Understanding one's own "15 minutes of fame"

[Peter Berger]

This is a light, chatty topic - enter at your own risk.

In another thread I don't want to disturb and enter Ajedrecista gave a link to a tome of an ancient magazine:
http://www.chesscomputeruk.com/SS_70.pdf

I've read all of the articles there with great interest, no idea how respected this magazine was back then, it was before my time , it partly felt a bit like a tabloid to me but it probably was a respectable source back then.

By far the most interesting and fun for me was about the "Rebel8 - Crafty Challenge" for personal reasons. It gave me the background to understand the true reasons for my own clearly most successful thread on CCC ever. I had no clue about the historical background back then, I was just honestly curious. And people clearly understood that I had no idea myself, else it wouldn't have worked.

I even dubbed it the "Junior - Crafty NPS challenge" or sth very similar to that with no idea of the background and its implications.

My motivation: Bob Hyatt did an emotional post that if he would ever lose a serious match at tournament time control to another engine although having a 10 times hardware advantage, he'd quit chess programming. I had an old P233MMX myself, by accident my fastest computer was about exactly 10 times faster. Crafty was more competitive than at other times back then. I thought, finally something I can really see and find out for myself. Having no clue about the old auto232 controversy it certainly helped that I said that all games would be run manually and with ponder on ( Bob always claimed the latter to be very relevant).

I announced the match in a kind of official lingo and that I'd report its progress and the games eventually. As I was really working a lot in real life, this was just my hobby, it didn't even go fast at all, there were some significant breaks.

Ed's statement in the old magazine that he was not in the "mood for these kind of explanations" really made my day. Because it was just true. Once Junior started to win games I really got many complaints. And then both programs clearly even showed some bugs you wouldn't expect. But I was honestly completely impartial and trying to do my best.

And the match was finished, Crafty eventually won it by the narrowest of margins, the contestants were probably close in strength under these conditions. In the end I got many replies I wouldn't have expected by good and able people not usually active on CCC. And Bob said in the end when I made my slight frustration feel with all his complaints: you did well. So we even worked together some later.

Gossip for people interested.

Peter

[Rebel]

With pleasure I read Eric's story about the NPS experiment, pretty solid and detailed. One imported thing was missing. My main point was that a good evaluation could cover many search plies. Hence the 1:100 challenge.

My first appearance was in 1986 the WCCC in Cologne. I spoke to the programmers of that time, all they could talk about was PVS with hash tables. It was never about eval, search search was all. I never heard of PVS or hash tables, heck I did not even had a Q-search. At that time all I had was a good eval.

And so was Crafty, I looked at the code, played some games and noticed not much had changed, well optimized for speed, underestimation of eval as a big eval would only lose NPS. And so I took my chances.

Of course it was only one game, statistically meaningless, but that one game proved my theory.