Author: Robert Hyatt
Date: 08:27:16 05/28/01
First, I think the two best programs won this event, which is the way it should always happen, but doesn't. I was reasonably happy with the way things went for me. The second day was a bit of a screw-up, but it was my fault. My 1st cousin just had a heart transplant in Ochsner's clinic in New Orleans. This happened late Saturday afternoon. My wife and I decided to drive down to New Orleans that night to see him and his family (about a 6 hour drive from here). I set up an automated script to set up crafty for the next morning's games (A GM wanted to play that night so I had it set up for normal play until 8:30am Sunday morning. I checked the script to be sure it worked, but I didn't notice a typo that set up the wrong opening book. I logged on Sunday morning from the hospital waiting room (people thought I was nuts) to check on the games and rounds 5/6 went ok. In round 7, Crafty played Ferret and was out of book very quickly in a Petrov's line. I couldn't understand why it was out of book so I went looking (this is a good example of "can't see the forest for the trees" as crafty would not normally play the petrov against a computer in the first place, but I didn't think about that at the moment.) What I found was that someone had played a game against Crafty with the Petrov's, and after e4 e5 Nf3 Nf6 Ne5 d6 Crafty lost badly due to some students running a final project on the machine and eating up all the cpu time. Crafty 'learned' that opening was bad and flagged it as "don't play". Unfortunately, the !/play % stuff in my book overrides learning so it played the first 2 moves then said "hey, I can't play this line" and went off on its own. It played miserably and got whacked convincingly by a tactical shot Ferret found. I had "found" the problem (learning had disabled all the petrov lines) and so gave it no more thought. I checked back in at the start of round 8 and found to my amazement that it had played 1. e4 as white. My tournament setup _never_ allows 1. e4 as a choice. It got into a castle-opposite Sicilian and lost easily. When I saw this opening I suddenly realized what I had done. I had not copied the "start book" (books.bin/bookc.bin) files to the right place. But then I remembered my script and thought "I remember doing that..." Of course, I had recently renamed some disk partitions and I used the old names in the script so I played with the normal human-version of those files. In retrospect, I was fortunate to go 2-2 that last day, as it could have easily played other bad non-computer-type openings that it will still play vs humans to provide them with more variety. At least I didn't see any bugs, nor evaluations I didn't agree with, so from a purely chess point-of-view, it did fine. The book problem was a one-time thing that doesn't cause me much concern, other than my silly mistake in trying to do an automatic setup since I wasn't sure I could log on on Sunday morning in time to set things up myself. Congrats to everyone. It was nice to have an event without insulting comments tossed around, people storming off and screwing up the pairings, etc. The pairings were a bit slow at first, and a few participants screwed up the colors which the TD didn't know how to handle. I do think I would prefer a 60-0 time control for future events. The increment let some games go _way_ over the time limit for the start of the next round, which is a headache for everyone. With 60 0, you can play a round every 2.5 hours with little chance for problems, leaving time to check and re-check the pairings, etc.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.