## Posts

## Comments

**bigjeff5**on Arguments for optimism on AI Alignment (I don't endorse this version, will reupload a new version soon.) · 2023-10-17T20:32:47.490Z · LW · GW

This is exactly right. To put it more succinctly: Memory corruption is a known vector for exploitation, therefore any bug that potentially leads to memory corruption also has the potential to be a security vulnerability. Thus memory corruption should be treated with similar care as a security vulnerability.

**bigjeff5**on Arguments for optimism on AI Alignment (I don't endorse this version, will reupload a new version soon.) · 2023-10-17T20:27:30.435Z · LW · GW

The reason Person A in scenario 2 has the intuition that Person B is very wrong is because there are dozens, if not hundreds of examples where people claimed no vulnerabilities and were proven wrong. Usually spectacularly so, and often nearly immediately. Consider the fact that the most robust software developed by the most wealthy and highly motivated companies in the world, who employ vast teams of talented software engineers, have monthly patch schedules to fix their constant stream vulnerabilities, and I think it's pretty easy to immediately discount anybody's claim of software perfection without requiring any further evidence.

All the evidence Person A needs is the complete and utter lack of anybody having achieved such a thing in the history of software to discount Person B's claims.

I've never heard of an equivalent example for AI. It just seems to me like Scenario 2 doesn't apply, or at least it cannot apply at this point in time. Maybe in 50 years we'll have the vast swath of utter failures to point to, and thus a valid intuition against someone's 9-9's confidence of success, but we don't have that now. Otherwise people would be pointing out examples in these arguments instead of vague unease regarding problem spaces.

**bigjeff5**on Markets are Anti-Inductive · 2015-01-21T04:50:56.056Z · LW · GW

That was four years ago, but I'm pretty sure I was using hyperbole. Pros don't bluff often, and when they do they are only expecting to break even, but I doubt it's as low as 2% (the bluff will fail half the time).

I'd also put in a caveat that the best hand wins among hands that make it all the way to the river. There are plenty of times where a horrible hand like a 6 2, which is an instant fold if you respect the skills of your fellow players, ends up hitting a straight by the river and being the best hand but obviously didn't win. Certainly more often than 1%, and there are plenty of better hands that you still almost always fold pre-flop that are going to hit more often.

So, at best it was poorly stated (i.e. hyperbole without saying so), at worst it's just wrong.

**bigjeff5**on Rationality Quotes July 2014 · 2014-07-10T17:16:52.846Z · LW · GW

Pretty much.

**bigjeff5**on Rationality Quotes July 2014 · 2014-07-10T16:41:12.310Z · LW · GW

At this point you have to ask what you mean by "theory" and "learning".

The original method of learning was "those that did it right didn't die" - i.e. natural selection. Those that didn't die have a pattern of behavior (thanks to a random mutation) that didn't exist in previous generations, which makes them more successful gene spreaders, which passes that information on to future generations.

There is nothing in there that requires one to ask any questions at all. However, considering that there is information gained based on past experience, I think the definition of learning could be stretched to cover it. Obviously there is no individual learning, but there is definitely species learning going on there.

Since the vast majority of creatures that use this method of learning as their primary method of learning don't even have brains, it seems obvious that there is no theory there. However, if we stretch the definition of theory to include any pattern of information that attempts to reflect reality (regardless of how well it does that job), well then even the lowliest bacteria have theories about how their world is supposed to work, and act accordingly.

That same broader definition of "theory" would cover wedrifid's theoryless algorithms as well, as all you care about are patterns of information attempting to reflect reality, and they certainly have those.

All that said, the point of the quote is that in order for you as an individual to learn, then you as an individual must have an underlying theory of how things are supposed to be that can be challenged when faced with reality, in order to learn.

I have no idea if it's actually true, I'm no psychologist or human learning expert or anything even remotely related, but it sounds like it has to be true even under the strict sense. It seems like it's practically a tautology to me. Even wedfrid's algorithms have a starting framework that attempts to reflect reality, however simplistic it may be. The algorithm itself is the theory there; it didn't come from nothing.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T23:29:33.679Z · LW · GW

I see that now, it took a LOT for me to get it for some reason.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T23:29:02.698Z · LW · GW

Wow.

I've seen that same explanation at least five times and it didn't click until just now. You can't distinguish between the two on tuesday, so you can only count it once for the pair.

Which means the article I said was wrong was absolutely right, and if you were told that, say one boy was born on January 17th, the chances of both being born on the same day are 1-(364/365)^2 (ignoring leap years), which gives a final probability of roughly 49.46% that both are boys.

Thanks for your patience!

ETA: I also think I see where I'm going wrong with the terminology - sampling vs not sampling, but I'm not 100% there yet.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T22:52:59.744Z · LW · GW

How can that be? There is a 1/7 chance that one of the two is born on Tuesday, and there is a 1/7 chance that the other is born on Tuesday. 1/7 + 1/7 is 2/7.

There is also a 1/49 chance that both are born on tuesday, but how does that subtract from the other two numbers? It doesn't change the probability that either of them are born on Tuesday, and both of those probabilities add.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T21:54:24.913Z · LW · GW

This statement leads me to believe you are still confused. Do you agree that if I know a family has two kids, I knock on the door and a boy answers and says "I was born on a Tuesday," that the probability of the second kid being a girl is 1/2? And in this case, Tuesday is irrelevant? (This the wikipedia called "sampling")

I agree with this.

Do you agree that if, instead, the parents give you the information "one of my two kids is a boy born on a Tuesday", that this is a different sort of information, information about the set of their children, and not about a specific child?

I agree with this if they said something along the lines of "One and only one of them was born on Tuesday". If not, I don't see how the Boy(tu)/Boy(tu) configuration has the same probability as the others, because it's twice as likely as the other two configurations that that is the configuration they are talking about when they say "One was born on Tuesday".

Here's my breakdown with 1000 families, to try to make it clear what I mean:

1000 Families with two children, 750 have boys.

Of the 750, 500 have one boy and one girl. Of these 500, 1/7, or roughly 71 have a boy born on Tuesday.

Of the 750, 250 have two boys. Of these 250, 2/7, or roughly 71 have a boy born on Tuesday.

71 = 71, so it's equally likely that there are two boys as there are a boy and a girl.

Having two boys doubles the probability that one boy was born on Tuesday compared to having just one boy.

And I don't think I'm confused about the sampling, because I didn't use the sampling reasoning to get my result*, but I'm not super confident about that so if I am just keep giving me numbers and hopefully it will click.

*I mean in the previous post, not specifically this post.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T18:54:39.358Z · LW · GW

The answer I'm supporting is based on flat priors, not sampling. I'm saying there are two possible Boy/Boy combinations, not one, and therefore it takes up half the probability space, not 1/3.

Sampling to the "Boy on Tuesday" problem gives roughly 48% (as per the original article), not 50%.

We are simply told that the man has a boy who was born on tuesday. We aren't told how he chose that boy, whether he's older or younger, etc. Therefore we have four possibilites, like I outlined above.

Is my analysis that the possibilities are Boy (Tu) /Girl, Girl / Boy (Tu), Boy (Tu)/Boy, Boy/Boy (Tu) correct?

If so, is not the probability for some combination of Boy/Boy 1/2? If not, why not? I don't see it.

BTW, contrary to my previous posts, having the information about the boy born on Tuesday is critical because it allows us (and in fact requires us) to distinguish between the two boys.

That was in fact the point of the original article, which I now disagree with significantly less. In fact, I agree with the major premise that the tuesday information pushes the odds of Boy/Boy closer 50%, I just disagree that you can't reason that it pushes it to exactly 50%.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T18:11:39.039Z · LW · GW

For the record, I'm sure this is frustrating as all getout for you, but this whole argument has really clarified things for me, even though I still think I'm right about which question we are answering.

Many of my arguments in previous posts are wrong (or at least incomplete and a bit naive), and it didn't click until the last post or two.

Like I said, I still think I'm right, but not because my prior analysis was any good. The 1/3 case was a major hole in my reasoning. I'm happily waiting to see if you're going to destroy my latest analysis, but I think it is pretty solid.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T17:48:50.619Z · LW · GW

Yes, and we are dealing with the second question here.

Is that not what I said before?

We don't have 1000 families with two children, from which we've selected all families that have at least one boy (which gives 1/3 probability). We have one family with two children. Then we are told one of the children is a boy, and given zero other information. The probability that the second is a boy is 1/2, so the probability that both are boys is 1/2.

The possible options for the "Boy born on Tuesday" are not Boy/Girl, Girl/Boy, Boy/Boy. That would be the case in the selection of 1000 families above.

The possible options are Boy (Tu) / Girl, Girl / Boy (Tu), Boy (Tu) / Boy, Boy / Boy (Tu).

There are two Boy/Boy combinations, not one. You don't have enough information to throw one of them out.

This is NOT a case of sampling.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T17:34:39.953Z · LW · GW

Yeah, probably the biggest thing I don't like about this particular question is that the answer depends entirely upon unstated assumptions, but at the same time it clearly illustrates how important it is to be specific.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T17:23:58.789Z · LW · GW

The relevant quote from the Wiki:

The paradox arises because the second assumption is somewhat artificial, and when describing the problem in an actual setting things get a bit sticky. Just how do we know that "at least" one is a boy? One description of the problem states that we look into a window, see only one child and it is a boy. This sounds like the same assumption. However, this one is equivalent to "sampling" the distribution (i.e. removing one child from the urn, ascertaining that it is a boy, then replacing). Let's call the statement "the sample is a boy" proposition "b". Now we have: P(BB|b) = P(b|BB)

P(BB) / P(b) = 11/4 / 1/2 = 1/2. The difference here is the P(b), which is just the probability of drawing a boy from all possible cases (i.e. without the "at least"), which is clearly 0.5. The Bayesian analysis generalizes easily to the case in which we relax the 50/50 population assumption. If we have no information about the populations then we assume a "flat prior", i.e. P(GG) = P(BB) = P(G.B) = 1/3. In this case the "at least" assumption produces the result P(BB|B) = 1/2, and the sampling assumption produces P(BB|b) = 2/3, a result also derivable from the Rule of Succession.

We have no general population information here. We have one man with at least one boy.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T17:17:40.638Z · LW · GW

Re-read it.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T17:08:17.540Z · LW · GW

http://en.wikipedia.org/wiki/Boy_or_Girl_paradox

I know it's not the be all end all, but it's generally reliable on these types of questions, and it gives P = 1/2, so I'm not the one disagreeing with the standard result here.

Do the math yourself, it's pretty clear.

Edit: Reading closer, I should say that both answers are right, and the probability can be either 1/2 or 1/3 depending on your assumptions. However, the problem as stated falls best to me in the 1/2 set of assumptions. You are told one child is a boy and given no other information, so the only probability left for the second child is a 50% chance for boy.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T17:04:52.994Z · LW · GW

How is it different? In both cases I have two independent coin flips that have absolutely no relation to each other. How does knowing which of the two came up heads make any difference at all for the probability of the other coin?

If it was the first coin that came up heads, TT and TH are off the table and only HH and HT are possible. If the second coin came up heads then HT and TT would be off the table and only TH and HH are possible.

The total probability mass of some combination of T and H (either HT or TH) starts at 50% for both flips combined. Once you know one of them is heads, that probability mass for the whole problem is cut in half, because one of your flips is now 100% heads and 0% tails. It doesn't matter that you don't know which is which, one flip doesn't have any influence on the probability of the other. Since you already have one heads at 100%, the entire probability of the remainder of the problem rests on the second coin, which is a 50/50 split between heads and tails. If heads, HH is true. If tails, HT is true (or TH, but you don't get both of them!).

Tell me how knowing one of the coins is heads changes the probability of the second flip from 50% to 33%. It's a fair coin, it stays 50%.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T16:50:05.790Z · LW · GW

No, it's the exact same question, only the labels are different.

The probability that any one child is boy is 50%. We have been told that one child is a boy, which only leaves two options - HH and HT. If TH were still available, then so would TT be available because the next flip could be revealed to be tails.

Here's the probability in bayesian:

P(BoyBoy) = 0.25 P(Boy) = 0.5 P(Boy|BoyBoy) = 1

P(BoyBoy|Boy) = P(Boy|BoyBoy)*P(BoyBoy)/P(Boy)

P(BoyBoy|Boy)= (1*0.25) / 0.5 = 0.25 / 0.5 = 0.5

P(BoyBoy|Boy) = 0.5

It's exactly the same as the coin flip, because the probability is 50% - the same as a coin flip. This isn't the monty hall problem. Knowing half the problem (that there's at least one boy) doesn't change the probability of the other boy, it just changes what our possibilities are.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T10:12:57.450Z · LW · GW

Lets add a time delay to hopefully finally illustrate the point that one coin toss does not inform the other coin toss.

I have two coins. I flip the first one, and it comes up heads. Now I flip the second coin. What are the odds it will come up heads?

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T06:06:39.978Z · LW · GW

The only relevant information is that one of the children is a boy. There is still a 50% chance the second child is a boy and a 50% chance that the second child is a girl. Since you already know that one of the children is a boy, the posterior probability that they are both boys is 50%.

Rephrase it this way:

I have flipped two coins. One of the coins came up heads. What is the probability that both are heads?

Now, to see why Tuesday is irrelevant, I'll re-state it thusly:

I have flipped two coins. One I flipped on a Tuesday and it came up heads. What is the probability that both are heads?

The sex of one child has no influence on the sex of the other child, nor does the day on which either child was born influence the day any other child was born. There is a 1/7 chance that child 1 was born on each day of the week, and there is a 1/7 chance that child 2 was born on each day of the week. There is a 1/49 chance that both children will be born on any given day (1/7*1/7), for a 7/49 or 1/7 chance that both children will be born on the same day. That's your missing 1/7 chance that gets removed inappropriately from the Tuesday/Tuesday scenario.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T05:35:18.222Z · LW · GW

In Boy1Tu/Boy2Tuesday, the boy referred to as BTu in the original statement is boy 1, in Boy2Tu/Boy1Tuesday the boy referred to in the original statement is boy2.

That's why the "born on tuesday" is a red herring, and doesn't add any information. How could it?

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T04:55:28.167Z · LW · GW

I see my mistake, here's an updated breakdown:

Boy1Tu/Boy2Any

Boy1Tu/Boy2Monday Boy1Tu/Boy2Tuesday Boy1Tu/Boy2Wednesday Boy1Tu/Boy2Thursday Boy1Tu/Boy2Friday Boy1Tu/Boy2Saturday Boy1Tu/Boy2Sunday

Then the Boy1Any/Boy2Tu option:

Boy1Monday/Boy2Tu Boy1Tuesday/Boy2Tu Boy1Wednesday/Boy2Tu Boy1Thursday/Boy2Tu Boy1Friday/Boy2Tu Boy1Saturday/Boy2Tu Boy1Sunday/Boy2Tu

See 7 days for each set? They aren't interchangeable even though the label "boy" makes it seem like they are.

Do the Bayesian probabilities instead to verify, it comes out to 50% even.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T04:48:30.698Z · LW · GW

Which boy did I count twice?

Edit:

BAny/Boy1Tu in the above quote should be Boy2Any/Boy1Tu.

You could re-label boy1 and boy2 to be cat and dog and it won't change the probabilities - that would be CatTu/DogAny.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T04:16:49.832Z · LW · GW

No, read it again. It's confusing as all getout, which is why they make the mistake, but EACH child can be born on ANY day of the week. The boy on Tuesday is a red herring, he doesn't factor into the probability for what day the second child can be born on at all. The two boys are not the same boys, they are individuals and their probabilities are individual. Re-label them Boy1 and Boy2 to make it clearer:

Here is the breakdown for the Boy1Tu/Boy2Any option:

Boy1Tu/Boy2Monday Boy1Tu/Boy2Tuesday Boy1Tu/Boy2Wednesday Boy1Tu/Boy2Thursday Boy1Tu/Boy2Friday Boy1Tu/Boy2Saturday Boy1Tu/Boy2Sunday

Then the BAny/Boy1Tu option:

Boy2Monday/Boy1Tu Boy2Tuesday/Boy1Tu Boy2Wednesday/Boy1Tu Boy2Thursday/Boy1Tu Boy2Friday/Boy1Tu Boy2Saturday/Boy1Tu Boy2Sunday/Boy1Tu

Seven options for both. For some reason they claim either BTu/Tuesday isn't an option, or Tuesday/BTu isn't an option, but I see no reason for this. Each boy is an individual, and each boy has a 1/7 probability of being born on a given day. In attempting to avoid counting evidence twice you've skipped counting a piece of evidence at all! In the original statement, they never said one and ONLY one boy was born on Tuesday, just that one was born on Tuesday. That's where they screwed up - they've denied the second boy the option of being born on Tuesday for no good reason.

A key insight that should have triggered their intuition that their method was wrong was that they state that if you can find a trait rarer than being born on Tuesday, like say being born on the 27th of October, then you'll approach 50% probability. That is true because the actual probability is 50%.

**bigjeff5**on Bayes' Theorem Illustrated (My Way) · 2013-12-22T01:18:46.815Z · LW · GW

Just so it's clear, since it didn't seem super clear to me from the other comments, the solution to the Tuesday Boy problem given in that article is a really clever way to get the answer wrong.

The problem is the way they use the Tuesday information to confuse themselves. For some reason not stated in the problem anywhere, they assume that both boys cannot be born on Tuesday. I see no justification for this, as there is no natural justification for this, not even if they were born on the exact same day and not just the same day of the week! Twins exist! Using their same bizarre reasoning but adding the extra day they took out I get the correct answer of 50% (14/28), instead of the close but incorrect answer of 48% (13/27).

Using proper Bayesian updating from the prior probabilities of two children (25% boys, 50% one each, 25% girls) given the information that you have one boy, regardless of when he was born, gets you a 50% chance they're both boys. Since knowing only one of the sexes doesn't give any extra information regarding the probability of having one child of each sex, all of the probability for both being girls gets shifted to both being boys.

**bigjeff5**on The Bias You Didn't Expect · 2012-10-11T02:32:18.950Z · LW · GW

I'm talking about probability estimates. The actual probability of what happened is 1, because it is what happened. However, we don't know what happened, that's why we make a probability estimate in the first place!

Forcing yourself to commit to only one of two possibilities in the real world (which is what all of these analogies are supposed to tie back to), when there are a lot of initially low probability possibilities that are initially ignored (and rightly so), seems incredibly foolish.

Also, your analogy doesn't fit brazil84's murder example. What evidence does the lottery win give that allows us to adjust our probability estimate for how the gun was fired? I'm not sure where you're going with that, at all.

The real probability of however the bullet was fired is 100%. All we've been talking about are our probability estimates based on the limited evidence we have. They are necessarily incomplete. If new evidence makes both of our hypotheses less likely, then it's probably smart to check and see if a third hypotheses is now feasible, where it wasn't before.

**bigjeff5**on The Bias You Didn't Expect · 2012-10-10T21:34:49.971Z · LW · GW

The probability of both, in that case, plummets, and you should start looking at other explanations. Like, say, that the victim was shot with a rifle at close range, which only leaves a bullet in the body 1% of the time (or whatever).

It might be true that, between two hypotheses one is now more likely to be true than the other, but the probability for both still dropped, and your confidence in your pet hypothesis should still drop right along with its probability of being correct.

So say you have hypothesis X at 60% confidence and hypotheses Y at 40% New evidence comes along that shifts your confidence of X down to 20%, and Y down to 35%. Y didn't just "win". Y is now even more likely to be wrong than it was before the new evidence came it. The only substantive difference is that now X is probably wrong too. If you notice, there's 45% probability there we haven't accounted for. If this is all bound up in a single hypothesis Z, then *Z* is the one that is the most likely to be correct.

Contradictory evidence shouldn't make you *more* confident in your hypothesis.

**bigjeff5**on Firewalling the Optimal from the Rational · 2012-10-09T04:18:56.969Z · LW · GW

4-step is what preceded 2-step. I say preceded, but it's not like 4-step has gone anywhere. It's still the most common beat pattern for electronic music. It's just a steady beat in 4/4 time with a kick drum on each beat, so it just goes boom boom boom boom with each measure, and it's super easy to dance to.

Techno and house are pretty much exclusively 4-step.

2-step runs at the same/similar speed as 4-step, and is still in 4/4 time, but the drum beat is split up and made more erratic. You'll often have several drum rhythms going on simultaneously. The effect is that the beat sounds like it is sort of stuttering, sort of like this: boom boom pause boom pause pause boom boom pause boom boom boom (that's three measure's worth there). I think Garage was the only real 2-step going on before dubstep, but I'm not real clear on that part of it.

Dubstep gets it beat patterns from 2-step (thus the "step" in the name).

The "Dub" comes from the reggae tradition of sampling pop songs to build a record in an afternoon. That's why the vast majority of dubstep tracks are remixes - it's just how you make dubstep. The build ups and drops that are so popular are not necessary for dubstep, and just because a song has that stuff doesn't make it dubstep. They are just a natural fit for DJ's (who like to control the energy of a crowd) and a 2-step beat pattern.

It's not really dubstep if it isn't heavily sampled with an erratic 4/4 beat (aka 2-step).

**bigjeff5**on The Bias You Didn't Expect · 2012-10-08T18:58:17.096Z · LW · GW

What else could it be?

The break distance bias found in the papers?

You can't use two pieces of contradictory evidence to support the same argument. If the most highly contested cases still have a chance at success, finding 0% success rate at the furthest distance from the last break (because they are the longest cases and therefore placed last) should not *increase* your belief that there is no bias at work. It should reduce it. How significantly your belief is reduced depends on just how likely you would see 0% success rates at a high distance from break due only to scheduling, but I can't see any way it could legitimately *raise* your belief that there is no bias.

**bigjeff5**on The Bias You Didn't Expect · 2012-10-08T16:35:37.295Z · LW · GW

I agree that any contested case should be longer than an uncontested, however are there not cases where the prosecution simply doesn't need to go through a lengthy argument to prove their case? Prosecution lays out X, Y, and Z evidence that is definitive, and therefore the prosecution doesn't need to spend a lot of time arguing. Are these types of cases not generally shorter than cases that are contested but more likely to succeed? Or does a lengthy defense attempting to weasel out of the evidence make up for a short prosecution? And are these specific cases few and far between?

I doubt it, but there chances of success are surely worse than uncontested applications.

In this context I could believe a very low success rate, but the researchers found a 0% success rate for a number of courts. That still makes me suspicious. I'm still not sure what "very low success rate" means for parole hearings though. Is 20% low? Is it more like 5%? Somewhere in between? Obviously, the lower a reasonable success rate for these types of cases the more likely you'll see 0% rates in different courts, just based on chance.

I don't know enough about the details of Israeli parole hearings to speak definitively about that. I can say >as a lawyer that many times I have made uncontested applications which were denied. This was not in a >criminal or parole context.

Fair point. Like I said above I'm not really sure what my expectations should be for a reasonable success rate in these types of cases (or cases in general). Question though, did these applications tend to be closer to or further from a break than your more successful uncontested applications? (obviously purely anecdotal, but I'm sure you see my point)

**bigjeff5**on The Bias You Didn't Expect · 2012-10-08T15:15:47.157Z · LW · GW

I can certainly buy that, but would there really be zero people who apply even though they don't have much chance of winning? I know a few stubborn people who I would expect to apply anyway even if they didn't have much chance of success. I'd be surprised to find out that the prison system has an insignificant number of people who are like that as well.

Also, do the most highly contested applications (and therefore the longest, and therefore placed last on the docket) really have 0% chance of success? If so, would not those applications be better off not applying at all? It seems to contradict the idea that an application with 0% chance of success would not be filed.

Lastly, with only a 65% approval rate for the early applications, I'm pretty surprised that the prosecution doesn't care about them very much. If they were completely uncontested wouldn't you expect closer to a 100% success rate?

**bigjeff5**on The Bias You Didn't Expect · 2012-10-07T23:55:07.138Z · LW · GW

I don't see any reason there wouldn't be the inverse as well. That is, applications which are immediately rejected, and therefore quite short.

I also find it suspicious that the most highly contested applications would also be the least likely to be approved. Presumably these are the ones which are borderline, and require much argument, pro and con, to come to a decision. Immediate rejections wouldn't require long arguments, and neither would immediate acceptances. Under the above hypothesis, both of these types of cases should be early in the session.

If there were no bias and the cases were arranged by length, I would expect to see nearly 50% of the lengthy contested cases, to be accepted. Or, if that many are simply not accepted, some number significantly greater than 0%. For a slightly different arrangement, if the immediate rejections were placed last in the session I would expect the acceptance rates to start at nearly 100% and progress down to 0%. Unless, of course, there is no such thing as a quick, immediate acceptance parole case, in which case the argument doesn't work anyway. If they are all long there isn't much point in arranging by likelihood to be accepted.

**bigjeff5**on How to deal with someone in a LessWrong meeting being creepy · 2012-09-13T21:09:41.847Z · LW · GW

Body language coaching doesn't just exist, it's an industry. It is typically associated with public speaking, salesmanship, etc, and there are a lot of places (and books, and online resources, etc) to get training. In fact, one of the linked blogs in the OP, "Paging Dr. NerdLove", is completely dedicated to helping men who are bad at inter-personal communication with women (i.e. socially awkward) get better at it, which includes quite a lot of body language training.

It's reasonably well known that body language comprises a significant portion of interpersonal communications, so just like you'd expect with other languages there are quite a lot of resources for learning the language, if you take some time to look for them.

And of course, like any language, the resources are of varying quality and usefulness. But the general idea of "you get what you pay for" holds.

**bigjeff5**on How to deal with someone in a LessWrong meeting being creepy · 2012-09-13T16:13:38.267Z · LW · GW

The heart of the problem is body language.

It's an actual language that must be learned and spoken, but a lot of people for some reason never learned it, or learned it poorly.

When these people interact with strangers, it's exactly like the guy with a bad understanding of a foreign language who tries to speak it, and instead of saying "Hi, are you friendly? Lets be friends!" he says "Hi, I want swallow your head!"

I hope you can see why people wouldn't like someone who goes around talking like that on a regular basis, and that the problem really does lie with the speaker, not the people he's speaking to.

What's worse, if he doesn't understand what others are trying to tell him (in the language he speaks poorly - aka body language) when he makes these kinds of statements he certainly can remain oblivious to the problem and be unable to fix it himself. If a person in that situation never meets a kind soul willing to help him speak correctly then he really is screwed, and there isn't much he can do about it unless he recognizes the problem on his own and seeks help.

**bigjeff5**on The Parable of the Dagger · 2012-09-03T00:23:52.389Z · LW · GW

I'm still not getting the difference. He chose the second box because he deduced the the key must be there based on the assumption that one of the inscriptions was true. There is no equivalence between assuming a key in the second box and deducing a key in the second box based on a false premise.

However, assuming one of the inscriptions is true and assuming a correlation between the inscriptions and the contents of the box seem the same to me. He can't deduce a correlation between them, because the only basis for such a correlation is the existence of the inscriptions and the basic format of the king's challenge (which was not identical to the jester's own riddle). There is nothing in the first inscription to suggest a correlation exists, particularly if he determined that the inscription must be false! It has to be a faulty assumption, and I don't see how it is different than assuming one of the inscriptions must be true, other than semantically.

I'm not trying to be obtuse here, I'm just not seeing the difference between what you've said and what I've said.

**bigjeff5**on The Parable of the Dagger · 2012-09-02T18:25:04.605Z · LW · GW

For the inscriptions to be either true or false, they would have to correlate with the contents of the boxes. If he didn't assume this correlation existed, why would he have bothered trying to solve the implied riddle, and then believe upon solving it that he could choose the correct box?

The assumption that one of the inscriptions is true is also the assumption that the contents of the boxes correlate with the truthfulness of the inscriptions. And the key point is that neither inscription need be true, because the contents of the boxes don't correlate with the truthfulness of the inscriptions. And in fact, neither inscription was true.

In other words, I don't understand why you're arguing a simple clarification of essentially the same point you made.

**bigjeff5**on 37 Ways That Words Can Be Wrong · 2012-09-02T18:14:50.430Z · LW · GW

I think that's basically the point - the argument is technically valid, but it is wrong, and you got there by using "human" wrong in the first place.

Socrates is clearly human, and the definition on hand is "bipedal, featherless, and mortal". If Socrates is mortal, then he is susceptible to hemlock. When Socrates takes hemlock and survives, you can't change the definition of "human" to "bipedal, featherless, not mortal". You're still using the word "human" wrong.

What's telling here is that you don't say "Socrates is not human" because you already know he is. If you do go down that route, even though your arguments are correct the conclusion will be intuitively wrong - just another valid but incorrect argument. There are undefined characteristics regarding what it is to be human which carry significantly more weight than the definition itself, and instead of encapsulating them in the definition you've tried to ignore them - tried to make reality fit your definition rather than the other way around.

**bigjeff5**on The Parable of the Dagger · 2012-09-01T23:42:32.661Z · LW · GW

Unlike the jester's riddle, the king never claimed there was any correlation between the contents of the boxes and the inscriptions on those boxes. The jester merely assumed that there was.

**bigjeff5**on Two More Things to Unlearn from School · 2012-09-01T18:29:11.557Z · LW · GW

It couldn't be that, I was raised among the proletariat. Not much prestige dialect signalling there. (There is some, of course, but nothing like the bourgeoisie.)

**bigjeff5**on Two More Things to Unlearn from School · 2012-09-01T18:17:28.597Z · LW · GW

I think in my previous post the implication is that I believe the punishment was unwarranted. That is not the case (though I certainly felt that way at the time). I simply felt the reason given for the detention was less important than the experience of realizing that authority figures can be wrong.

It was entirely appropriate for the teacher to give me detention, because I actually was interrupting class when she was trying to teach, and I don't think I was being particularly helpful to the rest of the students. What she was teaching was correct, as far as I can remember, however there isn't much a person can do about 40+ years of poor habits when it comes to speaking English.

She was in a bad position, and did a reasonable job under the circumstances. I was just a bratty little smart aleck making her life difficult.

**bigjeff5**on The Futility of Emergence · 2012-03-22T18:37:12.784Z · LW · GW

Yes, the point is to be sure you aren't using "Emergence" or "Emergent Phenomena" as stop signs. That you recognize that there is in fact a cause (or causes) for what you are seeing, and if the total seems to be more than the sum of its parts, that there is some mechanism that exists that is amplifying the effects.

Emergence is not an explanation by itself.

**bigjeff5**on Hearsay, Double Hearsay, and Bayesian Updates · 2012-02-22T00:07:26.554Z · LW · GW

The appellate system itself - of which cases involving new DNA evidence are a tiny fraction - is a much more useful measure.

There are a whole lot more exonerations via the appeals process than those driven by DNA evidence alone. This aught to be obvious, and the 0.2% provided by DNA is an extreme lower bound, not the actual rate of error correction.

Case in point, I found an article describing a study on overturning death penalty convictions, and they found that 7% of convictions were overturned on re-trial, and 75% of sentences were reduced from the death penalty upon re-trial.

One in fourteen sounds a lot more reasonable to me, and again that's just death penalty cases, for which you'd expect a higher than normal standard for conviction and sentencing.

The standard estimate is about 10% for the system as a whole.

**bigjeff5**on No Evolutions for Corporations or Nanodevices · 2012-01-05T23:36:31.990Z · LW · GW

The theory that you are familiar with is a little off. What stars can produce is solely a function of size, not generation. Already fused material from a previous star does not allow the new star to fuse more elements. Likewise, the longevity of stars is solely a function of size. It's a balance between the heat of fusion and the pressure of gravity. More matter in the star means more pressure, which means the rate of fusion increases and more elements can be fused, but the fuel is consumed significantly faster.

The smaller a star is the longer it burns, because there is less pressure being exerted by gravity to drive the fusion process. Big stars don't last long (the biggest only a few million years), but they produce the all of the naturally occurring elements - up to iron via normal fusion, and the heavier elements during supernova that occurs after iron fusion begins. Smaller stars like our sun will never get past the carbon stage and will never go supernova, and smaller stars still like brown dwarfs will never get past the hydrogen stage. These small stars last the longest because their rate of fusion is incredibly slow.

**bigjeff5**on Dissolving the Question · 2012-01-05T01:11:52.493Z · LW · GW

Interesting! I hadn't thought about quantum tunneling as a source of uncertainty (mainly because I don't understand it very well - my understanding of QM is very tenuous).

**bigjeff5**on Dissolving the Question · 2012-01-04T22:44:04.026Z · LW · GW

I'm not sure I understand how quantum events could have an appreciable effect on chemical reactions once decoherance has occurred. Could you point me somewhere with more information? It's very possible I misunderstood a sequence, especially the QM sequence.

I could also see giving different estimates for the population of Australia for slightly different versions of your brain, but I would think you would give different estimates given the same neuron configuration and starting conditions extremely rarely (that is, run the test a thousand times on molecule for molecule identical brains and you might answer it differently once, and I feel like that is being extremely generous).

Honestly I would think the decoherance would be so huge by the time you got up to the size of individual cells that it would be very difficult to get any meaningful uncertainty. That is to say, quantum events might be generating a constant stream of alternate universe brains, but for every brain that is functionally different from yours there would be trillions and trillions of brains that are functionally identical.

If you include electrons a single water molecule has 64 quarks, and many of the proteins and lipids our cells are made of have thousands of atoms per molecule and therefore tens of thousands of quarks. I am having a hard time envisioning anything less than hundreds of quarks in a molecule doing enough to change the way that molecule would have hooked into its target receptor, and further that another of the same molecule wouldn't have simply hooked into the receptor in its place and performed the identical function. There may be some slight differences in the way individual molecules work, but you would need hundreds to thousands of molecules doing something different to cause a single neuron to fire differently (and consequently millions of quarks), and I'm not sure a single neuron firing differently is necessarily enough for your estimate of Australia to change (though it would have a noticeable effect given enough time, a la the butterfly effect). The amount of decoherance here is just staggering.

To summarize what I'm saying, you'd need at least hundreds of quarks per molecule zigging instead of zagging in order for it to behave differently enough to have any meaningful effect and probably at least a few hundred molecules per neuron to alter when/how/if that neuron fires, or whether or not the next neuron's dendrite receives the chemical signal. I would think such a scenario would be extremely rare, even with the 100 billion or so neurons and 100 trillion or so synapses in the brain.

**bigjeff5**on Dissolving the Question · 2012-01-04T19:02:16.403Z · LW · GW

Do neurons operate at the quantum level? I thought they were large enough to have full decoherance throughout the brain, and thus no quantum uncertainty, meaning we could predict this particular version of your brain perfectly if we could account for the state and linkages of every neuron.

Or do neurons leverage quantum coherence in their operation?

**bigjeff5**on Two More Things to Unlearn from School · 2012-01-03T01:45:04.963Z · LW · GW

Yeesh, that's terrible. It kind of figures that he'd rather mislead a class full of students about the way physics works than own up to his mistake.

It reminds me of an error I had been taught about the way airfoils work that wasn't corrected until I read a flippin comic strip on the subject almost a decade after I graduated high school.

I was stunned, and spent the rest of the afternoon learning how airfoils *really* work. What makes this particular example so tragic is it leverages another principle of physics that you won't realize doesn't fit if you are taught to accept everything the teacher says as gospel. What's worse is I'm pretty sure the mistake is still there in the vast majority of textbooks.

**bigjeff5**on Two More Things to Unlearn from School · 2012-01-02T19:13:54.397Z · LW · GW

Ad hominem literally means "to the man" or "to the person".

It was most certainly an ad hominem question, but given the framing he probably wasn't intending to discredit the argument with the ad hominem and therefore didn't commit the ad hominem *fallacy*.

The *fallacy* is making an ad hominem attack *in order to distract from or discredit the argument without addressing the merits of the argument itself*. The traits can certainly be related to the argument, and in fact the more closely related the traits are the more effective the fallacy is at convincing others (e.g. He's wrong about QFT because he isn't a physicist vs he's wrong about QFT because he drinks milk - both fallacies, the first much more effective than the second). That doesn't mean the ad hominem isn't relevant nor worth discussing, it only means the ad hominem is not evidence against the argument. The fallacy lies in thinking that it is.

SORRY, idiots! (that's not a.h. either)

It's still ad hominem, it's just not a logical fallacy (but given that the word "idiot" means a person with extra-ordinarily low intelligence, it's almost certainly incorrect).

**bigjeff5**on Two More Things to Unlearn from School · 2012-01-02T18:11:30.617Z · LW · GW

I will say that I also had a high school English teacher who would use the wrong word or give a ridiculous interpretation in the hopes that a student would correct him and learn to not always trust authority.

I had a teacher somewhat similar to that my freshman year in high school, except she was a last-minute replacement and was not really an English teacher. Her grammar was atrocious, and I ended up getting detention for correcting her too often (interrupting class or lack of respect or some such was the reason given on the detention). It was probably my first real experience with an authority figure being so utterly and obviously wrong, and I wasn't sorry at all for the detention. It was well worth it.

**bigjeff5**on Absolute Authority · 2012-01-02T01:26:49.381Z · LW · GW

Just noticed this comment when I was looking through my messages for an old comment, and I wanted to respond.

It is the word "too" that is important there, and the usage you describe is only used as an affirmative for contradicting a negative statement (at least, that's proper grammar anyway).

For example, if the original statement had been "God must not make a boulder he cannot lift!" and I had responded with "God must too make a boulder he cannot lift!" you would be right, but the original statement is an affirmative statement ("God can make a boulder he cannot lift."), my own sentence before it is an affirmative (in the grammatical sense - not so much in the "uplifting" sense), so trying to contradict either with an affirmative doesn't make any sense.

Also, I did a Google search, and while using "too" between must and another verb is not common, using "must too" to mean "must also" is by far the most common usage I could find. I do admit that other combinations of *verb* "too" *verb* seem to imply contradicting a negation even without the proper context, so that usage is definitely not as clear as I originally thought it would be. I still think it's pretty, though.