OpenChess

Jeremy Bernstein wrote:New York Times: http://www.nytimes.com/2011/07/03/cross ... -code.html

jb

Please Log In ??

I don't get any login request, just the first page of the article pops up.

try this onehttp://www.nytimes.com/2011/07/03/cross ... ss&emc=rss

JcMaTe wrote:try this onehttp://www.nytimes.com/2011/07/03/cross ... ss&emc=rss

This is new and from the above:

"Larry Kaufman, a grandmaster who helped Mr. Rajlich in the development of Rybka, but who now works on a rival called Komodo, said in an e-mail that he believed only earlier versions of Rybka were based on Fruit and Crafty.

“In my opinion, there was reasonable basis for the disqualification of Rybka 1 and any tournament victories that occurred within a year or so of its release,” Mr. Kaufman wrote. “By the time Rybka 3 came out, it was for all practical purposes a completely new program,” he said."

So it appears a Larry, who helped develop versions of Rybka from 2.3.2a to 3, appears to agree with the ICGA decision for earlier Rybkas. SInce Larry tuned the evaluation function, he would have good inside knowledge about it.

mjlef wrote:So it appears a Larry, who helped develop versions of Rybka from 2.3.2a to 3, appears to agree with the ICGA decision for earlier Rybkas. SInce Larry tuned the evaluation function, he would have good inside knowledge about it.

Everything I've read from Larry would indicate that he sent Vasik the data used to tune the evaluation (probably in some iterative process), but never saw the code. I can try to find the post (it was on CCC) if you're curious, but I'm pretty sure that this was the situation.

Jeremy

that is my recollection as well. Somewhat akin to someone sending me a personality file for crafty, which can adjust any eval parameter, without them having to look at the source at all...

Should Vasik's case be picked up by the FSF, he may want to consider contacting this individual:
Source : New York Times

Jeremy Bernstein wrote:

mjlef wrote:So it appears a Larry, who helped develop versions of Rybka from 2.3.2a to 3, appears to agree with the ICGA decision for earlier Rybkas. SInce Larry tuned the evaluation function, he would have good inside knowledge about it.

Everything I've read from Larry would indicate that he sent Vasik the data used to tune the evaluation (probably in some iterative process), but never saw the code. I can try to find the post (it was on CCC) if you're curious, but I'm pretty sure that this was the situation.

Jeremy

I emailed Larry about this several months ago, and you are right. Larry tested and tuned some of the evaluation parameters only, starting around Rybka 2.3.2a but much more so for Rybka 3 where he also proposed many new evaluation features. Larry only had access to parameter files and not the source code. And I do not even know if Larry had seen or understood Fruit code at this time (Larry is very smart, but not a programmer). We made clear Larry's innocence in the report when we wrote:

"Note: Larry Kaufman is a Grandmaster long associated with helping programmers
improve chess engines. There is absolutely no suspicion or belief on the part of the
investigation panel or the ICGA that Kaufman has been in any way involved in any
aspect of the copying behavior alleged of Vasik Rajlich. So far as the panel secretariat
and the ICGA are concerned Kaufman’s reputation in this respect is squeaky clean"

kingliveson wrote:Should Vasik's case be picked up by the FSF, he may want to consider contacting this individual:

Code: Select all

Mark A. Lemley, a Stanford law professor who specializes in science and technology issues, wrote in an e-mail that because Fruit and Crafty are freely available may mean that Mr. Rajlich is not guilty of misconduct if he copied some of the code. But, Mr. Lemley added, “I can see why the Computer Gaming Association might want to prohibit it under its rules.”

Source : New York Times

Guy spoke without knowing what he was talking about. Crafty and Fruit are not "free software". They both have licenses. Which makes _all_ the difference in the world.

hyatt wrote:I don't get any login request, just the first page of the article pops up.

Firefox shows me allways a login request - with IE now i get instantly the article