All of rebellionkid's Comments + Replies

Harry Potter and the Methods of Rationality discussion thread, February 2015, chapter 104

Make the game end when the clock runs over, or when the snitch is caught, whichever is sooner. And make the snitch worth ~half the typical point spread of a match.

4Izeinwinter6yNope. Make it worth zero points. This keeps it's importance, but changes the job of the seeker. Behind, "Run interference", ahead, "catch snitch". But mostly, having one is still daft.
Why the tails come apart

Fantastic, I wish I'd had this back when almost everyone in LW/EA circles I met was reading the biography of everyone in the ' fortune 400 and trying to spot the common factors. A surprisingly common strategy that's likely not to work for exactly these reasons.

Why Are Individual IQ Differences OK?

Notice I said nothing at all about racism or our policy responses to race. Of course intra-group variation is more important, that's obvious and applies to height too. This much is well known and irrelevant to my point.

The thing I'm interested here is why it's commonly accepted that there ought (in a strong moral sense) to be no correlation. Not our response to the actual existence of that correlation.

Why Are Individual IQ Differences OK?

I hope it is false.

I think this is the most interesting sentence in the whole discussion.

Let's be clear. Racial groupings are really very significant pieces of evidence. There's huge amounts of genetics that correlates, huge amounts of culture that correlates, huge amounts of wider environment that correlates. It would be frankly astonishing if things like IQ, reaction time, hight, life expectancy, and rates of disease didn't also correlate.

So, we ought to expect to see a correlation, and in fact a whole bunch of studies say we do. ... And then those s... (read more)

3Epiphany8yCulture and environment are not race. Therefore, if you're studying race, those influences should be taken out of your scientific experiment. It's extremely difficult to remove things like culture and environment from a study on IQ. The fact that so much is correlated with it doesn't mean the results of studies intended to determine racial differences are significant so much as it means they're a tangled mess of cause and effect which we likely haven't sorted out adequately. A. We don't want black people to suffer needlessly. B. We don't want to encourage ourselves and others to be prejudiced against people when, regardless of what the average African's IQ is, it is still both logically incorrect (hasty generalization) and ethically wrong to prejudge individual Africans. However, knowing how humans behave, we figure that if people believe Africans have lower IQs, that will result in an increase in prejudice. Actually, I bet some people are not happy saying that there are correlations there. This is one of those notions you might want to double check.
0private_messaging8yWell, there's a correlation between race and height, too, no doubt, but such correlation is utterly insignificant comparing to variance within either race - you don't say whites are taller or blacks are taller, there's very short black populations and very tall ones, and ditto for the whites whose variance is smaller. The quantitative differences make a qualitative difference here. Racists believe the correlation to be of greater significance than that of correlation between the height and intelligence. Based on fairly poor evidence - Raven's matrices are not this culture fair, they're culture fair in the sense that you can test British, Germans, French, Russian, and Chinese and Japanese with it, not in the sense that you can go and test some tribe that doesn't do much arithmetic nor is exposed to similar visual stimuli. By the way given the diversity of blacks it would be utterly surprising if there is not a single ethnicity there with an average IQ greater than 100, as well as IQ with greater or smaller variance. (I would expect blacks to have larger variance than whites because they're plain more diverse, and mixed to have greater variance still) Then the "rational" racists also object to use or even the existence of correlation between such racism and intelligence, conscientiousness, education, and other factors.
Harry Potter and the Methods of Rationality discussion thread, part 25, chapter 96

Thanks for that. Sorry, should have explained my meaning better. I was looking for a clue of the form "the chapter translates it as 'defeated' but it actually means 'banished'" or similar. Along the lines of Harry's massive mistranslation of 'nihil supernum'.

Harry Potter and the Methods of Rationality discussion thread, part 25, chapter 96

Ok, I'm now frustrated and bored by online translators. Can someone give me a hand and translate that? I get some of it, but never enough to actually check meaning properly.

3ArisKatsaris8yThe chapter itself provides the translation in the line immediately following: "Three shall be Peverell's sons and three their devices by which Death shall be defeated."
CEA does not seem to be credibly high impact

Can I just make clear my role here. 1) I've had general conversation with Jonathan about CEA and MIRI, in which several of these criticisms were raised. 2) I checked over the numbers for the GWWC impact assessment. 3) I've also said that criticism in public is often productive, and that public scrutiny of CEA on LW would be helpful for people choosing between it and MIRI. 4) I saw a draft of the post before it was published.

I want to make it very clear that: 1) I do not endorse this post. 2) I did not do detailed fact-checking for it. 3) I do not want t... (read more)


Is it obvious that the second attitude would be terribly much harder to achieve? By which I mean is it clear that the cost of becoming sufficiently conservative is higher than the cost of the pain?

2Nisan9yCertainly not. On the contrary, successfully cultivating that sense of personal responsibility seems like a rewarding activity on its own. In fact, the only way of acquiring pain asymbolia that I've heard of involves treating pain as a warning signal and taking it seriously. So it seems like there are only benefits and no downsides to acquiring pain asymbolia, as long as you aren't cruel to yourself.
Pre-commitment and meta at the Cambridge UK meetup

Have updated the post, sorry for misreading your writing.

2Paul Crowley9yI am amazed you were able to make as much sense of it as you were! Thank you for posting this!
Optimizing Sleep

Update. I've discovered I dont function nearly so well at night as I had anticipated. I was running with core sleep 4-8, after shifting this to 2-6 I can see an improvement.

Also notable features: I can shift the naps around a fair amount, but too much core sleep, or hitting the snooze button after waking from core sleep throws me off for a long time after. The amount of sleep in each session doesn't seem as important in how I feel afterwards as how I wake up. Waking up in the middle of deep sleep going blurg and snoozing for 10 mins is followed by feeling... (read more)

Meetup : Cambridge UK

Sorry, trying to be cunning and failing.

Sorry for confusion. Mistyped the date. The meeting is on Sunday not Saturday.

Dutch Books and Decision Theory: An Introduction to a Long Conversation

I'd always thought "What bets do I take" was the justification for Bayesian epistemology. Every policy decision (every decision of any kind) is a statement of the form "I'm prepared to accept these costs to receive these outcomes given these events", this is a bet. If Bayesian epistemology lets you win bets then that's all the justification it could ever need.

0Jack9yThe above discussion about "what bets do I take?" is about literal, monetary wager-making. The sense in which any decision can be described in a way that is equivalent to such a wager is precisely the question being discussed here.
Optimizing Sleep

At the moment I'm tracking sleep with a pen and paper, just started with an Android app. Zeo was totally unknown to me. Thankyou, will investigate this.

My plan is to go by the metric that matters to me, university work. I get enough data there to notice something going wrong. Not the same detail as proper Anki-style benchmarking, however there is an extra variable in my case (resulting in high variance on the day/week scale), that would spoil such results.

0gwern9yI'm a big fan of the Zeo [], I should mention. You may be better off with one of the cellphone apps (I think some are even free, while the cellphone-based Zeo is like $100, plus something like $20-40 in headband replacement a year depending on how fast you use them up).
Optimizing Sleep

I'm starting an experiment today. The program is Everyman: 3 20 min naps at midday, the afternoon and the evening with core sleep gradually decreasing, hopefully to a small number of hours. I have early morning lectures and function well at night, so short of being nocturnal decreasing my core sleep looks the best way to be efficient. Will keep this tread updated with results.

0rebellionkid9yUpdate. I've discovered I dont function nearly so well at night as I had anticipated. I was running with core sleep 4-8, after shifting this to 2-6 I can see an improvement. Also notable features: I can shift the naps around a fair amount, but too much core sleep, or hitting the snooze button after waking from core sleep throws me off for a long time after. The amount of sleep in each session doesn't seem as important in how I feel afterwards as how I wake up. Waking up in the middle of deep sleep going blurg and snoozing for 10 mins is followed by feeling awful. Being in light enough sleep that my phone screen turning back on before the alarm goes off is enough to wake me and then getting straight out of bed is followed by feeling really energetic. It's incredible how much sleep I'm able to get in the naps. Sometimes schedules mean I can only really afford 10/15 mins, and I feel myself waking up from real sleep after this.
2gwern9yAre you tracking your sleep with a Zeo or at least a cellphone accelerometer? How are you benchmarking your mental state? Memory may take a serious hit, are you using something like Anki which records statistics on forgetting/memory performance? (Working memory and executive function are also issues, so I would suggest dual n-back.)
Meetup : Cambridge UK

There were certainly enough people interested to make this regular. To get involved if you weren't here today join the google group at

Meetup : Cambridge UK

Hopefully future meetups will be on a regular basis so they can be advertised in good time.

Archimedes's Chronophone

Archimedes died in 212 BC. Justinian closed the pagan schools and killed of the classical research project in 529 AD. The first book of the Quran was delivered in 610 AD. So sharing with the Islamic scientists (who had all read Aristotle anyway) doesn't make sense.

But an expansion of the communication between places like Alexandria, Athens, Cos etc could be a really good plan.

Archimedes's Chronophone

My understanding is not that he "didn't think to" run experiments, rather he actively rejected the doing of experiments. The idea is that one can study nature, study things as they naturally are, because that is them acting in accordance with their inherent properties. But it makes no sense to study things in a lab, in an artificial environment. If I launch something out of a cannon then I disturb it from how it inherently acts. Of course something acting against how it inherently acts is pointless to study. Likewise doing any kind of experiment will only tell you about your experiment (at best) and tell you nothing of the natural state of things.

Einstein's Arrogance

Just to point out what may be a nitpick or a clarification. It's perfectly possible for infinity many positive things to sum to a finite number. 1/2+1/4+1/8+...=1.

There can be infinitely many potential murderers. But if the probability of each having done it drops off fast enough you can avoid anything that is literally infinitesimal. Almost all will be less than 1/3^^^^^^3 of course, but that's a perfectly well defined number you know how to do maths with.

-1Daemon8yHate to nitpick myself, but 1/2+1/4+1/8+... diverges (e.g., by the harmonic series test). Sum 1/n^2 = 1/4 + 1/9 + ... = (pi^2)/6 is a more fitting example. An interesting question, in this context, is what it would mean for infinitely many possibilities to exist in a "finite space about any point that can be reached at sub-speed of light times." Would it be possible under the assumption of a discrete universe (a universe decomposable no further than the smallest, indivisible pieces)? This is an issue we don't have to worry about in dealing with the infinite sums of numbers that converge to a finite number.
Pascal's Mugging: Tiny Probabilities of Vast Utilities

Enormously higher probability is not 1. This still doesn't mean the statement is zero evidence.

2dlthomas9yI don't know - if he did actually have reality breaking powers, he would likely be tempted to put them to more effective use. If he would in fact be less likely to be making the statement were it true, then it is evidence against, not evidence for, the truth of his statement.
Newcomb's Problem and Regret of Rationality

An amusing n=3 survey of mathematics undergrads at Trinity Cambridge:

1) Refused to answer. 2) It depends on how reliable Omega is/but you cant (shouldn't) really quantify ethics anyway/this situation is unreasonable. 3) Obviously 2 box, one boxing is insane.

3 said he would program an AI to one box. And when I pointed out that his brain was built of quarks just like the AI he responded that in that case free will didn't exist and choice was impossible.

Harry Potter and the Methods of Rationality discussion thread, part 9

A lot of the "hate" between colleges is perpetuated in the same way as most Cambridge traditions. Freshers read on wikipedia what Cambridge traditions are and then emulate them. There is belief in belief that people from St Johns are horrible but not much more.

1Sarokrae10yThat's pretty much the way Hogwarts works, right?
4Eugine_Nier10yWell, these traditions somehow propagated before the internet.