Originally posted by Fat LadyI'm amazed the Fritz match-up was so high in the Fischer-Spassky match, current wisdom is that the 1972 Fischer would be slaughtered by the likes of Anand, Topalov etc. It looks to me like this is simply incorrect.
I'm amazed the Fritz match-up was so high in the Fischer-Spassky match, current wisdom is that the 1972 Fischer would be slaughtered by the likes of Anand, Topalov etc. It looks to me like this is simply incorrect.
I just don't know what significance these figures have when trying to detect cheats on this site. At the very least they prove that >90% matc ave longer to think about their moves but will often have dozens of games on the go at once.
Higher matchup may not always mean "better play". For example in 1st game of their match Fisher lost but his matchup was higher. Also when higher matchup is result of better quality you should also take into account your opponent`s strenght - its easier to play good against weaker players. So I doubt if Fisher would reach such a matchup (and quality) against modern GM`s.
At the very least they prove that >90% match-ups are possible in OTB games.
In round 11 it was Spassky who reached 100% matchup. And in few games Fisher have reached >90%. So I dont believe that matchup of single games can give overhelming evidence. Which can`t ber said about matchup of many games, selected by objective criteria (consecutivity, particular tournament,etc.)
Result:
Fischer
Top 1 Match: 385/658 (58,5% )
Top 2 Match: 509/658 (77,4% )
Top 3 Match: 563/658 (85,6% )
Fascinating result. Assuming you have analysed out of DB moves you also need to reflect on the fact that Fischer had no DB and that opening theory has moved on a lot since then so he would have left any DB many moves earlier than now. Factoring this in and his match up might be even higher.
Also this is OTB with, what was it generally then, 2h30m for 40 moves and with days to take on each move or even just 15 minutes a move Fischers match up would be much higher.
Now I'm not a GM nor are the top players on this site but if we take players we are 100% (??) sure of like David Tebb and Northern Lad (seeing we know who these 2 fellows are) and give them correspondence time limits against Fischer's 2h30m for 40 moves and I am prepared to bet they would give Fischer a run for his money and regularly win so comparable (if not higher) match ups for such players must be possible.
So match ups at 85% to 90% alone might not prove anything (although maybe 95% to 100% does) and you need to look at the games for elusive engine play (i.e. engine type moves) and these are extremely difficult to find. I think we need to trust that the mods know what they are doing and that the proof they find is conclusive.
Most bannings I have seen seem 100% justified and I have only been concerned about a couple (English Tal for example) but it maybe that those couple were genuinely strong players who were tempted in critical games (Championship finals) and that overwhelming evidence arose in those specific games. For example in a complex ending following a TB exactly with the ONLY winning move for a sequence of 30 moves would be somewhat conclusive of TB use.
Originally posted by CrowleyI always felt the banning of User 282210 may have been iffy.
So, we've either banned cheats or Fischer?
If this guy was 2343 rated Fide Master Stefan Docx as he claimed to be, then a strong OTB player with access to modern databases and 3 days or more to make a single move could possibly get higher matchup %'s than the 60%, 75% & 85% thresholds without resorting to cheating IMO.
Stephane also always used to have very few games in progress as I recall, thus reducing time pressures further.
If Stephane was an imposter with an extreme matchup % then that is a different story.
Whatever the case, as my analysis of Fischer-Spassky '72 shows, we must be careful when making sweeping judgements with regards to matchup rates.
Originally posted by SquelchbelchStephane had very extreme matchup (more than 90% top 3 choices) with rybka. Also from my games against him I can say that he have played very fast making his replies in blitz tempo. So he was obvious cheat.
I always felt the banning of User 282210 may have been iffy.
If this guy was 2343 rated Fide Master Stefan Docx as he claimed to be, then a strong OTB player with access to modern databases and 3 days or more to make a single move could possibly get higher matchup %'s than the 60%, 75% & 85% thresholds without resorting to cheating IMO.
Stephane ...[text shortened]... y '72 shows, we must be careful when making sweeping judgements with regards to matchup rates.
Originally posted by adam warlockThanks for your comments.
...
By the way have you thought about doing the same for some other (more recent) championship matches?
There's not a chance I'll do any more 20+ game matches. I estimate analysis & write-up for these 20 took over 40 hours of my life.
Originally posted by SquelchbelchYes, all they have to do is be as good as Bobby Fischer.
[Event "World Championship 28th"]
[Site "Reykjavik"]
[Date "1972.07.11"]
[Round "21"]
[White "Spassky, Boris V"]
[Black "Fischer, Robert James"]
[Result "0-1"]
[ECO "B46"]
[PlyCount "81"]
[EventDate "1972.??.??"]
1. e4 c5 2. Nf3 e6 3. d4 cxd4 4. Nxd4 a6 5. Nc3 Nc6 6. Be3 Nf6 7. Bd3 d5 8.
exd5 exd5 9. O-O Bd6 10. Nxc6 {Takes game out of book ...[text shortened]... scher achieved this against Spassky, over-the-board nearly 40 years ago![/i]
I note there are a number of the games where his match ups drop into the 60's. Yet there are players here who NEVER drop that low. Guess they are just better players than Fischer.
Originally posted by SquelchbelchA work of art Squelch.
Thanks for your comments.
There's not a chance I'll do any more 20+ game matches. I estimate analysis & write-up for these 20 took over 40 hours of my life.
You will have enjoyed looking at the games - there are great.
I've played over the F/S match games from about 6 different books.
I'm not suprised the OTB rate was high - the man was a genius.
(which shows just how strong these boxes are - constant cheats
will stand out like IQP's - to avoid detection they are going to
have to down-grade).
I see the Box too took the dreaded h-pawn in game one.
Good work Squelch.
Originally posted by greenpawn34Thanks 🙂
A work of art Squelch.
You will have enjoyed looking at the games - there are great.
I've played over the F/S match games from about 6 different books.
I'm not suprised the OTB rate was high - the man was a genius.
(which shows just how strong these boxes are - constant cheats
will stand out like IQP's - to avoid detection they are going to ...[text shortened]... down-grade).
I see the Box too took the dreaded h-pawn in game one.
Good work Squelch.
In game 1 Fritz takes 2:15 to decide that 29...Bxh2?! isn't the top 1 choice.
At 2:39 the top 3 are
1.=(0.10): 29...a6
2.=(0.13): 29...Ke7
3.+/=(0.26): 29.Bxh2
so even then this famous "blunder" doesn't seem so bad.
The line given is 29...Bxh2 30.g3 Ke7 31.Bd2 Kd6 32.Bb4+ Ke5 33.e4 f5 from this position
Originally posted by SquelchbelchThe famous blunder was worked out to a draw although obviously it was not practical.
In game 1 Fritz takes 2:15 to decide that 29...Bxh2?! isn't the top 1 choice.
The issue in game moderation is engine-specific and human-specific moves and plans; although I think match-up rates as a useful criterion, such statistics require appropriate contexts.
The nature of a move or plan must be considered objectively and practically in regards to how the move or plan in question increases or reduces winning possibilities. The evaluation of imbalances and advantages [symbolic, static, dynamic, material, positional/structural, tactical], and subsequent realization of a position I daresay warrants serious consideration in game moderation.
To those with serious intentions with influencing game moderation, I think that reporting suspected engine users with evidence is much more effectual than writing about it in the forums. And what differentiates humans and engines and concordantly what constitutes evidence for engine use should be at least latently obvious to those who have some understanding of them.
[Edit: The statistics should be interesting in the context of the quality of the games; thanks for posting - I appreciate that to obtain the statistics may have taken much effort on Squelchbelch's part.]
This is sort of off topic, but I think this thread may be a good place to ask: Are the novelties GMs play OTB often 2nd or 3rd or worse computer choices(for surprise value)?
Or has anyone noticed that some book moves aren't in the top three? (Specifically, I'm thinking of one of the games Topalov used the Cochrane gambit against Kramnik's Russian Defense. Of course, it may very well be in the top three as the knight doesn't have too many options about where to go, but it's still a surprising move.)
I have a small point to make along with these questions: Computer moves are good moves--there's no doubting that, but we are still very far off from finding the (provably) ^best^ move/s in a position.
And now back on topic: Often, the top 3 computer choices may be within .3 or less of each other in terms of evaluation score. Maybe we should just consider all computer moves within a certain range of each other(evaluation-wise, of course) to be the best moves in the position and test that against a person's play?
Originally posted by Dragon FireAlso this is OTB with, what was it generally then, 2h30m for 40 moves and with days to take on each move or even just 15 minutes a move Fischers match up would be much higher.
he states that is already factored out. he has used database up to year 1972.
and I am prepared to bet they (northern lad or david tebb) would give Fischer a run for his money and regularly win so comparable (if not higher) match ups for such players must be possible.
I don't know what you mean by "regularly winning", but I would be extremely surprised if either of them would have won 1 game out of, say, 30 games in the time controls you have mentioned.
Originally posted by Squelchbelcha very nice research, thank you for the efforts. One comment I would like to make, but hesitate a little, is that one also should not overestimate the possibility of spotting "engine-like" play. here I'll post two games, one supposedly is played between two humans and one by two engines. I don't know about you, but I think the games suggest the opposite of one's probable guess about which pair played which game, at least I would guess it wrongly.
Since I’ve been analysing games I have heard the same questions asked time & time again:
How would the best GM’s fare against Fritz when using the 30 second top 3 matchup system?
How do you know what is humanly possible and what is cheating?
Well the short answer is that I’ve had to previously go on the word of much stronger and more ...[text shortened]... : 28/42 (66,7% )
Top 2 Match: 39/42 (92,9% )
Top 3 Match: 39/42 (92,9% )
1st game:
Originally posted by diskamylsecond game:
a very nice research, thank you for the efforts. One comment I would like to make, but hesitate a little, is that one also should not overestimate the possibility of spotting "engine-like" play. here I'll post two games, one supposedly is played between two humans and one by two engines. I don't know about you, but I think the games suggest the opposite of 8. Kb5b6 Kd7c8 89. Kb6c6 Kc8b8 90. b4 Bf4d2 91. Ba5c7 Kb8c8 92. b5 Bd2e3 93. b6 1-0
[/pgn]
Result:
White: Spassky
Top 1 Match: 19/42 (45,2% )
Top 2 Match: 30/42 (71,4% )
Top 3 Match: 36/42 (85,7% )
Black: Fischer
Top 1 Match: 28/42 (66,7% )
Top 2 Match: 39/42 (92,9% )
Top 3 Match: 39/42 (92,9% )
I am fairly new to chess and absolutely love the game. This discussion seems very interesting as I am currently reading all about the 1972 World Championship Match. I'm not 100% sure what the above statistics mean. Can someone kindly explain?