Author: Robert Hyatt
Date: 06:14:24 01/12/99
Go up one level in this thread
On January 11, 1999 at 12:28:58, blass uri wrote: > >On January 11, 1999 at 09:31:03, Robert Hyatt wrote: > >>On January 11, 1999 at 01:15:55, blass uri wrote: >> >>> >>>On January 10, 1999 at 20:03:37, Robert Hyatt wrote: >>> >>>>On January 10, 1999 at 01:28:23, blass uri wrote: >>>> >>>>> >>>>>On January 09, 1999 at 22:08:43, KarinsDad wrote: >>>>> >>>>>>On January 09, 1999 at 09:25:23, blass uri wrote: >>>>>> >>>>>>> >>>>>>>On January 09, 1999 at 06:23:08, Micheal Cummings wrote: >>>>>>> >>>>>>>> >>>>>>>>On January 08, 1999 at 20:07:06, Mark Young wrote: >>>>>>>> >>>>>>>>>If Microsoft devoted substantial resources to the creation of a chess playing >>>>>>>>>program, and devoted one year to the project, how would that program compare (in >>>>>>>>>terms of playing strength only) to the best professional chess playing programs >>>>>>>>>on today's market? >>>>>>>>> >>>>>>>>>** >>>>>>>>> >>>>>>>>>What is the point of this poll question? In a sense, IBM has already done this. >>>>>> >>>>>>The point of the poll question is that it is a controversial topic (hence, all >>>>>>of the posts on it). Additionally, nobody else suggested another poll question. >>>>>> >>>>>>>>> >>>>>>>>>If the point of this question were to mean the chess program must be run on a >>>>>>>>>desktop computer, the results would still be the same. It would be better then >>>>>>>>>the other professional chess programs, if that would be the goal for Microsoft. >>>>>>>>> >>>>>>>>>For Microsoft, substantial resources would mean millions of dollars, a team of >>>>>>>>>programmers, a team of computer chess experts (ex. Bob Hyatt), and a team of >>>>>>>>>high rated professional chess players. All working together to meet the goal. >>>>>>>>> >>>>>>>>>With that kind of resources I don?t think the goal of just topping, the best >>>>>>>>>professional chess programs in terms of playing strength currently out would be >>>>>>>>>much of a challenge. >>>>>>>> >>>>>>>>I would like to know one thing, why would Microsoft want to invest the time and >>>>>>>>money to create a chess program. >>>>>> >>>>>>There are no reasons for them to do it. The poll question was "If". >>>>>> >>>>>>>> Its not as though it is big market. Plus what >>>>>>>>are we going to need to run this program, I suggest with the millions of dollars >>>>>>>>invested that the program would be graphically power hungry along with the power >>>>>>>>and memory to then go and run the chess engine. >>>>>>>> >>>>>>>>I woulds suggest 128 Ram, PII450. 16Meg 2D graphics Card. >>>>>>>> >>>>>>>>I think too many people when hearing this will jump on the bandwagon thinking >>>>>>>>that a big company will create a powerful program, This from a company who takes >>>>>>>>2 years over to release operating systems and even when they do is full of bugs. >>>>>>>> >>>>>>>>Its like Deep Blue, used on a super computer >>>>>>> >>>>>>>I do not think that IBM payed to the right people to help them. >>>>>> >>>>>>Why not? Kasparov is still complaining about it. Sounds like they were >>>>>>successful to me. You can almost always get better people, but you shouldn't >>>>>>argue with success. >>>>> >>>>>I expected deeper blue to play better >>>>>It won only because of stupid mistakes of kasparov in the 2 games that it >>>>>won(resigning in a draw position and going to a line that you were not ready to >>>>>go). >>>>> >>>>>Their previous machine failed to win the computer championship >>>>>and did only 3.5 out of 5 (they lost against Fritz3 and drew against wchess when >>>>>the Fritz3 and wchess used pentium90). >>>> >>>> >>>>Blass.. generally your chess comments are right on the mark. But the above >>>>statement is so far wrong I really don't know where to start, other than suggest >>>>that you look back thru prior ICCA journals and find out who won each of the >>>>ACM events starting in 1987. Here's a hint: 1987, 1988, 1989, 1990, 1991, 1994 >>>>were *all* won by Deep Thought. They didn't participate in the 1992 >>>>computer chess event, and the only other event held was the 1995 tournament that >>>>Fritz won. So out of all those years, they won *every event* except for one >>>>that they participated in. What micro has come even close to that? IE what >>>>micro has won even 1 acm event? Rebel (I think) won in 1992 when no "big iron" >>>>was present. But other than that exception, they've been completely buried by >>>>the deep thought/deep blue guys.. >>> >>>I admit that they were better than other programs because of hardware advantage >>>but I saw some games of them >>>and in at least in 2 games that I saw(I did not see most of the games) they won >>>after the opponent got an advantage after the opening and probably missed a >>>win. >> >>That I wouldn't argue with. However, it proves a different point than you >>intended... that they could get _inferior_ opening positions (because their >>opening book wasn't great in the early days) and _still_ win. In 1994 at the >>ACM event they had to forfeit round 1 because of a power failure at the Watson >>computer center, and they _still_ won the tournament. When you win even after >>a bad opening, or after losing a game to a time forfeit when your computer is >>inaccessible, and you _still_ win the tournaments, you must be doing something >>right, because it can't be all luck... > >I do not think it was all luck but I think they could win bad positions because >of better hardware. >I agree I should not say that the better hardware did not help them but only >that it did not help them the same as I expect. > >The fact that they played bad in the opening in some games because of the engine >proves that IBM did not hire the right people for deep thought. note that playing the right opening moves is by far the _easiest_ part of the game. Building a machine fast enough and good enough to play with Kasparov was a real challenge. They knew that creating a good opening book was not nearly so much of a challenge, and this was obviously put off until the end. However, if you look at each of the ACM events, you will find Genius, Rebel, Mchess, Wchess, Fritz and so forth _all_ playing bad openings on occasion. And you will find deep thought, cray blitz, hitech and the 'big iron' programs doing exactly the same. I can recall several games we played over a 20 year period where we left book a pawn down with no compensation of any kind, yet we were good enough (fast enough) to still win the games. _every_ program plays bad openings at present, although many have modes that at least avoid blunders. > >I know that they hired better people for deeper blue but I still believe they >could do better program with other people. You'd have to know the people. The primary team is at least as good as any other programmers working on chess, and they are actually probably better than most. They aren't dummies. The opening work for the last couple of matches was done by GM players. I doubt there is anyone better (at present) than GM Joel Benjamin to choose openings that fit DB's style of play... > >even if I assume that deeper blue in the same level of the top programs(assuming >the top programs are 2000 times faster) that's impossible to judge. But of course, they spent years of hardware research and development to get that speed advantage... >> >> >> >> >>> >>>I was not impressed by the games. >>> >>>I believe they were better than their oppenents but my impression is that their >>>hardware advantage against micros was something like Pentium400 against 386 and >>>with hardware advantage like this I could expect better results. >> >> >>it is more like a Pentium 400 vs a calculator. They are some 2000 times faster >>than the best micros at present... > >I did not mean to deep blue but to deep thought when I said that it is something >like Pentium400 against 386 >If it is more than that then I am more convinced that other programmers could do >better in the same hardware. As a note, deep thought running at under 2M nodes per second produced a true GM rating over 24 consecutive games to claim the fredkin level 2 prize. I don't know of any program that could do that today. IE fritz only needs to go 3-4x faster to reach this speed. Yet I haven't seen it win a game vs a GM at tournament time controls, much less perform at 2600+ over 24 consecutive games at 40/2 games... >> >> >> >> >>> >>>I believe that Fritz3 (if it was 20 times faster than pentium90 and without >>>losing lines in the opening book) could have better results against the same >>>opponents. >> >>Remember they lost 2 games to computers in 10+ tournaments. On equal hardware, >>fritz didn't do very well in the last two WCCC events. Certainly nothing like >>DT/DB did for a sustained period of time... >> >> >> >>> >>>(I know that Fritz3 lost to Sos in the first round in 1995 only because of the >>>opening book,deeper blue > >I know that I should write deep thought >Sorry for the mistake. >I intended to deep thought(I do not like deeper blue and it may be an >unconscious reason for this mistake) > >>> lost to Fritz3 because of a tactical mistake and not as >>>a simple result of the opening book) >> >>Deep Thought lost to fritz after following a bad opening line, and then playing >>one move on its own and getting into severe trouble. It had a very weak king- >>safety analysis (remember, this was not deep blue II, or even deep blue I, but >>the original deep thought processors).
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.