Quake 3 bots left running for 4 years, find optimal strategy?

i.imgur.com

32 points·polemic··10 comments

Feels like the Monkey banana and water spray experiment happened with the bots (http://wiki.answers.com/Q/Did_the_monkey_banana_and_water_sp...)

I bumped it, I agree, very interesting. A link to the full thread would be nice.

4chan, so it'll disappear pretty soon.

Thread is screencapped from 2011, and 4chan threads disappear after a few hours.

Luckily for you: http://chanarchive.org/4chan/v/22226/discussion-on-bots-a-i

This submission deserves more attention that it is getting!

If it's real. ...chan boards are the home of (sometimes very elaborate) trolls. Additionally: 22:52 "I'm gonna download q3a", 23:09 description of the game. Yet it took >20min to get a screenshot later.

I don't buy the "not moving" part either. If they were using NN, then there's no reason they'd include / train connections to scores - only to current environment. NN with any kind of randomisation added between games (to get out of local maxs) would not stand still either.

TL;DR - screenshot is not a proof, ...chan boards are not reliable news services.

It'd depend on the function you use to measure error, wouldn't it?

Computational intelligence / nature inspired computing tends first of all to reveal flaws in the problem statement.

Cool. If true, seems like they had evolved tit-for-tat: keep the peace or we'll all punish you for it.

The behaviour of the bots, after 4 years, is fascinating!

"Ice cream? god I love ice cream!"

Reply on news.ycombinator.com