Page 1 of 2

The same as 10 years ago

Posted: Mon Oct 01, 2012 9:00 am
by Lyudmil Tsvetkov
When I played 10 years ago on a Pentium 200 MHz against engines like
AnMon and Comet, most of my wins would come by using larger structures of fixed (blocked) pawns.
To my surprise, when I play nowadays on a 4 core machine against engines like Houdini and Stockfish, I am still able to win a fair number of games (although much less and altogether few and far between) by resorting to such pawn structures. Maybe 70% of all winning games would come through this.
Of course, I am happy when I am able to win a game, but at the same time I am a bit annoyed, because the fact certainly also points to a less of progress in engine development for an entire decade concerning this specific element of play. How could this actually happen?
My personal view is that nowadays top engines (Houdini, for example) are at the 2700 level positionally, not worse, but not better. This will make 150 points below the best humans.
At the same time top engines are some 500 elo points above the best humans tactically, Carlsen and Kasparov for that reason. This is due mainly to the fact that humans have difficulties visualizing all relevant moves with all piece movements having taken place, maybe not on move 3 or 5, but certainly at move 7 and deeper. There are just too many piece and pawn movements taking place for the human brain.
Surprisingly, the advantage of engines disappears with larger fixed structures of pawns and humans still may enjoy advantage there. I see 2 main reasons for that:
- one is that programmers, although aware of the issue, are reluctant to tackle it, because their testing ground consists mainly of engines and they consider the casual games won by humans in this way as just some kind of chance events.
- and the second one is that the main engine weapons (attacks and mobility) fail them with larger fixed structures on the board.

With bigger fixed structures attacks occur much less frequently, and what concerns mobility, it is simply not the same as in usual situations - with fixed structures less mobile pieces can gradually but forcefully gain mobility, while more mobile ones gradually but forcefully become useless, something difficult to track by the engines moves ahead.

Of course, it is not a big problem that engines lose some games even when multi-headed, but the annoying thing, at least for me, is that engines generally have not made a considerable progress for an entire decade!, while they are much better in almost any other situation on the board. The problem might not appear in engine-engine matches, but it is disgusting to a point to see an engine player 300 elo points above the best humans lose like a novice, and not a single game at that.

I think it is about time that authors took to solving this issue, which could lead to a much improved positional game.
I would dare to propose an easy solution without resorting to subtleties for trying to at least partly eliminate the problem. I think that assigning bonus points for the most advanced pawns of a group of diagonally connected pawns (assigning bonus points in terms of the size of the group, in terms of ranks and files, in terms of fixed structures, especially important, and in terms of closeness to the enemy king, also very important) could be the cure without implementing much additional knowledge - I do not know how many engines do that. At the same time attacks and mobility could be weighted down a bit in relation to the growing number of pairs of fixed pawns, especially when part of a single continuous structure.

Another point that possibly fails the engines is that they do not seem to consider a blocking knight when part of a group of a diagonally connected pawns with fixed structures as an asset, but it certainly is. Such a knight retains good mobility, cannot be attacked by enemy pawns and in a way makes the structure of pawns whole, because the best the adversary can do is try to exchange that knight. But engines seem to think that the side with the knight just has a deficient pawn structure.

I am sorry if I said something out of place, I just wanted to be of little help and hopefully a day will come when humans will not be able to exploit such engine weaknesses.

It is interesting to know what you think of this.

Best regards,
Ludmil

Re: The same as 10 years ago

Posted: Mon Oct 01, 2012 10:06 am
by Vinvin
You pointed out the right things : the main effort from programmers port to engine-engine matches. Improvement focused on human-engine are non-existent. I asked for more than 10 years to have 2 versions per engine (or one parameter) to chose best strength against humans or against engines !

A lot of programmers says "it's easy to do, I just give a bonus to open lines". But very few version offer this parameter ...

Re: The same as 10 years ago

Posted: Mon Oct 01, 2012 10:12 am
by melajara
Very interesting.
To add some substance to this thread, could you post a selection of such recent wins against top engines, e.g. as a set of pgn games?
And if you could take care of adding some annotations on critical moves regarding this topic, that would be fantastic, thanks.

Re: The same as 10 years ago

Posted: Mon Oct 01, 2012 11:25 am
by IWB
Yes it is the same, but it is now quite fair.

You KNOW that you are playing against a computer, it would be just fair if you tell the computer that it is plaing agains a human.

How to do that? Just tell him that a knight is worth less than 2 pawns and that connected pawns are worth even more. It will throw in a knight for a few pawns (sometimes just one) and open your lines.

I made that setting with S12 on playchess years ago and was the first account to have 400 won games (maximum history) in a row agains opponents better than 2600+ (Playchess Elo) in the human room. Many of them tried your attempt and all failed (actually some where angry that I use such "unfair" methods :-) ) Of course the method is objectivily lowering the playing strength, but even S12 was good enough against humans (and some with a 3000 rating and higher)

The best would be a long blind testing long tourney. No one should know against what he is plaiyng ... impossible, I know.

Bye
Ingo

Some games against Houdini 1.5

Posted: Tue Oct 02, 2012 5:27 am
by Lyudmil Tsvetkov
Hi, Mel.
Here some games against Houdini 1.5 (just because this is the engine I recently have played most games against, but wins against other top engines would come in more or less the same way)

Conditions for the games:
4 Core Dell xps laptop 4x1.7 GHz
Windows 7 64 bit
hyperthreading on or off (I am not certain if I have specified 4 threads for these games, as usually the OS will set 8 threads; I am totally clueless on hyperthreading, from what I have read somewhere Vasik Railich claims that manually setting affinities and using hyperthreads would be marginally better, Robert Houdart is adamant that hyperthreading damages performance, most tests show - Sedat Canbaz for example - that the difference would be negligible, and to put a final touch on the issue, on my computer Stockfish sets the number of threads to 5 by default, I do not know why)
Time control would be 2' + 2'' per game, but I allow myself to use more time, maybe 2 times more than the engine overall, as I would hate that my flag falls in a perfectly decent position after say, move 25. I think this is about fair as the opponents are extremely strong and they use multiple cores. Unfortunately, I do not have time for longer games, which are way more meaningful, of course.
Hash tables set to 1024 Mb
No book for Houdini and I usually deviate very early from known theory to make things more interesting
I could adjudicate some games if the position is beyond doubt of final result
Smoking strictly prohibited during games and Houdini will get a loss if its mobile rings during the game :D

Below I tried to post the pgns of 4 recent games against Houdini with some comments and some bitmap images attached, but unfortunately the bitmaps will not show. Maybe someone can advise on a way to proceed with bitmaps. Or maybe there is a way to attach to this message a Word document with the bitmaps?

[PlyCount "86"]
[Event "Blitz 2m+2s"]
[Site "Sofia"]
[Date "2012.08.16"]
[White "Tsvetkov, Lyudmil"]
[Black "Houdini 1.5 x64"]
[Result "1-0"]
[EventDate "2012.??.??"]
[ECO "D05"]
[TimeControl "120+2"]
[Annotator "Tsvetkov,Lyudmil"]
[MLNrOfMoves "43"]
[MLFlags "000100"]

{1024MB, Dell XPS 4Cores} 1. e3 {[%emt 0:00:00]} 1... Nf6 {[%emt 0:00:04]} 2.
d4 {[%emt 0:00:03]} 2... e6 {[%emt 0:00:06]} 3. Nf3 {[%emt 0:00:04]} 3... d5
{[%emt 0:00:04]} 4. Bd3 {[%emt 0:00:03]} 4... c5 {[%emt 0:00:07]} 5. c3
{[%emt 0:00:03]} 5... Be7 {[%emt 0:00:04]} 6. O-O {[%emt 0:00:05]} 6... O-O
{[%emt 0:00:04]} 7. Nbd2 {[%emt 0:00:06]} 7... b6 {[%emt 0:00:00]} 8. Qe2
{[%emt 0:00:12]} 8... Nc6 {[%emt 0:00:04]} 9. Ne5 {[%emt 0:00:04]} 9... Nxe5
{[%emt 0:00:05]} 10. dxe5 {[%emt 0:00:03]} 10... Nd7 {[%emt 0:00:01]} 11. f4
{[%emt 0:00:10]} 11... c4 {[%emt 0:00:00]} 12. Bc2 {[%emt 0:00:04]} 12... Qc7
{[%emt 0:00:00]} 13. Qh5 {[%emt 0:00:16]} 13... g6 {[%emt 0:00:06]} 14. Qh6
{[%emt 0: 00:07]} 14... Rd8 {[%emt 0:00:20]} 15. Nf3 {[%emt 0:00:11]} 15... Bf8
{[%emt 0:00:02]} 16. Qh4 {[%emt 0:00:24]} 16... Nc5 {[%emt 0:00:05]} 17. g4
{[%emt 0:00:13]} 17... Re8 {[%emt 0: 00:07]} 18. g5 {[%emt 0:00:12]} 18... Be7
{[%emt 0:00:04]} 19. h3 {[%emt 0:00:18]} 19... a5 {[%emt 0:00:05]} 20. Nh2
{[%emt 0:00:03]} 20... h5 {[%emt 0:00:04]} 21. Bd1 {[%emt 0: 00:42]} 21... Nd3
{[%emt 0:00:04]} 22. Bxh5 {[%emt 0:00:09]} 22... a4 {[%emt 0:00:12]} 23. Be2
{[%emt 0:00:26]} 23... Ba6 {[%emt 0:00:00]} 24. Ng4 {[%emt 0:00:05]} 24... Reb8
{[%emt 0:00:03]} 25. Qh6 {[%emt 0:00:41]} 25... Qd8 {[%emt 0:00:00]} 26. a3
{[%emt 0:00:26]} 26... Bf8 {[%emt 0:00:04]} 27. Qh4 {[%emt 0:00:06]} 27... Bg7
{[%emt 0:00:14]} 28. Nf6+ {[%emt 0:01:12]} 28... Bxf6 {[%emt 0:00:00]} 29. gxf6
{[%emt 0:00:02]} 29... Nc5 {[%emt 0: 00:05]} 30. Rf2 {[%emt 0:00:18]} 30... Ra7
{[%emt 0:00:07]} 31. Rg2 {[%emt 0:00:03]} 31... Rc7 {[%emt 0:00:03]} 32. Rg5
{[%emt 0:00:13]} 32... Ne4 {[%emt 0:00:04]} 33. Rg2 {[%emt 0:00:08]} 33... Nc5
{[%emt 0:00:03]} 34. Bh5 {[%emt 0:00:15]} 34... Qe8 {[%emt 0:00: 01]} 35. Kh1
{[%emt 0:00:23]} 35... d4 {[%emt 0:00:05]} 36. exd4 {[%emt 0:00:09]} 36... Bb7
{[%emt 0:00:11]} 37. Be3 {[%emt 0:00:11]} 37... Nd3 {[%emt 0:00:00]} 38. Rg1
{[%emt 0:00:09]} 38... Rd8 {[%emt 0:00:02]} 39. Kh2 {[%emt 0:00:05]} 39... Bxg2
{[%emt 0:00:09]} 40. Rxg2 {[%emt 0:00:03]} 40... Rdd7 {[%emt 0:00:00]} 41. Bf3
{[%emt 0:00:18]} 41... Ra7 {[%emt 0:00:07]} 42. Be4 {[%emt 0:00:09]} 42... Ne1
{[%emt 0:00:04]} 43. Qxe1 {[%emt 0:00:05]} 43... b5 {[%emt 0:00:00]} 1-0

A backward-fated pawn part of the king shelter is a tremendous disadvantage


Let us see how black reacts to Ne5...



Leaving the excellent d4 square for the knight is not a wise decision.

wpg5 is not the only winning option here, but now black is absolutely helpless



This is the picture just a couple moves later. A lead pawn on a weak spot of the enemy king position could beat a piece.


Houdini is resorting to some tricks, maybe white will not see the knight on e1 is undefended, and it could even capture the white rook! Well, this could happen in a tournament game with people kiebitzing and making noises, but not now:) wpf5 is also winning

[PlyCount "63"]
[Event "Blitz 2m+2s"]
[Site "Sofia"]
[Date "2012.08.18"]
[White "Tsvetkov, Lyudmil"]
[Black "Houdini 1.5 x64"]
[Result "1-0"]
[EventDate "2012.??.??"]
[ECO "D04"]
[TimeControl "120+2"]
[Annotator "Tsvetkov,Lyudmil"]
[MLNrOfMoves "31"]
[MLFlags "000100"]

{1024MB, Dell XPS 4Cores} 1. e3 {[%emt 0:00:00]} 1... Nf6 {[%emt 0:00:06]} 2.
d4 {[%emt 0:00:03]} 2... g6 {[%emt 0:00:06]} 3. Nf3 {[%emt 0:00:02]} 3... Bg7
{[%emt 0:00:11]} 4. Bd3 {[%emt 0:00:02]} 4... O-O {[%emt 0:00:04]} 5. O-O
{[%emt 0:00:03]} 5... d5 {[%emt 0:00:15]} 6. Nbd2 {[%emt 0:00:07]} 6... c5
{[%emt 0:00:04]} 7. c3 {[%emt 0:00: 02]} 7... Qc7 {[%emt 0:00:15]} 8. Ne5
{[%emt 0:00:03]} 8... Nc6 {[%emt 0:00:06]} 9. f4 {[%emt 0:00:03]} 9... a5
{[%emt 0:00:02]} 10. h3 {[%emt 0:00:14]} 10... Be6 {[%emt 0:00:07]} 11. g4
{[%emt 0:00:04]} 11... h5 {[%emt 0:00:03]} 12. g5 {[%emt 0:00:03]} 12... Nxe5
{[%emt 0:00:01]} 13. fxe5 {[%emt 0:00:06]} 13... Nd7 {[%emt 0:00:00]} 14. Qf3
{[%emt 0:00:34]} 14... a4 {[%emt 0:00:00]} 15. a3 {[%emt 0:00:03]} 15... Nb6
{[%emt 0:00:04]} 16. Qg3 {[%emt 0:00:23]} 16... Qd7 {[%emt 0:00:06]} 17. Kg2
{[%emt 0:00:03]} 17... Rac8 {[%emt 0:00:05]} 18. Rf4 {[%emt 0:00:29]} 18... Qc6
{[%emt 0:00:04]} 19. Kh2 {[%emt 0:00:05]} 19... Na8 {[%emt 0:00:04]} 20. Qh4
{[%emt 0:00:26]} 20... Bd7 {[%emt 0:00:04]} 21. Nf1 {[%emt 0:00:31]} 21... Nc7
{[%emt 0:00:07]} 22. Ng3 {[%emt 0:00:18]} 22... Ne6 {[%emt 0:00: 00]} 23. Rf2
{[%emt 0:00:57]} 23... Bh8 {[%emt 0:00:04]} 24. Nxh5 {[%emt 0:00:11]} 24... Ng7
{[%emt 0:00:02]} 25. Nf6+ {[%emt 0:01:07]} 25... exf6 {[%emt 0:00:02]} 26. gxf6
{[%emt 0:00:02]} 26... Ne6 {[%emt 0:00:11]} 27. Bd2 {[%emt 0:00:11]} 27... Rc7
{[%emt 0:00: 05]} 28. Rg1 {[%emt 0:00:06]} 28... Be8 {[%emt 0:00:05]} 29. Qh6
{[%emt 0:00:00]} 29... Bg7 {[%emt 0:00:12]} 30. fxg7 {[%emt 0:00:11]} 30... Nxg7
{[%emt 0:00:00]} 31. Rf6 {[%emt 0:00:08]} 31... c4 {[%emt 0:00:00]} 32. Rg4
{[%emt 0:00:27]} 1-0

Mating Houdini is not that difficult after all



Avoiding trodden paths at the cost of ugly development


bph5 is definitely not the move to play


wnh5 could not be called a sacrifice, because mate is more or less evident



f6 is a better place for the knight to perish. Black king looks almost like being mated..

Difficult to comment on.


[PlyCount "81"]
[Event "Blitz 2m+2s"]
[Site "Sofia"]
[Date "2012.09.09"]
[White "Tsvetkov, Lyudmil"]
[Black "Houdini 1.5 x64"]
[Result "1-0"]
[EventDate "2012.??.??"]
[ECO "A00"]
[TimeControl "120+2"]
[Annotator "Tsvetkov,Lyudmil"]
[MLNrOfMoves "40"]
[MLFlags "000100"]

{1024MB, Dell XPS 4Cores} 1. h3 {[%emt 0:00:00]} 1... e5 {[%emt 0:00:09]} 2. d3
{[%emt 0:00:03]} 2... Nc6 {[%emt 0:00:07]} 3. Nf3 {[%emt 0:00:04]} 3... d5
{[%emt 0:00:04]} 4. e3 {[%emt 0:00:03]} 4... Nf6 {[%emt 0:00:04]} 5. Nc3
{[%emt 0:00:06]} 5... a6 {[%emt 0:00:04]} 6. Be2 {[%emt 0:00:06]} 6... Bd6
{[%emt 0:00:00]} 7. O-O {[%emt 0: 00:11]} 7... h6 {[%emt 0:00:10]} 8. e4
{[%emt 0:00:03]} 8... d4 {[%emt 0:00:03]} 9. Nb1 {[%emt 0:00:07]} 9... Be6
{[%emt 0:00:12]} 10. Nh2 {[%emt 0:00:06]} 10... O-O {[%emt 0:00: 06]} 11. Ng4
{[%emt 0:00:02]} 11... Nd7 {[%emt 0:00:09]} 12. Nd2 {[%emt 0:00:05]} 12... Qe7
{[%emt 0:00:02]} 13. a3 {[%emt 0:00:07]} 13... a5 {[%emt 0:00:03]} 14. Nh2
{[%emt 0: 00:18]} 14... a4 {[%emt 0:00:00]} 15. Bg4 {[%emt 0:00:04]} 15... Rfd8
{[%emt 0:00:02]} 16. Bf5 {[%emt 0:00:05]} 16... Bxf5 {[%emt 0:00:10]} 17. exf5
{[%emt 0:00:04]} 17... Qf6 {[%emt 0:00:00]} 18. Qf3 {[%emt 0:00:15]} 18... Ne7
{[%emt 0:00:00]} 19. g4 {[%emt 0: 00:04]} 19... Nc5 {[%emt 0:00:02]} 20. Re1
{[%emt 0:00:31]} 20... Nc6 {[%emt 0:00:14]} 21. Nhf1 {[%emt 0:00:07]} 21... Nb8
{[%emt 0:00:07]} 22. Ng3 {[%emt 0:00:04]} 22... Nbd7 {[%emt 0:00:00]} 23. Nde4
{[%emt 0:00:31]} 23... Qh4 {[%emt 0:00:00]} 24. Kg2 {[%emt 0:00:08]} 24... Rdb8
{[%emt 0:00:00]} 25. Bd2 {[%emt 0:00:13]} 25... Ra6 {[%emt 0:00:06]} 26. Rad1
{[%emt 0:00:20]} 26... Rc6 {[%emt 0:00:03]} 27. Rc1 {[%emt 0:00:16]} 27... Rb6
{[%emt 0:00:00]} 28. Rb1 {[%emt 0:00:02]} 28... Re8 {[%emt 0:00:12]} 29. Nxc5
{[%emt 0:00:14]} 29... Bxc5 {[%emt 0:00:00]} 30. Ne4 {[%emt 0:00:06]} 30... Bf8
{[%emt 0:00:07]} 31. g5 {[%emt 0:00:00]} 31... hxg5 {[%emt 0:00:05]} 32. Bxg5
{[%emt 0:00:06]} 32... Qh7 {[%emt 0:00:00]} 33. Kh2 {[%emt 0:00:16]} 33... f6
{[%emt 0:00:03]} 34. Bd2 {[%emt 0: 00:03]} 34... Kh8 {[%emt 0:00:07]} 35. Rg1
{[%emt 0:00:08]} 35... Kg8 {[%emt 0:00:00]} 36. Rg6 {[%emt 0:00:29]} 36... c5
{[%emt 0:00:07]} 37. Rbg1 {[%emt 0:00:42]} 37... Rxb2 {[%emt 0:00:10]} 38. Rh6
{[%emt 0:00:15]} 38... Qxh6 {[%emt 0:00:00]} 39. Bxh6 {[%emt 0:00: 09]} 39...
Re7 {[%emt 0:00:01]} 40. Bc1 {[%emt 0:01:17]} 40... Rxc2 {[%emt 0:00:00]} 41.
Qd1 {[%emt 0:00:02]} 1-0

Again that nasty knight!



Bbf5 is not the right move.


At last white finds the winning move. Look at the structure c2-d3-e4-f5 - does not it look like a single whole?



Let me just once enjoy winning against Houdini in a beautiful way:)
But maybe it was all forced after white's 1.h3?

[PlyCount "62"]
[Event "Blitz 2m+2s"]
[Site "Sofia"]
[Date "2012.09.15"]
[White "Houdini 1.5 x64"]
[Black "Tsvetkov, Lyudmil"]
[Result "0-1"]
[EventDate "2012.??.??"]
[ECO "C65"]
[TimeControl "120+2"]
[Annotator "Tsvetkov,Lyudmil"]
[MLNrOfMoves "31"]
[MLFlags "000100"]

{1024MB, Dell XPS 4Cores} 1. e4 {[%emt 0:00:07]} 1... e5 {[%emt 0:00:03]} 2.
Nf3 {[%emt 0:00:09]} 2... Nc6 {[%emt 0:00:03]} 3. Bb5 {[%emt 0:00:09]} 3... Nf6
{[%emt 0:00: 03]} 4. O-O {[%emt 0:00:04]} 4... Be7 {[%emt 0:00:01]} 5. Re1
{[%emt 0:00:04]} 5... d6 {[%emt 0:00:04]} 6. d4 {[%emt 0:00:04]} 6... Bd7
{[%emt 0:00:02]} 7. d5 {[%emt 0:00: 04]} 7... Nb8 {[%emt 0:00:03]} 8. Bc4
{[%emt 0:00:02]} 8... O-O {[%emt 0:00:04]} 9. Nc3 {[%emt 0:00:05]} 9... Ne8
{[%emt 0:00:04]} 10. Qd3 {[%emt 0:00:05]} 10... Kh8 {[%emt 0:00: 04]} 11. h3
{[%emt 0:00:10]} 11... f5 {[%emt 0:00:06]} 12. a3 {[%emt 0:00:07]} 12... f4
{[%emt 0:00:02]} 13. b4 {[%emt 0:00:06]} 13... g5 {[%emt 0:00:04]} 14. a4
{[%emt 0:00: 05]} 14... h5 {[%emt 0:00:22]} 15. Nd2 {[%emt 0:00:03]} 15... g4
{[%emt 0:00:02]} 16. hxg4 {[%emt 0:00:05]} 16... hxg4 {[%emt 0:00:03]} 17. Nb3
{[%emt 0:00:04]} 17... Nf6 {[%emt 0: 00:15]} 18. Bd2 {[%emt 0:00:00]} 18... a6
{[%emt 0:00:13]} 19. g3 {[%emt 0:00:04]} 19... f3 {[%emt 0:00:03]} 20. Na5
{[%emt 0:00:03]} 20... Qe8 {[%emt 0:00:21]} 21. Nxb7 {[%emt 0:00:03]} 21... Qh5
{[%emt 0:00:11]} 22. Qe3 {[%emt 0:00:00]} 22... Ng8 {[%emt 0:00:02]} 23. Nb5
{[%emt 0:00:12]} 23... Rf7 {[%emt 0:00:07]} 24. N5xd6 {[%emt 0:00:05]} 24...
cxd6 {[%emt 0:00:11]} 25. Nxd6 {[%emt 0:00:01]} 25... Bxd6 {[%emt 0:00:08]} 26.
Qg5 {[%emt 0:00:08]} 26... Rh7 {[%emt 0:00:18]} 27. Qxh5 {[%emt 0:00:11]} 27...
Rxh5 {[%emt 0:00:01]} 28. Rec1 {[%emt 0:00:03]} 28... Ra7 {[%emt 0:00:08]} 29.
Bxa6 {[%emt 0:00:08]} 29... Nxa6 {[%emt 0:00:04]} 30. Be3 {[%emt 0:00:05]} 30...
Rb7 {[%emt 0:00:05]} 31. Ra3 {[%emt 0: 00:05]} 31... Be8 {[%emt 0:00:04]} 0-1

A Ruy Lopez or a King's Indian?


Bkh8 is necessary. bpf5 would be premature because of wne5!


Capturing on f5 looks like the only solution, but Houdini thinks wpa3 is better:)



The consequences are not happy. Difficult to imagine how the white king could escape. Look again at the structure c7-d6-e5-f4


Stoicism is the only word for the white king



Game is over, but soon I will not be able to score another win against Houdini. Maybe in 20 games' time.

Best regards,
Ludmil

Re: The same as 10 years ago

Posted: Tue Oct 02, 2012 5:55 am
by Lyudmil Tsvetkov
Hi Ingo!
You are perfectly right.
When playing against an engine without specific conditions
the human is enjoying some vantage points - he knows whom he is playing against, while the computer does not, he usually knows the style of the engine, he might resort to some well-tested techniques, he is not under a psychological pressure, usually, he spends less physical energy,
etc. But at the same time computers do not get tired, they do not get distracted, they do not miscalculate, usually, are better at shorter time control, etc. So overall some balance might be evident.

I completely agree that trying to complicate the game will serve the computer. Computers are tactically much stronger and humans like to stick to some well-known structures. Even Kasparov would play FRC only after extensive preparation on a given set of positions. Some weaker engines, not to mention Shredder, are really quite more dangerous than
top guns in specific situations. But I am not certain that with longer time control and a piece down Shredder would prevail against stronger opponents. You know what Tal said when hearing that Fischer wanted to play the best women player at the time a knight down, claiming being able to win - 'Fischer is Fischer, but the knight is a knight' :D

Best regards,
Ludmil

Re: The same as 10 years ago

Posted: Tue Oct 02, 2012 10:59 am
by IWB
Lyudmil Tsvetkov wrote:H... You know what Tal said when hearing that Fischer wanted to play the best women player at the time a knight down, claiming being able to win - 'Fischer is Fischer, but the knight is a knight' :D
I disagree with the first part as you use your knowledge that you play against a computer. The compute is not using the fact that you can get distracted or get nervous ... that is a difference.

About the quote above I can only say that I played 400 games with the weakend Shredder without a loss or just draw. Many of the games the human opponents where up a knight for a pawn and not one was won by the 2600+ players ... that is saying enough I guess.

Bye
Ingo

PS; I dnt know if you play the games on playchess. Another thing the comp cant do there is "premove". IT is a simple implementation for human play, but should be switched off against comps as they simply do not have that possibility of the GUI feature (and it has nothing to do with game play)

PPS: The biggest problem with the 400 games in a row was not to get a disconnect as the human opponent imediatly claim that as a win against a 3400+ computer :-) ...

Re: The same as 10 years ago

Posted: Tue Oct 02, 2012 11:22 am
by Lyudmil Tsvetkov
[/quote]I disagree with the first part as you use your knowledge that you play against a computer. The compute is not using the fact that you can get distracted or get nervous ... that is a difference.

About the quote above I can only say that I played 400 games with the weakend Shredder without a loss or just draw. Many of the games the human opponents where up a knight for a pawn and not one was won by the 2600+ players ... that is saying enough I guess.

Hi Ingo.
I do not see what the difference is.
Both are competitive advantages for one or the other side.
When I play against a human, I might get distracted, but my opponent could get too.

Concerning the Shredder games.
If the computer is rated 3400 at playchess, then the 2600-rated opponents are simply too weak. I do not know how this correlates to elo,
but experimental evidence shows that if an opponent is 100 elo points better than you, he can play a pawn down on equal terms. 300 elo difference would mean that the opponent could play on equal terms a minor piece down, etc. Obviously, with an 800 points difference, Shredder could safely sacrifice a queen for a pawn and still win the game.

Best,
Ludmil

Re: The same as 10 years ago

Posted: Tue Oct 02, 2012 12:39 pm
by IWB
It all comes down that the knowledge of the opponent is important.
You think you should have the knowledge but not your opponent (computer) I think this is unfair, you don't. You don't think this is unfair because some other things the computer is better at and this is just to level the chances. I disagree to that as the advantages of the computer are system internal and not given to it because you are a human, while your advantage is given to you because you have human disadvantages. Basicaly to have a proper test setup whereever possible you should give the same knowledge in a field to everyone participating. You break that rule.

There where some human computer games in the past where the humans where given the knowledge about the used book. As this becomes useless the human-computer matches died completly as it becomes ridiculous to emulate equal conditions ... you try to do that again - but it is still ridicolous in my eyes, how about running just on one core, you just have one brain, or how about the computer having to recognize the bord by a camera as you have to use eyes, or how about limiting the calculated positions to 1 per second as you cant do more ...

And yes, the opponents with 2600 were too weak, but there are no stronger humans ... it even won vs the 3000+ players on Playchess and I remember two top 15 players out of the Fide list loosing to that program. If you bring some humans having a fide rating of 3400 I am absolutly willing to play against them with my weakend Shredder even if I expect to lose then :-)

Bye
Ingo

Re: The same as 10 years ago

Posted: Tue Oct 02, 2012 2:37 pm
by Uri Blass
Lyudmil Tsvetkov wrote:

Hi Ingo.
I do not see what the difference is.
Both are competitive advantages for one or the other side.
When I play against a human, I might get distracted, but my opponent could get too.

Concerning the Shredder games.
If the computer is rated 3400 at playchess, then the 2600-rated opponents are simply too weak. I do not know how this correlates to elo,
but experimental evidence shows that if an opponent is 100 elo points better than you, he can play a pawn down on equal terms. 300 elo difference would mean that the opponent could play on equal terms a minor piece down, etc. Obviously, with an 800 points difference, Shredder could safely sacrifice a queen for a pawn and still win the game.

Best,
Ludmil
I disagree here.
Queen against a pawn is a huge advantage and I expect to win against everybody including players who are 800 elo stronger with that material advantage.

I can add that you cannot translate material to elo difference.

I expect a player with rating 1100 to score more than 50% against
a player with rating 1000 with a pawn handicap.

I expect a player with rating 3100 to score less than 50% against
a player with rating 3000 with a pawn handicap.