gwern comments on Open Thread September, Part 3 - Less Wrong

2 Post author: LucasSloan 28 September 2010 05:21AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (203)

You are viewing a single comment's thread.

Comment author: gwern 28 September 2010 11:49:44PM *  7 points [-]

I'd like to remind everyone that I have continued to work on predictionbook.com, and now it's up to ~1800 predictions, and for those of you in a hurry, there are dozens/hundreds of interesting predictions coming due in the next year or 3: http://predictionbook.com/predictions/future

Remember, signing up for Intrade is hard and it's not profitable to wager on many of its long-term contracts, but PB is absolutely free!

(One thing is for sure: with ~443 predictions registered, I should eventually be pretty well-calibrated...)

EDIT: Hanson on the value of track records: http://www.overcomingbias.com/2010/09/track-records.html

Comment author: JoshuaZ 29 September 2010 12:05:11AM 0 points [-]

Also it would be nice if you had a small amount of additional explanation of how to use the interface.

Comment author: gwern 29 September 2010 12:40:10AM *  0 points [-]

Well, it strikes me as pretty intuitive to me. The only part that seems to trip people up is that a 0-49% prediction for is translated into a 100-x% prediction against.

Comment author: JoshuaZ 29 September 2010 12:42:12AM 1 point [-]

Right, that was the main issue. I had to think for a second to figure out how to put that in.

Comment author: Douglas_Knight 29 September 2010 05:38:10AM 0 points [-]

Unless something has changed, a UI problem is that people often "judge" when they should "estimate" or "wager" or something.

Comment author: gwern 29 September 2010 12:25:45PM 0 points [-]

That explains some judgements, but there is a line of text below the judge buttons specifically to forestall that; I'm not sure what easy solutions there are to that.

Comment author: JoshuaZ 29 September 2010 12:01:05AM 0 points [-]

A large number of these are things where my confidence on them is much too low to bet. Almost anyone already willing to bet on them would likely have a lot more of thought and relevant data. Still, some of these look interesting enough to maybe play with.

Comment author: gwern 29 September 2010 12:35:03AM 0 points [-]

Well, isn't that all the more reason to use PB rather than Intrade?

And your confidence may be too low. For example, looking at my own profile, I am so far significantly underconfidence, which is a problem to be fixed just like overconfidence.