hgm wrote: ↑Wed Jul 18, 2018 9:42 pm
Before every WCCC there is a players meeting, which basically decides everything. As a former participant I have often been approached by e-mail for my opinion on some proposed rule change.
As a former participant, I quite vividly remember a sudden rule change that invalidated at least one year of my work, hadn't been presented to the participants beforehand and when pointed out how silly it was, the response was that the comments would be considered for the next year.
Obviously, I never played again after that.
Gian-Carlo,
Do you remember what the rule change was? I would like to see if it is still a problem.
tiptongrange wrote: ↑Wed Jul 18, 2018 10:34 pm
Laskos, what do you mean by "... and another game of Leela where she did not played..."?
One Leela game was postponed, because its operator was absent due to a domestic emergency.
Postponed but it was played. Leela had a time penalty because of that so it played with time odds.
Instead of 105'+15" for both players, it was Leela 30'+15" versus 45'+15" for opponent if i recall correctly.
Ah, yes? Didn't know. I am not sure whether to include it as a regular game.
I was there when the TD and GridGinko operator and Leela operator negotiated the time difference. By being late for the game Leela would normally have forfeited the game. To discourage people being late, and wanting the game to be played, I think a 15 minute time penalty quite reasonable. Note Leela had a lot of time at game end, so it did not get into time trouble even with the 15 minute penalty. Although people might not like the decisions of the TD and operators, the game happened and should be counted.
I might note all the programmers were very nice about this. All agreed to a delay for the next round so Leela would play. I lent the Leela operator a CAT-5 Ethernet to USB adapter since his laptop had no CAT-5 port. People buy each other candy. It is not a cuthroat tournament. People get along.
Meanwhile Leela didn't manage to win a single game and finished near the bottom. Probably a disappointing result. On the upside: I am sure the Leela Chess Zero people have learned many important lessons about tournament preparation.
mjlef wrote: ↑Thu Jul 19, 2018 7:45 am
Do you remember what the rule change was? I would like to see if it is still a problem.
Hi Mark,
I'll tell the story in a private message.
The rule itself is irrelevant, as the point was about the (non-democratic) procedure.
The rule was to max the number of cores to 8. That excluded custer programs. This rule was subsequently revoked for WCCC on programmer's demand, and the WCSC was spun off to keep the rule (and refine it to totally equal hardware).
hgm wrote: ↑Thu Jul 19, 2018 2:17 pm
The rule was to max the number of cores to 8. That excluded custer programs. This rule was subsequently revoked for WCCC on programmer's demand, and the WCSC was spun off to keep the rule (and refine it to totally equal hardware).
There was also this case where Vincent was arbitrarily banned, and then someone pointed out that by the same reasoning used, Harvey Williamson should also have been banned...
I had forgotten about this, but this discussion lead me to an email from you about that situation. I have mixed memories of my time in computer chess, but that email thread was a positive highlight. Even if it didn't lead to much of anything in the end.
I'm curious about Leela's final game against Jonny. In their blog, they characterize it as "Leela blunders its queen to a discovered attack" but to my eyes it was neglecting king safety to try trapping a piece with g5-f6, and actually gaining material in the end (RBN-Q) but those pieces were uncoordinated. Is this a well-known trap in the Catalan?
An obvious improvement for future competitions without betraying their "zero knowledge" principles would be to construct their own opening book from self-play, simply to avoid losing time on the clock during the opening.
IanO wrote: ↑Thu Jul 19, 2018 10:41 pmIs this a well-known trap in the Catalan?
No. It looks like 9. Rc1 was Johnny's first move out of book, and that is a very rare move (normal is 9. Rd1 or 9. Bf4). So book knowledge would not have helped Leela here.
13. .. Nh5 looks bad. Stockfish prefers 13. .. g5. And 15. .. f6? was also bad, but Black is gaining material, so it is easy to see how this is tempting.
--Jon
Last edited by jdart on Fri Jul 20, 2018 12:06 am, edited 1 time in total.
pretty sure its just a straight up blunder, leela can occasionally recover brilliantly from them. if there was no book, it seems to me leela did awfully well in the opening.
one famous one was a limit on number of processors, were the limit was completely arbitrary and announced at the last minute. Never understood that and it was eliminated the next year...
