gwern comments on New Singularity.org - Less Wrong

14 Post author: lukeprog 18 June 2012 02:29PM

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

Comments (100)

You are viewing a single comment's thread.

Comment author: gwern 20 June 2012 11:35:13PM 1 point [-]

I see a lot of mistakes pointed out here, but also a lot of claims about preferences, many contradictory. Are there any plans to A/B test stuff or at least something more rigorous than 'I agree with X, that blue is too dark'?

Comment author: lukeprog 22 June 2012 02:21:13AM 0 points [-]

Probably not; too expensive.

Comment author: gwern 22 June 2012 02:35:40AM 0 points [-]

Oh; I didn't know your web guys charge extra for that. Never mind then.

Comment author: John_Maxwell_IV 22 June 2012 08:05:17PM *  -1 points [-]

BTW everyone, this is the coolest A/B testing tool evar:

http://genetify.com/demo/

(Or it certainly looks like it, I haven't actually used it.)

Comment author: gwern 22 June 2012 08:19:41PM 0 points [-]

From the description, it reads like an ad hoc unmaintained multi-armed bandit.

Comment author: John_Maxwell_IV 22 June 2012 08:26:44PM 0 points [-]

You make that sound like it's a bad thing. Are you sure humans are better at multi-armed bandit problems than algorithms?

Comment author: gwern 22 June 2012 08:33:58PM 0 points [-]

Are you sure humans are better at multi-armed bandit problems than algorithms?

I'm pretty sure they're not, which is why I am questioning your enthusiasm for such an ad hoc library.