Posts
Comments
“How many Oxford dons does it take to change a lightbulb?”
CHANGE‽‽‽
To my knowledge, there is currently no method that will generate a reasonably exhaustive list of all the languages a given book has been translated into. I use a combination of Worldcat, Wikipedia, Amazon and Google.
Greg Lewis discusses this at length here.
Metaculus generates lots of valuable metrics besides the "meaningless internet points" about which Zvi and others complained. If Yudkowsky had predicted regularly, he would have been able to know e.g. how well-calibrated he is, how his Brier score evolved over time, how it compares to the community's, etc.
There is no need to create a Flashcards tag when we already have Spaced repetition.
Though I understood what you meant, perhaps a clearer terminology is all-things-considered beliefs vs. independent impressions.
Keynes's view of Leninism seems similar to Russell's, and may have been influenced by it. Here's a quote from The Practice and Theory of Bolshevism (published in 1920):
Bolshevism is not merely a political doctrine; it is also a religion, with elaborate dogmas and inspired scriptures. When Lenin wishes to prove some proposition, he does so, if possible, by quoting texts from Marx and Engels. A full-fledged Communist is not merely a man who believes that land and capital should be held in common, and their produce distributed as nearly equally as possible. He is a man who entertains a number of elaborate and dogmatic beliefs—such as philosophic materialism, for example—which may be true, but are not, to a scientific temper, capable of being known to be true with any certainty. This habit, of militant certainty about objectively doubtful matters, is one from which, since the Renaissance, the world has been gradually emerging, into that temper of constructive and fruitful scepticism which constitutes the scientific outlook. I believe the scientific outlook to be immeasurably important to the human race. If a more just economic system were only attainable by closing men’s minds against free inquiry, and plunging them back into the intellectual prison of the middle ages, I should consider the price too high.
Thanks for the explanation.
Speaking personally, I would much prefer that contributions judged to make an article net worse be reverted than downvoted, both because this gives the contributor much clearer feedback and because it improves the quality of the article, relative to the opinion of the user deciding whether to downvote or revert. If other users disagree with that assessment, the karma system can perhaps then be used to resolve the disagreement, or to at least contribute to a resolution by conveying a signal whose meaning is now much easier to discern.
Concerning the specific edit: on reflection, I agree with you that the information I included would have been more appropriate in the Overcoming Bias article, and I am inclined to agree with you that it is best omitted from Hanson's article. Accordingly, I have restored the previous version of the entry, though with the external link fixed.
I'm not sure how to interpret the downvoting of a recent edit I made. I simply clarified a sentence and replaced a dead link with the version archived by the Wayback Machine. Downvoting substantive posts without providing an explanation is often justified given the difficulty associated with pinpointing the nature of a complex disagreement, or because the signal the downvote is meant to convey is relatively clear from the context. But with a simple, atomic edit that seems unambiguously (very mildly) positive, at least from a commonsense perspective, such unexplained downvotes are apt to leave an editor puzzled and unable to draw any valuable lessons, except perhaps that they should simply abstain from contributing in the future.
This week, I’ll spend 40–50 hours in Virtual Reality (Immersed), like I did last week and every (work) week for the last 2½ years. It’s not just fun and games — there are plenty of those, along with exercise, meditation, creativity, socializing, etc. — but for this article, I’m only focusing on (and counting) the work.
Yes, really: 8–10 hours a day strapped in. I’ve encountered a fair amount of skepticism about both the technology and the general premise, many nit-picks about the software, or how it fails to match some preconception about how things “should” work.
I decided to switch to Emacs 1.5 years ago, and I feel it's the most important computing decision I made since... starting to use a computer? I may write a more detailed post on what things I use Emacs for, but here I just wanted to endorse the above recommendation, including its caveats ("Don't bother with it though if you don't have some time to invest in learning it"), and emphasize that Emacs can fulfill many needs besides "code editor" (I am not a programmer myself).
Thanks, but this post is no longer updated and the link is not broken on my website. (If you think that's confusing, despite the notice at the top, I may consider replacing its contents with just a link, though retaining the content may make it more discoverable.)
Thanks for the update!
chaotic History of Economic Analysis
The word 'chaotic' was an adjective I chose to describe the book's content, rather than part of the book's title. :)
Related to the ReplicationMarkets example: on Metaculus, there is an entire category of self-resolving questions, where resolution is at least in part determined by how users predict the question will resolve. We have seen at least one instance of manipulation of such questions. And there is even a kind of meta-self-resolving question, asking users to predict what the sentiment of Metaculus users will be with regard to self-resolving questions.
The probability I would assign to #8 intuitively is about 0,41. Math based on my other three predictions yields (doing the calculation now) 0.476. I am going to predict the math output rather than my intuition.
I think the correct response to this realization is not to revise your final answer so as to make it consistent with the first three. It is to revise all four answers so that they are maximally intuitive, subject to the constraint that they be jointly consistent. Which answer comes last is just an artifact of the order of presentation, so it isn't a rational basis for privileging some answers over others.
The contracts are denominated in USD, and they pay in that currency. But you trade on margin, and the collateral can be in any currency (crypto or fiat). In your example, you get back the BTC plus 15% of what that BTC was worth in USD when you made the trade.
Incidentally, TRUMPFEB is now trading at 0.16 (i.e. implied 16% chance that Trump is president next February). This looks insane to me (and I have bet accordingly). I'd be curious if you or others have further thoughts on what might be going on.
I'm not sure I understand your argument, given that FTX allows traders to keep balances in both USD and BTC, but in any case historically FTX prices have been in line with Betfair/PredictIt prices, so I doubt this consideration is relevant.
I'm too lazy to look it up, but I did research this a couple of weeks ago and found that 538 had indeed outperformed the markets both in 2008 and 2016 (I wasn't able to find data for 2012). This is not very informative, though, since it's just a couple of cases. Much better is to look at the state-level predictions and use brier scores as a measure of forecasting performance.
man do I wish that there was just one large market with minimal fees
Such a market exists, though unfortunately it is restricted to countries where most LW users are not citizens of.
Only the first article in the comment is by Silver, on whose expertise the original poster is basing his recommendation. That article doesn't discuss mail-in ballots or voter suppression, and in fact his main point is that the time remaining until election day (almost three months when the article was written) combined with uncertainties due to Covid-19 meant that the race was still open back then. Those considerations have much more limited force at present, when only 16 days remain, and Biden's lead has widened considerably.
If you've been at all listening to Silver recently, you'll know that he thinks his model probably underestimates Biden's chances. This shouldn't be surprising, since as Silver acknowledges, in this new version of the model he has made a special effort to build conservative assumptions into it.
In any case, I would encourage people hesitant to bet for Biden to resist the temptation of "throwing in a bunch of considerations" for why the models may be wrong, and instead try to calculate what the correct forecast should be in light of those considerations. For example, if you think mail-in ballots will be a big factor, try to estimate the magnitude of this effect.
Following my own advice, I just built a simple Guesstimate model of the impact of mail voting on the popular vote. I created the model very quickly, so if anyone spots any errors, please mention them below. And if you think some of the parameters should be different, simply copy the model and adjust those parameters to your satisfaction. Note that the effect of "rejected" in-person ballots is not modeled. This effect favors Biden, since a greater proportion of Trump votes will be in person, and hence susceptible to being "rejected" (i.e., not cast due to failure to bring an ID, long lines, inability to find a polling station, etc).
ETA: The upshot of the model is that mail voting shrinks the expect popular vote gap between Biden and Trump by about 2%. If we assume that the electoral college gives Trump a ~2% popular vote advantage, the model implies a drop in Biden's chances of winning the election from 87% to about 79%. [I modified the model and improved some of the estimates, and now the effect is less than 1%.]
(Disclosure: I have bet a total of USD 12k on Biden, mostly back when his odds where roughly equal with Trump's.)
The book is dedicated "for Peter, who convinced me". Maybe that mysterious Peter is the ultimate cause of Christian's interest in Al alignment and his decision to write a book about it?
Makes sense! Thanks for the explanation.
a frustratingly well-paywalled, yet exhaustive, complete and informative overview of the IARPA's FOCUS tournament
Since you quote from a section that is behind the paywall, I assume you have access to the article. If so, could you make it available? Or just send it to me (name@surname.com) and I'll upload it to my site and post a link to it here and on LW. Thanks!
Thanks for writing this—just a couple of days ago I thought it might be a good idea to get food pedals.
Since you use Karabiner, have you considered using goku to create "complex modifications"? It might help you make your keyboard more ergonomic and hence ease your wrist pain. I personally like to use the spacebar as a modifier key, and control the arrow keys with spacebar-j / k / l / i. You can also set spacebar-a / s / d / f to delete letter/word forward/backward. I actually have hundreds of modifications, but these are amongst the most useful.
Also, you may already know this, but just in case: on Gmail, you can enable 'auto-advance' under preferences/advanced, and then use 'e' instead of '[', which is easier to reach on the keyboard (so perhaps that pedal is best used for some other function).
Acausalism.
Also it is good that you had here an example of something that a lot of people would view as a negative case (making the invention of the hydrogen bomb faster).
There's also the example of a work without which the Russian Revolution, and the subsequent deaths of tens of millions of people in famines and mass killings, may not have occurred. But until you mentioned it, I hadn't realized that fiction appears to be more often credited with having a positive than a negative influence, whereas for philosophy the reverse seems to be the case. Would be interesting to move beyond impressions and come up with a more rigorous way of testing this.
Some examples (I'm considering fiction generally and not just written fiction):
- The film The Day After was seen by 100 million Americans and was instrumental in changing Reagan’s nuclear policy.
- «President Ronald Reagan watched the film several days before its screening, on November 5, 1983. He wrote in his diary that the film was "very effective and left me greatly depressed," and that it changed his mind on the prevailing policy on a "nuclear war". The film was also screened for the Joint Chiefs of Staff. A government advisor who attended the screening, a friend of Meyer's, told him "If you wanted to draw blood, you did it. Those guys sat there like they were turned to stone." Four years later, the Intermediate-Range Nuclear Forces Treaty was signed and in Reagan's memoirs he drew a direct line from the film to the signing.» (Wikipedia)
- «Director Meyer and writer Hume produced The Day After to support nuclear disarmament with the ‘grandiose notion that this movie would unseat Ronald Reagan’, and the nuclear freeze groups heavily exploited the ABC movie as a propaganda.» (Hänni, A chance for a propaganda coup?)
- Arthur Koestler’s Darkness at Noon deeply influenced Edward Teller, whose views about the Soviet Union were central in his efforts to persuade the US to develop the hydrogen bomb.
- The short documentary film If you love this planet influenced Canadian PM Pierre Trudeau.
- The mini-series Holocaust motivated the abolition of the statute of limitations for war crimes in Germany:
- «In 1978 the major breakthrough into general consciousness of US citizens came with the showing on prime-time television of a four-part series simply entitled Holocaust, which was watched by nearly 100 million Americans. The fictional drama that followed the lives of a Jewish family, exposed to the full horrors of the Holocaust, and an SS man who rose to a leading position in the implementation of the extermination programme, captured the imagination in ways that scholarly literature could never do. Jewish organizations maximized the subsequent publicity opportunities presented by the success of the series to spread awareness of the Holocaust still further, both in Jewish and non-Jewish communities.
In West Germany a year later the showing of the series was a sensation. Holocaust was watched by around 20 million viewers (around half of the West German viewing population), who were transfixed by the personalized and highly emotional dramatic depiction of persecution and extermination. People empathized with the victims and recognized the monumentality of the crime as they had never done before. ‘A nation is shocked’ was the verdict of one scholarly analysis of the impact of the film.“ ‘Holocaust has shaken up post-Hitler Germany in a way that German intellectuals have been unable to do,’ commented the widely read weekly Der Spiegel. More than three decades after the end of the war an American film, criticized by some as reducing the destruction of the Jews to the level of a ‘soap opera’, had opened up the sense of national guilt. The following year the Federal Parliament (the Bundestag) abolished the statute of limitations on war crimes, permitting further legal prosecution of perpetrators of the Holocaust. The film was widely seen as playing a significant role in the decision.» (Ian Kershaw, The Global Age, ch. 8) - Nikolai Chernyshevsky's What is to be done had a more profound influence on Lenin than even Marx's Kapital, and is plausibly a causal antecedent to the Russian Revolution.
To reasonably conclude that PredictIt's limits are "limits of prediction markets"—as your title asserts—you need to show either that the other existing prediction markets also exhibit these limits, or that there is a fundamental theoretical reason for expecting such limits to be exhibited by any prediction market. As far as I can tell, you do neither. (You do say that «similar analysis is applicable to any [prediction market]», but you never justify this assertion. In fact, of the six problems you note, I think the only one that may be plausibly claimed to be inherent to prediction markets is #4, and even that one may be potentially solvable.)
Of course genetics isn't everything. This is recognized in the third law of behavioral genetics. Researchers who rely on twin studies do not assume otherwise.
The post addresses this worry:
you might worry about a correlation/causation problem with that kind of statement. However, there have also been several twin studies that help eliminate this bias.
There is, however, another worry unaddressed by those studies, which wolajacy raises in their comment. This is the debate between the 'human capital' and 'signaling' theories of education, covered extensively in Bryan Caplan's book, The case against education. Even if years of education cause—rather than correlate with—increased quality of life and length of life for individual people, reducing years of education for the population as a whole may not reduce those measures much if signaling is the main causal mechanism.
Thanks for answering my question. I'd personally assign a ~5% chance [EDIT: on reflection, perhaps closer to 10%] to that hypothesis. If you can think of a way to operationalize our disagreement, I'd be interested in arranging a bet.
The one that seems most likely to me is Pinker preemptively canceling himself to inoculate against future attempts. I don't think it's outlandish. And I think it is quite possible that Pinker has some Machiavelli in him.
What's your credence in this hypothesis?
That would be because they disagree with the consensus in EA about what constitutes 'the most impactful,' 'the greatest welfare,' and/or 'rigorous reasoning.'
I said that the belief must be reached from welfarist premises and rigorous reasoning, not from what the organization believes are welfarist premises and rigorous reasoning.
If they were sufficient, any NPO that could identify as an EA-aligned organization would do so.
I'm not sure what you mean by this. And it seems clear to me that lots of nonprofit orgs would not classify as EA orgs given my proposed criterion (note the clarification above).
I think it would be correct to classify it entirely as an x-risk org and not as an EA org. I don't think it does any EA-style analysis of what it should work on that is not captured under x-risk analysis, and I think that people working to do things like, say, fight factory farming, should never expect support from BERI (via the direct work BERI does).
I think it's worth noting that an org can be an EA org even if it focuses exclusively on one cause area, such as x-risk reduction. What seems to matter is (1) that such a focus was chosen because interventions in that area are believed to be the most impactful, and (2) that this belief was reached from (a) welfarist premises and (b) rigorous reasoning of the sort one generally associates with EA.
Another example I just discovered: Wikipedia classifies Quillette as an unreliable source; by contrast, Vox, The Nation, Mother Jones are all considered reliable sources. I don't often read Quillette, but my sense is that a criterion that generates this classification can't be defended as unbiased.
Whether a source is classified as reliable or unreliable can shape the content of Wikipedia articles in major ways, because only statements backed up by sources deemed reliable are admissible. If the list of reliable sources is skewed in a particular direction, so will be the articles.
There's also a similar question on Polymarket, a new prediction market. (Note that the Metaculus question is conditional on the NYT publishing a story on Scott, whereas the Polymarket is unconditional.)
When Eliezer and others talk about "civilizational inadequacy", they generally refer to something much broader than the United States. Eliezer mentions the example of Japan's monetary policy, for instance. He also contrasts the civilizational inadequacy thesis with "the view that in general, on most issues, the average opinion of humanity will be a better and less biased guide to the truth than my own judgment." (emphasis added) And he relies (I think) on that thesis to draw conclusions about how he expects humanity to handle AI risk; such an inference wouldn't work if the thesis was restricted to the prevailing culture or institutions of a particular country. At the very least, if usage deviates from this established meaning I think this should be made clear.
I'm not sure I understand your response. Yes, the series of updates is clearly focused on the United States, but your claim is that "civilization" explains why the US handled Covid-19 so poorly. Since human civilization is a factor present in all countries in the world, the fact that other countries handled Covid-19 very differently constitutes evidence against the "civilizational inadequacy" hypothesis. That your post wasn't focused on these other countries seems irrelevant.
Thanks. I'm not sure how to implement those suggestions on Wordpress, but I'll add them to the list of possible improvements.
I maintain EA Blogs. Although this is tangential to the original question, if anyone has suggestions on either additional blogs to include or possible general improvements, please contact me or leave a comment below.
Metaculus question: Will the New York Times doxx Scott Alexander?
As a side note, I strongly recommend the uBlacklist extension Mati mentions for preventing toxic websites from appearing on your search results (e.g. a certain "rational" wiki that writes cruel stuff about people they dislike).
Great post.
Gwern points to expert opinion that visual thinking ability might be second only to IQ in terms of intellectual importance.
Do you happen to remember the source for this? A quick Google search didn't help.
EDIT: Found it (I think):
an emphasis on spatial reasoning & visualization ability was one of the reasons behind SMPY choosing to use SAT-M for screening, as one of the theses is that, after general intelligence, visuospatial reasoning (as opposed to the more academically-prized glibness & verbal ability) may be the next most important requirement for major STEM achievement.
This is my anecdotal impression as a long-time Wikipedia editor (I started contributing in 2003). I can't offer concrete evidence other than my testimony, because this impression was formed in the course of observing subtle instances of bias on countless occasions, rather than encountering any one egregious incident. (Though on reflection I can cite two not-so-subtle examples illustrative of the phenomenon I have in mind: first, the labelling of cryonics as "quackery"; and secondly, the blacklisting of Econlib.)
The biases I noticed are in the left-wing and "skeptical" directions (meaning by the latter something like what Eliezer calls "traditional rationality", as opposed to the "Bayesian rationality" folks in the rationalist and EA communities generally endorse). Think of it as Wikipedia moving slightly in the direction of RationalWiki.
I just found that such a resource exists for philosophy, in case it is of interest to others: Meta-Encyclopedia of Philosophy.
On reflection, I guess it makes sense that the philosophers would be among the first to "go meta".
Red Pen Reviews does something like this for health and nutrition books. I am not aware of similar initiatives for books in other scientific fields.
I like FiveThirtyEight, but it's not the sort of publication you can refer to for "what happened in the past 3 days" (except for very specific events like 'how much Trump's popularity changed in the intervening period').
I second the Financial Times recommendation.
This argument has long been known, and much discussed. See e.g. Jacob Ross, Rejecting ethical deflationism and William MacAskill, The infectiousness of nihilism.
Recurring Freedom session across all my devices (laptop, phone, tablet) set to disable all apps and most websites (including messaging, news, and discussion sites) 30 minutes before bedtime every night.