noise have the same variance. That's why in the example in this post, we divide the raw performance by 2, rather than converting it to standard deviations and dividing by 2.
I think it's important to understand the relationship between the expected value of an unknown and the value of a noisy measurement of it, so it's nice to see a whole post about this relationship. I do think it's worth explicitly stating the relationship on a standard deviation scale, which this post doesn't do, but I've done that here in my comment.
Some other comments brought up that the heme iron in meat is better absorbed, which is true, see figure 1. But the good news is that Impossible burgers have heme iron. They make it in yeast by adding a plasmid with the heme biosynthesis enzymes, pathway in Figure 1 of their patent on the 56th page of the pdf.
I think we'll have an internet full of LLM-bots "thinking" up and doing stuff within a year.
Did this happen? At least not obviously.
Yes, it seems like biotech will provide the tools to build nanotech, and Drexler himself is still emphasizing the biotech pathway. In fact, in Engines of Creation, Drexlerian nanotech was called "second-generation nanotech", with the first generation understood to include current protein synthesis as well as future improvements to the ribosome.
I don't really see the point of further development of diamondoid nanotech. Drexler made his point in Nanosystems: certain capabilities that seem fantastical are physically possible. It conveniently opens with a list of lower bounds on capabilities, and backs them up with what is, as far as I'm concerned, and enough rigor to make the point.
Once that point has been made, if you want to make nanotechnology actually happen, you should be focused on protein synthesis, right? What you need is not better nanotech designs. It's some theory of why protein engineering didn't take over abiotic industry the way people expected, why we're building iridium-based hydrogen electrolyzers at scale and have stopped talking about using engineered hydrogenases and so on. A identification of the challenges, and a plan for addressing them. What's the point of continuing to hammer in that second-generation nanotech would be cool if only we could synthesize it?
I didn't feel chills from music for a long time, and then started to get them again after doing physical therapy and learning exercises to straighten my back and improve my posture. It was a notable enough change that I reported it to my physical therapists, but I don't recall how I interpreted it at the time ("I'm getting chills again" vs "chills are real??" or what).
An example important in my life is planning: I "couldn't" make long-term plans or complete my to-do list as long as my "to-do list" was just a list of obligations rather than anything I really wanted done. More generally, I think plans "on paper" are especially easy case, since they don't take a telepath. For example, see the planning fallacy and Robin Hanson's comment that managers prefer the biased estimates. Getting to a corporate level, there's accounting. A cool related image is in episode two of Twin Peaks when Josie opens the safe and finds two ledgers, one making the mill look profitable and the other tracking the real debts. That's an example of "occlumency" I guess. But how common is it to have two account books like that, one internally and the other for investors? Or two schedules, a real one for real planning and a fake one to project optimism to the higher-ups? Or two to-do lists, one of stuff I plan to really do and the other I'm just saying I'll do to avoid an argument? Like, actually two files or pieces of paper? Certainly in a corporate context there's good legal reasons not to, since liability for fraud often depends on what you can be proven to know, right?
I wonder if there's also an analogy to the Gibbs sampling algorithm here.
For a believer, it will mostly bounce back and forth from "Assuming God is real, the bible is divinely inspired" and "Assuming the bible is divinely inspired, God must be real". But if these are not certainties, occasionally it must generate "Assuming God is real, the bible is actually not divinely inspired". And then from there, probably to "Assuming the bible is not divinely inspired, God is not real." But then also occasionally it can "recover", generating "Assuming the bible is not divinely inspired, God is actually real anyway." So you need that conditional probability too. But given all the conditional probabilities, the resulting chain generates the joint distribution over whether or not the bible is divinely inspired and whether or not God is real.
The reason nobody else talks about the A_p distribution is because the same concept appears in standard probability expositions as a random variable representing an unknown probability. For example, if you look in Hoff's "A First Course in Bayesian Statistics", it will discuss the "binomial model" with an unknown "parameter" Θ. The "event" Θ=p plays the same role as the proposition A_p, since P(Y=1|Θ=p) = p. I think Jaynes does have something to add, but not so much in the A_p distribution chapter as in his chapter on the physics of coin flips, and his analysis of die rolls which I'm not sure if is in the book. He gets you out of the standard Bayesian stats mindset where reality is a binomial model or multinomial model or whatever, and shows you that A_p can actually have a meaning in terms of a physical model, such as a disjunction of die shapes that lead to the same probability of getting 6. Although your way of thinking of it as a limiting posterior probability from a certain kind of evidence is interesting too (or Jaynes's way of thinking of it, if it was in the book; I don't recall). Anyway, I wrote a post on this that didn't get much karma, maybe you'll be one of the few people that's interested.
Make sense. I suppose we assume that the insurance pays out the value of the asset, leaving our wealth unchanged. So assuming we buy the insurance, there's no randomness in our log wealth, which is guaranteed to be log(W-P). The difference between that, and our expected log wealth if we don't buy the insurance, is V. That's why log(W-P) is positive in the formula for V, and all the terms weighted by probabilities are negative.
My guesses at what the spoiler was going to be:
-
Ten non-independent trials, a 10% chance each (in the prior state of knowledge, not conditional on previous results,), and only one trial can succeed. You satisfy these conditions with something like "I hid a ball in one of ten boxes", and the chance really is 100% that one is a "success".
-
Regardless of whether the trials are independent, the maximum probability that at least one is a success is the sum of the probabilities per trial. In this case that doesn't yield a useful bound because we already know probabilities are below 100%, but in general it's useful.
Yeah, it's cool that "I did n trials, with a 1/n chance each, so the probability of at least one success is... " does have a general answer, even if it's not 100%. Just noting that it's not the only small modification of the title yielding a useful and interesting correct statement.
The ones that came to my mind still involved the sum of the per-trial probabilities. If it was clear that we were looking for something preserving the "n trials with 1/n chance", rather than the summation, I think it would have been more obvious where you're going with this.
Another consideration, though maybe not a fundamental one, is that past and future selves are the only beings we know for sure that we have lots of subjunctive dependence with, just from "structural" similarity like calculators from the same factory (to use an example from the TDT paper). Tumblr usr somnilogical elaborated on this a bit, concluding "future selves do help past selves!" An upload is a future self in the way that matters for this conclusion.
Yes, and I did look at something like four of the individual studies of depression, focusing on the ones testing pills so they would be comparable to the Prozac trial. As I said in the post, they all gave me the same impression: I didn't see a difference between the placebo and no-pill groups. So it was surprising to see the summary value of -0.25 SMD. Maybe it's some subtle effect in the studies I looked at which you can see once you aggregate. But maybe it's heterogeneity, and the effect is coming from the studies I didn't look at. As I mentioned in the post, not all of the placebo interventions were pills.
From a quick look on Wikipedia I don't see anything. Except for patients that report side effects from placebo, but of course that could be symptoms that they would have had in any case, which they incorrectly attribute to the placebo.
I don't see how you could get an accurate measure of a nocebo effect from misdiagnoses. I don't think anyone is willing to randomize patients to be misdiagnosed. And if you try to do it observationally, you run into the problem of distinguishing the effects of the misdiagnosis from whatever brought them to the doctor seeking diagnosis.
I took a look at The Secret of Our Success, and saw the study you're describing on page 277. I think you may be misremembering the disease. The data given is for bronchitis, emphysema and asthma (combined into one category). It does mention that similar results hold for cancer and heart attacks.
I took a look at the original paper. They checked 15 diseases, and bronchitis, emphysema and asthma was the only one that was significant after correction for multiple comparisons. I don't agree that the results for cancer and heart attacks are similar. They seem within the range of what you can get from random fluctuations in a list of 15 numbers. The statistical test backs up that impression.
If this is a real difference, I would expect it has something to do with lifestyle changes, rather than stress. However, it's in the opposite direction of what I would expect. That is, I would expect those with a birth year predisposed to lung problems to avoid smoking. I did find some chinese astrology advice to that effect. Therefore they should live longer, when in fact they live shorter. So that doesn't really make sense.
This result seems suspicious to me. First of all because it's just a couple of diseases selected out of a list of mostly non-significant tests, but also because people probably do lots of tests of astrology that they don't publish. I wouldn't bet on it replicating in another population.
It's okay, the post is eight pages long and not super internally consistent, basically because I had to work on Monday and didn't want to edit it. I don't make a post like that expecting everyone to read every paragraph and get a perfectly clear idea of what I'm saying.
Observation of the cosmic microwave background was a simultaneous discovery, according to James Peebles' Nobel lecture. If I'm understanding this right, Bob Dicke's group at Princeton was already looking for the CMB based on a theoretical prediction of it, and were doing experiments to detect it, with relatively primitive equipment, when the Bell Labs publication came out.
"Norman Borlaug’s Green Revolution" seems to give the credit for turning India into a grain exporter solely to hybrid wheat, when rice is just as important to India as wheat.
Yuan Longping, the "father of hybrid rice", is a household name in China, but in English-language sources I only seem to hear about Norman Borlaug, who people call "the father of the Green Revolution" rather than "the father of hybrid wheat", which seems more appropriate. It seems that the way people tell the story of the Green Revolution has as much to do with national pride as the facts.
I don't think this affects your point too much, although it may affect your estimate of how much change to the world you can expect from one person's individual choices. It's not just that if Norman Borlaug had died in a car accident hybrid wheat may still have been developed, but also that if nobody developed hybrid wheat, there would still have been a Green Revolution in India's rice farming.
Green fluorescent protein (GFP). A curiosity-driven marine biology project (how do jellyfish produce light?), that was later adapted into an important and widely used tool in cell biology. You splice the GFP gene onto another gene, and you've effectively got a fluorescent tag so you can see where the protein product is in the cell.
Jellyfish luminescence wasn't exactly a hot field, I don't know of any near-independent discoveries of GFP. However, when people were looking for protein markers visible under a microscope, multiple labs tried GFP simultaneously, so it was determined by that point. If GFP hadn't been discovered, would they have done marine biology as a subtask, or just used their next best option?
Fun fact: The guy who discovered GFP was living near Nagasaki when it was bombed. So we can consider the hypothetical where he was visiting the city that day.
I like it a lot. I'm mainly a tumblr usr, and on tumblr we're all worried about the site being shut down because it doesn't make any money. I love having LessWrong as a place for writing up my thoughts more carefully than I would on tumblr, and it also feels like a sort of insurance policy if tumblr goes under, since LessWrong seems to be able to maintain performance and usability with a small team. The mods seem active enough that they frontpage my posts pretty quickly, which helps connect them with an audience that's not already familiar with me, whereas on tumblr I haven't gotten any readers through the tag system in years and I'm coasting on inertia from the followers I already have.
As someone who grew up with Greg Egan on the shelf, I want to note that Greg Egan said basically the same thing about "Neuromancer" (that it cares more about being fashionable than having the characters think through their situation), and "Quarantine" and "Permutation City" were in part responses to cyberpunk, so perhaps all is not lost.
Backing that up with Greg Egan interview quotes.
From the Karen Burnham interview, on hating "Neuromancer", and on the influence of cyberpunk on "Quarantine":
I read Neuromancer in 1985, because I was voting for the Hugos that year and I thought I ought to read all the nominated novels. I really hated it; aside from the style and the characters, which definitely weren't to my taste, a lot of things about the technology in the book seemed very contrived and unlikely, especially the idea that anyone would plug in a brain-computer interface that they knew a third party could use to harm them.
Over the next few years I read some Rucker and Sterling novels, which I definitely enjoyed more than Gibson. So there was some reasonable stuff written under the cyberpunk banner, but none of it felt very groundbreaking to anyone who'd been reading Dick and Delany, and if it hadn't been wrapped in so much hype I probably would have enjoyed it more. In fact, the way cyberpunk as a movement influenced me most was a sense of irritation with its obsession with hipness. I don't think there's much doubt that “Axiomatic” and the opening sections of Quarantine have a kind of cyberpunk flavour to them, but my thinking at the time would have been less “Maybe I can join the cyberpunk club!” and more “Maybe I can steal back private eyes and brain-computer interfaces for people who think mirror shades are pretentious, and do something more interesting with them.”
From the Marisa O’Keeffe interview, something that corroborates what Eliezer Yudkowsky said about "Neuromancer" characters worrying how things look on a t-shirt:
A lot of cyberpunk said, in effect: “Computers are interesting because cool, cynical men (or occasionally women) in mirrorshades do dangerous things with them.” If that really is the most interesting thing you can imagine about a computer, you shouldn’t be writing SF.
From the Russell Blackford interview, on the influence of cyberpunk on "Permutation City":
I recall being very bored and dissatisfied with the way most cyberpunk writers were treating virtual reality and artificial intelligence in the ’80s; a lot of people were churning out very lame noir plots that utterly squandered the philosophical implications of the technology. I wrote a story called “Dust”, which was later expanded into Permutation City, that pushed very hard in the opposite direction, trying to take as seriously as possible all the implications of what it would mean to be software. In the case of Permutation City that included some metaphysical ideas that I certainly wouldn’t want to repeat in everything I wrote, but the basic notions about the way people will be able to manipulate themselves if they ever become software, which I developed a bit further in Diaspora, seem logically unavoidable to me.
Something depressing is certainly going on in mainstream culture, since for example "The New York Times" hasn't had a review of a Greg Egan book since "Diaspora" in 1998, except to suggest "that Egan doesn’t fully understand how oppression works — or that he is trying to make an inappropriate point".
But science fiction seems alright, if it reacted to "Neuromancer" exactly along the lines of Eliezer Yudkowsky's reaction to this post, producing some of the most beloved (by sci-fans) science fiction of the 90s. And I still see every new Alastair Reynolds book in the sci-fi sections of non-specialty bookstores.
I'm looking into the history of QM interpretations and there's some interesting deviations from the story as told in the quantum sequence. So, of course, single-world was the default from the 1920s onward and many-worlds came later. But the strangeness of a single world was not realized immediately. The concept of a wavefunction collapse seems to originate with von Neumann in the 1930s, as part of his mathematicization of quantum mechanics–which makes sense in a way, imagine trying to talk about it without the concept of operators acting on a Hilbert space. I haven't read von Neumann's book, but the 50s, 60s, and 70s discussions of a measurement problem seem to draw on him directly. And the idea that QM requires fundamental irreducible minds seems to date to Wigner's "Remarks on the Mind-Body Question", published in 1961. Wigner mentions that Heisenberg thought QM was fundamentally describing our knowledge of the world, but that seems different from consciousness specifically, causing collapse specifically, though I don't know Heisenberg's views well. What makes this weird is this is after many-worlds! Notably, DeWitt's 1970 article which popularized many-worlds seems to associate the "consciousness-causes-collapse" thing with Wigner specifically, giving Wigner more credit for it than Wigner gives himself. It's not quite correct to say that "consciousness-causes-collapse" originated with Wigner's article, since the "Wigner's friend" thought experiment was actually discussed by Everett. Unsurprisingly, since Wigner was a professor at Everett's school, so they likely discussed these issues. So the deviation from the story in the quantum sequence is that "consciousness-causes-collapse" was not the default theory which many-worlds had to challenge. Instead, they were contemporary competitors, introduced at basically the same time, with the same motivation. Of course, it remains the case that single-world was the default, and Wigner was arguably just following that where it led. But the real "Copenhagen" opinion, it seems to me, was against talking about a picture of the world at all. To say that there is some non-linear irreversible consciousness-initiated collapse, actually occurring in the world, is already a heresy in Copenhagen.
Yes, Sleeping Beauty has to account for the fact that, even if the result of the coin flip was such that she's being woken up on both Monday and Tuesday, if she bets on it being Monday, she will surely lose one of the two times. So she needs an extra dollar in the pot from the counterparty: betting $1 to $2 rather than $1 to $1. That pays for the loss when she makes the same bet on Tuesday. In expectation this is a fair bet: she either puts $1 in the pot and loses it, or puts $1 in the pot and gets $3 and then puts $1 in the pot and loses it, getting $2 total.
Anyway, feeling something is an action. I think it's a mistake when people take "anticipation" as primary. Sure, "Make Beliefs Pay Rent (In Anticipated Experiences)" is good advice, in a similar way as a guide to getting rich is good advice. Predictive beliefs, like money, are good to pursue on general principle, even before you know what you're going to use them for. But my anticipations of stuff is good for me to the extent that the consequences of anticipating it are good for me. Like any other action.
I think as usual with rationality stuff there's a good analogy to statistics.
I'm very happy I never took Stats 101 and learned what a p value was in a math department "Theory of Statistics" class. Because as I understood it, Stats 101 teaches recipes, rules for when a conclusion is allowed. In the math department, I instead learned properties of algorithms for estimation and decision. There's a certain interesting property of an estimation algorithm for the size of an effect: how large will that estimate be, if the effect is not there? Of a decision rule, you can ask: how often will the decision "effect is there" be made, if the effect is not there?
Frequentist statistical inference is based entirely on properties like these, and sometimes that works, and sometimes it doesn't. But frequentist statistical inference is like a set of guidelines. Whether or not you agree with those guidelines, these properties exist. And if you understand what they mean, you can understand when frequentist statistical inference works decently and when it will act insanely.
I think what statistics, and LessWrong-style rationality have in common, is taking the procedure itself as an object of study. In statistics, it's some algorithm you can run on a spreadsheet. On LessWrong, it tends to be something more vague, a pattern of human behavior.
My experience as a statistician among biologists was, honestly, depressing. One problem was power calculations. People want to know what power to plug into the sample size calculator. I would ask them, what probability are you willing to accept that you do all this work, and find nothing, even though the effect is really there? Maybe the problem is me, but I don't think I ever got any engagement on this question. Eventually people look up what other people are doing, which is 80%. If I ask, are you willing to accept a 20% probability that your work results in nothing, even though the effect you're looking for is actually present, I never really get an answer. What I wanted was not for them to follow any particular rule, like "only do experiments with 80% power", especially since that can always be achieved by plugging in a high enough effect size in the calculation they put in their grant proposal. I wanted them to actually think through whether their experiment will actually work.
Another problem--whenever they had complex data, but were still just testing for a difference between groups, my answer was always "make up a measure of difference, then do a permutation test". Nobody ever took me up on this. They were looking for a guideline to get it past the reviewers. It doesn't matter that the made-up test has exactly the same guarantee as whatever test they eventually find: only positive 5% of the time it's used in the absence of a real difference. But they don't even know that's the guarantee that frequentist tests come with.
I don't really get what was going on. I think the biologists saw statistics as some confusing formality where people like me would yell at them if they got it wrong. Whereas if they follow the guidelines, nobody will yell at them. So they come to me asking for the guidelines, and instead I tell them some irrelevant nonsense about the chance that their conclusion will be correct.
I just want people to have the resources to think through whether the process by which they're reaching a conclusion will reach the right conclusion. And use those resources. That's all I guess.
I think this thought has analogues in Bayesian statistics.
We choose a prior. Let's say, for the effect size of a treatment. What's our prior? Let's say, Gaussian with mean 0, and standard deviation equal to the typical effect size for this kind of treatment.
But how do we know that typical effect size? We could actually treat this prior as a posterior, updated from a uniform prior by previous studies. This would be a Bayesian meta-analysis.
I've never seen anyone formally do a meta-analysis just to get a prior. At some point, you decide your assumed probability distributions are close enough, that more effort wouldn't change the final result. Really, all mathematical modeling is like this. We model the Earth as a point, or a sphere, or a more detailed shape, depending on what we can get away with in our application. We make this judgment informally, but we expect a formal analysis to back it up.
As for these ranges and bounds... that reminds me of the robustness analysis they do in Bayesian statistics. That is, vary the prior and see how it effects the posterior. Generally done within a parametric family of priors, so you just vary the parameters. The hope is that you get about the same results within some reasonable range of priors, but you don't get strict bounds.
Very interesting. A few comments.
I think you mentioned something like this, but Drexler expected a first generation of nanotechnology based on engineered enzymes. For example, in "Engines of Creation", he imagines using enzymes to synthesize airplane parts. Of course the real use of enzymes is much more restricted: cleaning products such as dishwasher detergent, additives in food, pharmaceutical synthesis. It has always seemed to me that someone who really believed Drexler and wanted to bring his imagined future about would actually not be working on anything like the designs in "Nanosystems", but on bringing down the cost of enzyme manufacturing. From that perspective it's interesting that you note that the most promising direction in Drexlery mechanosynthesis is DNA origami. Not quite what Drexler imagined (nucleic acid rather than protein), but still starting with biology.
Also, I think it's very interesting that silicon turned out to be easier than diamond. While I agree with Yudkowsky that biology is nowhere near the limits of what is possible on the nanometer-scale due to constraints imposed by historical accidents, I disagree with Yudkowsky's core example of this, the weak interactions holding proteins in the folded configuration. Stronger bonds make things harder, not easier. Maybe the switch from diamond to silicon is an illustration of that.
Editing to add one more comment... Drexler's definition of "diamondoid" is indeed strange. If we take it literally, it seems that glass is "diamondoid". But then, "diamondoid" microbes already exist, that is, diatoms. Or at least, microbes with "diamondoid" cell walls.