Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Comment author: Martin-2 06 August 2015 08:27:23AM 0 points [-]

One of my favorite lessons from Bayesianism is that the task of calculating the probability of an event can be broken down into simpler calculations, so that even if you have no basis for assigning a number to P(H) you might still have success estimating the likelihood ratio.

Comment author: steven0461 21 April 2011 09:40:37PM 3 points [-]

It's one thing to say they can be changed, and another to say they can be changed just by being informed of the relevant evpsych.

Comment author: Martin-2 11 March 2015 02:52:07PM 3 points [-]

In the spirit of OP, since there's no guaranteed way to overcome this form of social anxiety and the afflictee will need to try many things to see what works for them, listening to a good evpsych story is as good a thing to try as any.

Comment author: Martin-2 12 November 2014 07:02:15AM *  20 points [-]


In response to comment by gjm on Probability puzzles
Comment author: steven0461 22 April 2011 09:56:12PM *  11 points [-]

A more general lesson is that whenever the answer to a puzzle causes you to go, "oh how wondrous that this question could have such a strange answer", you were probably tricked into accepting an anti-helpful framing of the problem, and one of the reasons why the puzzle-poser didn't guide you into a helpful framing instead was probably exactly that such anti-helpful framings cause people to feel that way.

Comment author: Martin-2 12 August 2014 01:04:55AM *  0 points [-]

This post is not evidence for that lesson. When OP's puzzle is stated as intended it indeed has a wonderful and strange answer. The meta-puzzle: "Are these two puzzles essentially the same?" referring to the puzzle as intended and as presented also has a wonderful and strange answer; in fact, John Baez and maybe all of his commenters have been getting it wrong for several years. Our intuition is imperfect, and whether the puzzles you come across tend to use this fact or just trick you with sneaky framing depends on where you get your puzzles.

Comment author: gwern 01 April 2014 04:56:37PM *  22 points [-]

Fairly often. One strategy I've seen is to compare meta-analyses to a later very-large study (rare for obvious reasons when dealing with RCTs) and seeing how often the confidence interval is blown; usually much higher than it should be. (The idea is that the larger study will give a higher-precision result which is a 'ground truth' or oracle for the meta-analysis's estimate, and if it's later, it will not have been included in the meta-analysis and also cannot have led the meta-analysts into Milliken-style distorting their results to get the 'right' answer.)

For example: LeLorier J, Gregoire G, Benhaddad A, Lapierre J, Derderian F. "Discrepancies between meta-analyses and subsequent large randomized, controlled trials". N Engl J Med 1997;337:536e42

Results: We identified 12 large randomized, controlled trials and 19 meta-analyses addressing the same questions. For a total of 40 primary and secondary outcomes, agreement between the meta-analyses and the large clinical trials was only fair (kappa ϭ 0.35; 95% confidence interval, 0.06-0.64). The positive predictive value of the meta-analyses was 68%, and the negative predictive value 67%. However, the difference in point estimates between the randomized trials and the meta-analyses was statistically significant for only 5 of the 40 comparisons (12%). Furthermore, in each case of disagreement a statistically significant effect of treatment was found by one method, whereas no statistically significant effect was found by the other.

(You can probably dig up more results looking through reverse citations of that paper, since it seems to be the originator of this criticism. And also, although I disagree with a lot of it, "Combining heterogenous studies using the random-effects model is a mistake and leads to inconclusive meta-analyses", Al khalaf et al 2010.)

Comment author: Martin-2 18 April 2014 01:32:36AM *  6 points [-]

I'm not sure how much to trust these meta-meta analyses. If only someone would aggregate them and test their accuracy against a control.

Comment author: Martin-2 10 March 2014 07:18:54PM *  2 points [-]

I can't do anything on purpose.

  • Professor Utonium, realizing he has a problem
Comment author: Nornagest 20 January 2014 09:20:17PM 3 points [-]

It is unlikely that a person who likes classical music and computer science will be able to self-modify into a person who likes heavy metal and stealing cars.

Can't speak for stealing cars, but there's more overlap between classical and metal fans than you might think; there exists a subgenre of neoclassical heavy metal, even.

Comment author: Martin-2 14 February 2014 02:50:22PM 2 points [-]

Also, since cars are now quite integrated with computers this person might have lots of fun stealing them. And if ze watches Breaking Bad there's a whole lot of inspiration there for intellectuals looking to turn to a life of blue-collar crime.

Maybe I should be steel-manning Locaha's argument but my point is I don't think the limits of this sort of self-mod are well understood, so it's premature to declare which mods are or aren't "real world".

Comment author: gjm 26 November 2013 11:49:12PM 2 points [-]

I do so hope so.

Comment author: Martin-2 27 November 2013 01:19:45AM 1 point [-]

I'm a musician if that's any hint.

Comment author: Martin-2 26 November 2013 08:23:03PM 14 points [-]

Done. I hate to get karma without posting something insightful, so here's a song about how we didn't land on the moon.

Comment author: Eugine_Nier 02 September 2013 04:44:42AM 17 points [-]

One of the penalties for refusing to participate in politics is that you end up being governed by your inferiors.


Comment author: Martin-2 06 September 2013 05:13:16AM 4 points [-]

One of the penalties for participating in politics is that your superiors end up being governed by their inferiors.

View more: Next