Author: Robert Hyatt
Date: 13:32:15 06/12/01
Go up one level in this thread
On June 12, 2001 at 05:07:01, Mark Young wrote: >On June 11, 2001 at 23:21:20, Robert Hyatt wrote: > >>On June 11, 2001 at 08:36:04, Mark Young wrote: >> >>>On June 10, 2001 at 01:24:56, Robert Hyatt wrote: >>> >>>>On June 09, 2001 at 22:03:39, Mark Young wrote: >>>> >>>>>. >>>> >>>> >>>>Sorry but that is a bit too sarcastic for my tastes. I don't believe _anybody_ >>>>has said computers are not "GM-level" at fast time controls. These two games >>>>were 30 0. We had a 30 0 event on chess.net two years ago with 4 computers >>>>and 4 human GM players. The 4 computers finished in the first 4 places. >>>> >>>>This is old news. >>>> >>>>40 moves in 2 hrs is something different. >>> >>>Why? Crafty may not be a 2500+ elo program, but there is a program called Deep >>>Junior (You may have heard of it) that has played the best players in the world >>>at 40 moves in 2 hrs at Dortmund and walk way with a rating of 2700+. >>> >>>Please explain why this result and others 2500+ results by other programs don't >>>count. >> >> >>Several reasons... but the #1 reason is that the computers are simply _not >>there_ yet. Yes, they will produce a 2700 rating here. But then they also >>produce a 2100 rating there, but _nobody_ points out the 2100 results because >>that doesn't show the point they want to make. >> >>If you think computers are 2700, fine. Your opinion. I happen to disagree >>at present. I don't think they are 2700, or 2600, and for the most part I >>don't think they are much if any over 2500. >> >>I've produced way too many of these 2600+ results myself, yet I _know_ what >>my code has and lacks. And in some cases I don't lack anything that is in >>the commercial programs, in some cases I do. But so long as a program plays >>a game where a GM says "2100" it is _never_ going to be legitimate to call >>that program a 2500+ program. A human GM just will _not_ play a game like >>that. They _will_ make an occasional tactical blunder, to be sure. But they >>won't play a series of moves that leaves a real GM chuckling... >> >>That is what I base it on. I watch these programs play on ICC all the time, >>I watch them enter dead lost endgames, saying they are winning. I watch them >>struggle to keep the bishop pair when one bishop is so bad you could replace it >>with a pawn and not notice the difference. I watch them fiddle around while >>their king gets mercilously attacked. I watch them fight for a rook on the 7th >>when the opponent's king is beyond the 7th and there are no pawns there. The >>list goes on and on. Just ask a good GM on ICC to give you a quick set of 2-3 >>_major_ deficiencies in a program he just played, whether he won or not... > >I?m sorry but your arguments don?t hold water. Elo ratings are not negated by >the chuckle factor of grandmasters. Computers are not human and will defend any >position no matter how hopeless and that can make all of us chuckle. The fact >still remains Elo is determined by wins and losses and draws, not how funny a >grandmaster thinks the program is playing? It may be determined solely by wins and losses, technically. But in reality, that isn't enough because if the "problems" get taken advantage of, things go wrong, quickly. > >The fact remains that the top programs today when playing grandmasters at 40 >moves in 2 hours are beating the grandmasters. Where are the 2100 results you >are taking about when the top programs are playing grandmasters at 40 moves in 2 >hours? How many games were played? Where are the game scores? > >The best you can do is pull out a loss and say look how stupid the computer >played in this game, but this does not negate the Elo rating achieved by the >programs. > >If your arguments are correct why are the computers besting the grandmasters >almost 2 to 1 after 49 games? >Is the game data incomplete, if so show the complete data record for us to >examine. The data is definitely incomplete. You only see the data from very visible events where computers are just rising up over the threshold of "Ho Hum... another computer fish to play". We sneak by a long way on tactics. But only until the human player says "enough"... which happens more and more frequently... But the fact remains, the programs have _many_ "holes". Many more holes than the human GM players have. _I_ can see them. The humans can see them. Most authors that are _honest_ can see them. > >If your only response is to say some anonymous grandmaster thinks that the top >programs are only 2100 and we are suppose to turn off our own brain and ignore >the date as if they were gods your dead wrong. I didn't post the 2100 story, someone else did. He wasn't an "anonymous GM" either. All you have to do is talk to them politely on ICC. They at least answer my questions reasonably honestly, IMHO. > >Let me clue you in Bob, the grandmasters have made many statements about >computers and there abilities that have been shown to be dead wrong. How many >times have we heard the grandmasters say no computer can solve this position, or >no computer can understand that position, only to be show the computer programs >are up to the challenge? Easy to explain... based on "time". The last time the GM looked, it couldn't. With hardware advances coming like they have, what was true 3 years ago isn't necessarily true today. But that just addresses tactics. There are _still_ plenty of positional/strategical things the machines are lost at... > >The point is grandmasters make many mistakes on and off the chessboard. If this >is the basis of your augments that we are to suspend the game data and only rely >on the word of this grandmaster or that grandmaster then your argument is >thinner then a anorexic Ethiopian. > >> >> >> >>> >>>1. Junior was lucky? >> >> >>In a way, yes. But just take his account on ICC and find an IM that >>beat him a couple of games in a row. Then find a GM that that IM beat >>a couple of games in a row. Then talk to the IM about the difference >>in the two. > >What games has Junior played at 40 moves in 2 hrs on ICC against Grandmasters? >But if you like we can look at Junior's ICC record... If you think that will >help your case. if the ICC search command comes back, look for standard games to see who has played it. ADOLF (IM Brian Hartman) used to play it from time to time at longer time controls... > >Information about ban(C) (Last disconnected Sun May 27 2001 18:06): > > rating [need] win loss draw total best >Bughouse 1566 [6] 15 29 0 44 1630 (12-Jan-1996) >Bullet 2186 [8] 1 0 0 1 >Blitz 2766 1472 1190 645 3307 3138 (14-Oct-1999) >Standard 2701 [3] 41 20 20 81 2735 (24-Nov-2000) > > 1: Junior - The 15th World micro computer chess champion (1997-1999) > 2: Shared Third Place in World Computer Ch. HK '95 > 3: developed in the state of Israel > 4: By Amir Ban & Shay Bushinsky > 5: 2 x 750 Mhz > 6: running Deep Junior6 or experimental Junior versions > 7: sometimes manually operated sometimes not > > > >> >> >> >> >>> >>>2. Did Mr. Ban payoff some of the GM's to lose to the Deep Junior? >>> >>>3. Other programs could not have achive such results so we can ignore this >>>result? >> >>No.. please feel free to pick all the _good_ results and harp on those. Don't >>take into consideration all the _ugly_ things that happen. Just pick the good. > >I did not pick any results, that what you are doing. I am showing all wins and >losses, and the fact remains the computers are winning more games then they are >losing. IE you run the 100 yard dash in 9.0 seconds today, again tomorrow, then the third run takes you 20 seconds, and you are still happy? A good sprinter doesn't have that kind of variability. The computers still do... >> >> >>> >>>4. Computer programs still have weaknesses, and as we know human Grandmaster >>>don't, as a result computers can not be considered Grandmasters in strength >>>regardless on any results? >> >>GMs have a tactical weakness that the computers can exploit. So long as the >>GM players play into positions where tactics abound. When they don't, then >>your statement is essentially correct. > >I Know...Its just hard to understand why the programs are not Grandmasters, when >the grandmasters are having so much trouble winning with all the computer >weaknesses... > >>> >>>5. Grandmasters have very fragile egos, and it is in some of our interest to >>>placate the Grandmaster community to have continued access? >> >> >>Has nothing to do with anything, IMHO. > >Maybe........Maybe not... > >> >>> >>>6. Total results don't count as long as computers are still beatable and we can >>>show losses to prove this point? >> >>Until you have played chess long enough to understand the issue >, there is little >>I can say to change your mind. > >Here it comes...Can't beat the message so attack the messenger.:P >I have only been playing chess for 31 years, I started playing when I was 4. >How much longer do I need to play until I understand wins, losses, and draws.:) > I'm simply saying that I know enough to recognize (a) a good C programmer from (b) an outstanding C programmer. (a) might be all a business needs. But I can tell them apart after 5 minutes of talking. You only have to look at a few human vs GM games. The kramnik match will be interesting to see how my feeling holds up. Kramnik is only 100 points "better" than what many think the programs are now doing. Let's see how the final score goes. if it is "close" I will probably change my mind. If it is "ugly" I won't. > If you ask _any_ programmer that knows something >>about chess, or _any_ GM that knows something about computers, you will discover >>they feel similar to what I feel. > >Really, give me names....Other then Bob Hyatt. Because if you are trying to say >that there is a majority of experts that agree with you I would like to know who >they are, and when did you have this meeting to vote on this issue? Ask Ed. Someone asked him a year or two ago. But rather than spoil the punch-line, drop him an email to see what he thinks... > > Unless you ask a programmer that has a motive >>for "stretching" the truth a bit (ie sales). > >Please please...Name the programmers that are liars, because they disagree with >you Bob, and are saying this just for money. I don't have to "name them". It is _obvious_ who they are... :)
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.