Author: Robert Hyatt
Date: 19:14:01 08/11/01
Go up one level in this thread
On August 11, 2001 at 13:56:37, Uri Blass wrote: >On August 11, 2001 at 08:40:53, Robert Hyatt wrote: > >>On August 11, 2001 at 03:43:37, Mark Young wrote: >> >>>On August 10, 2001 at 17:45:56, Robert Hyatt wrote: >>> >>>>On August 10, 2001 at 14:18:15, Theo van der Storm wrote: >>>> >>>>>On August 10, 2001 at 13:25:10, Bruce Moreland wrote: >>>>>>On August 10, 2001 at 02:57:51, Peter Berger wrote: >>>>>... >>>>>>>I don't think many reporters were really thinking of this Hongkong game. More >>>>>>>likely they were confused by the qualifier against Deep Junior. >>>>> >>>>>On the other hand here's what journalists world-wide use >>>>>as a reference to the history of computer-chess: >>>>>http://www.braingames.net/index.php?f=chess_center/events&p=mvm_history >>>>>Evidently the list has been carefully designed with regard to >>>>>the information that has been left out. >>>> >>>>It is also wrong in at least one place. Cray Blitz never had "one processor >>>>per square". That was HiTech, but some incompetent got it wrong. And then >>>>they have Fritz beating Deep Blue in 1995, before it existed. That was deep >>>>thought. >>> >>>It was reported at the time in Chess Life, or Inside chess that Deep Thought had >>>some kind of hardware problem. This problem slowed down the search speed of Deep >>>Thought. Do you know if that report was accurate? If I remember correctly Deep >>>thought played one bad move caused by this problem, which cost Deep Thought the >>>game. It was reported that a correctly running Deep Thought was able to play the >>>correct and saving move in a matter of seconds. >> >> >>What happened was that Deep Thought's communication link was broken. Which >>killed the program. It was pondering correctly and it had found a reasonable >>move that would have avoided the instant loss. They reconnected, and then >>restarted the program and it moved much quicker than it should have because >>it was started "cold" and the time was set which lost a good bit of time. >> >>If not for the comm failure, the game most likely would have ended differently, >>but things happen. > >It means that the Deep thought made a bad job in explaining the problems >after the game. > >I read in the israeli chess newspaper after the game that the Deep thought >team explained after the game that Fritz3 surprised them in the opening. >They said nothing about the comm failure that is a more convincing excuse. I don't know what you read/heard, but the comm failure was reported when the game was over. That's when I first heard about it. I believe that the DB team has also mentioned this in several of their talks. Most likely the reporter that wrote the story you read simply didn't ask the right people, or else reported based on incomplete or inaccurate quotes... Happens all the time in the press... > >I also agree that without c4 the result could be different(I know that >at least it was the case in genius3-genius3 game >and I did not check the new programs) > > They lost a game at the last ACM event due to a monster >>thunderstorm knocking out power to the Watson phone system. Deep Thought was >>on a huge UPS in the lab, but the phones were dead and after two hours, the >>game was given up as unplayable so the next round could be paired. They _still_ >>won the event cleanly (no need of tie-breaks). > >I think that the rules should be changed and if >there is an objective problem of phone connection or internet connection >the game should be delayed. > >Uri The rule says that if there is a network/communication problem, the clock can be stopped. But there are limits. It became obvious that if the game was attempted later, pairings for round 2 (round 1 was at something like 1 or 2pm in the afternoon, round 2 was set for 7 pm that evening if I recall correctly) would be impossible to complete. After a lot of discussion with the participants, it was decided by the TD (with the ACM computer chess committee involved as well as all the participants) that a forfeit was the only possible result. The normal phone system didn't go dead. The PBX at IBM Watson was the problem. Deep Thought was still running as it had a big battery backup available. The normal phone system was up (we were using a phone link to Minneapolis to access our Cray C90. Private hardware failures are really not well covered by the rules, which say that if it is not a comm failure, you can stop your clock no more than twice, for a total of 40 minutes total time. The general consensus was that the rule fit and was applied. None of us liked it, and it was a nice result that DT won the event in spite of that forfeit...
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.