Lisp has average performance in Google AI Competition

jcole.posterous.com

2 points·JoshCole··4 comments

Interesting, but I don't think there's enough data on the Lisp side to interpret much. From memory, wasn't it a few dozen Lisp entries against well over a thousand for Java?

Raw statistical uncertainty aside, if we're trying to judge the languages themselves it would need to be done with programmers skilled with their chosen language. I'd guess that most of the Java entries were by people that knew Java reasonably well from working with it every day, but I wouldn't be surprised if many in the Lisp section were people that had little or no prior experience with Lisp and decided that this would be a fine opportunity to learn it.

There were only a handful of Lisp entrants while Java had more than any other language in the competition. I agree with you that we shouldn't read much into this. I also think we shouldn't be reading much into the posts about how it was Lisp that won the competition.

... because it is hard to compete with strongly typed, optimized in runtime code.

Yea. If you get the same program in two different languages, you would expect the faster language to win, it can do a more thorough search. I glanced at the results of the previous competition and the first page was drenched in C++. It is too bad that Clojure wasn't offered as a language in this competition, I remember hearing that it's performance is right up there with Java.

Maybe I should do a on the correlation between language speed and ranking next.

Reply on news.ycombinator.com