chessgames.com
Members · Prefs · Laboratory · Collections · Openings · Endgames · Sacrifices · History · Search Kibitzing · Kibitzer's Café · Chessforums · Tournament Index · Players · Kibitzing
Golden Executive / Houdini vs Hosea / Schiller / Wall
CG.com Masters - Machines Invitational (2011), Yahoo Chess, rd 3, Apr-29
Spanish Game: Marshall Attack. Modern Main Line (C89)  ·  0-1

ANALYSIS [x]

FEN COPIED

Annotations by Eric Schiller.      [185 more games annotated by E Schiller]

explore this opening
find similar games 1 more Golden Executive / Houdini/Hosea / Schiller / Wall game
PGN: download | view | print Help: general | java-troubleshooting

TIP: You can get computer analysis by clicking the "ENGINE" button below the game.

PGN Viewer:  What is this?
For help with this chess viewer, please see the Olga Chess Viewer Quickstart Guide.
PREMIUM MEMBERS CAN REQUEST COMPUTER ANALYSIS [more info]

Kibitzer's Corner
< Earlier Kibitzing  · PAGE 4 OF 4 ·  Later Kibitzing>
Mar-10-12
Premium Chessgames Member
  AylerKupp: Since this was the only game that the machines lost, I was curious when they determined that Houdini's position had become worse. While 22.h4 did not seem to be the strongest move, allowing Black to open up lines in White's k-side, all the engines (except for Spike and eventually Komodo which didn't consider it in its top 5 moves) evaluated the position after 22.h4 as no worse than even. Lets see what other engines determined what a better move for White could have been after 21...Qd7.


click for larger view

Time: 07:13:49 00:26:07 36:51:29 09:02:14 01:23:46 25:34:38

Depth: d=32 d=27 d=33 d=28 d=31 d=39
Engine: Critter Houdini Komodo Rybka Spike Stockfish

22.a4 [0.00] - [+0.12] [0.00] [+0.49] [+0.28]

22.g4 [0.00] [0.00] [+0.11] - [+0.32] [+0.32]

22.Qf4 [0.00] [+0.02] [+0.10] [0.00] [+0.26] [+0.24]

22.h4 [0.00] [0.00] - [0.00] - [+0.24]

22.Qh5 [0.00] [0.00] [+0.10] [0.00] [+0.52] [+0.32]

22.Rxe8 - [0.00] [0.00] [0.00] - -

22.Be3 - - - - [+0.23] -

Initially only Spike and Stockfish reached d=31 before I ran out of time and patience, and those were the only two engines which evaluated the position as somewhat favorable for White. So I thought that perhaps there was a magic threshold at d=31 so I re-ran the analyses using Critter and Komodo, allowing them to reach d>31. Conversely, I re-ran the analysis with Stockfish to see if it would reach an epiphany at d>31. But the engines' evaluation didn't change substantially, Critter and Komodo still evaluated the position as basically equal and Stockfish still evaluated the position as somewhat favorable for White, although its evals dropped somewhat, sometimes by half.

So all the engines still considered the position after 21...Qd7 as either equal or no more than slightly favorable for White in spite of its 2 pawn material advantage. So, at least as far as the engines were concerned, while Black has pretty much equalized the losing move for White would have had to come later.

Mar-10-12  ajile: <35.Rh1 f5 36.Kg1 Bf3 White resigned. 0-1>

So what's the finish?

37.Rh2 f4 maybe?

Mar-11-12  SChesshevsky: <<AylerKupp: While 22.h4 did not seem to be the strongest move, allowing Black to open up lines in White's k-side, all the engines (except for Spike and eventually Komodo which didn't consider it in its top 5 moves) evaluated the position after 22.h4 as no worse than even.>>

I guess while some machines might consider 22. h4 even, I can't believe they would consider after 23...h6 even.

I'm thinking White might have been better off with somethng like 22. Rxe8 Rxe8 23. Re1 Rxe1 24 Bxe1.

It looks ugly but gives white some more open lines for the Queen, increases any advantage the extra pawn gives w/o the rooks and maybe most importantly costs Black a tempo to create luft and maybe can weaken the Kside on the dark squares to get some counterplay.

Mar-11-12
Premium Chessgames Member
  Penguincw: I don't think the machine played its best.
Mar-14-12
Premium Chessgames Member
  AylerKupp: <Penguincw> I agree. One thing that I don't think is widely known is that at the time that this game was played <Golden Executive> has the weakest computer of the 3 of us, a laptop with only 2 cores and (I think), only 2 BG of memory, and the slowest clock speed. It also overheated on occasion, as shown by the "Flying Papers" variation (ask <Golden Executive> to explain that one to you). So, even though Houdini might have been the strongest engine at that time (and it probably still is), it was handicapped with the hardware that it was running in compared to <kutztown46> and myself. Therefore, if the masters were going to win a game, they probably had their best chance when playing against Houdini than when playing against either Rybka or Stockfish.

And, since I suspect that I know what you might be thinking, when <kutztown46> and I played against <Golden Executive>, we deliberately configured our computers to use only 2 cores and a smaller hash table to make the games between us as even as possible.

If you want to see another example of humans playing better than the computers, look at my recent post in chessgames.com chessforum, Reshevsky vs Mednis, 1963.

Mar-14-12
Premium Chessgames Member
  Golden Executive: <SChesshevsky: <<AylerKupp: While 22.h4 did not seem to be the strongest move, allowing Black to open up lines in White's k-side, all the engines (except for Spike and eventually Komodo which didn't consider it in its top 5 moves) evaluated the position after 22.h4 as no worse than even.>> I guess while some machines might consider 22. h4 even, I can't believe they would consider after 23...h6 even.

I'm thinking White might have been better off with somethng like 22. Rxe8 Rxe8 23. Re1 Rxe1 24 Bxe1.>

I found in Fritz 12 database this game:

[Event "Leningrad"]
[Site "Leningrad"]
[Date "1987.??.??"]
[Round "?"]
[White "Chandler, Murray G"]
[Black "Nikolic, Predrag"]
[Result "1/2-1/2"]
[ECO "C89"]
[WhiteElo "2595"]
[BlackElo "2600"]
[PlyCount "108"]
[EventDate "1987.05.??"]
[EventType "tourn"]
[EventRounds "13"]
[EventCountry "URS"]
[EventCategory "14"]
[Source "ChessBase"]

1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 O-O 8. c3 d5 9. exd5 Nxd5 10. Nxe5 Nxe5 11. Rxe5 c6 12. d4 Bd6 13. Re1 Qh4 14. g3 Qh3 15. Bxd5 cxd5 16. Qf3 Bf5 17. Qxd5 Rae8 18. Bd2 Bd3 19. Na3 Bxa3 20. bxa3 Bc4 21. Qf3 Qd7 <22. Rxe8 Rxe8 23. Re1 > h6 24. Rxe8+ Qxe8 25. h3 Bxa2 26. Kh2 Qe6 27. Qe3 Qc6 28. g4 Be6 29. Kg3 Qc4 30. f3 Qa2 31. Bc1 Qc2 32. Qd2 Qb3 33. h4 Qc4 34. Qe1 Qd3 35. Bf4 h5 36. gxh5 Qf5 37. Bg5 f6 38. Bf4 Kf7 39. Qe4 Qh3+ 40. Kf2 Qxh4+ 41. Bg3 Qxh5 42. Qb7+ Kg8 43. Qxa6 Qd5 44. Bf4 Kh7 45. Qd6 Qa2+ 46. Kg3 Qe2 47. Bd2 Bf5 48. Qf4 Kg6 49. d5 Qf1 50. Qd4 Qh3+ 51. Kf2 Qh2+ 52. Ke1 Qg3+ 53. Ke2 Qh2+ 54. Ke1 Qg3+ 1/2-1/2

After <23.Re1>:

184611: Chandler,M - Nikolic,P, Leningrad 1987


click for larger view

Analysis by Houdini 1.5a w32 depth 24:

<1. = (0.00): 23...Rxe1+ 24.Bxe1 h6> 25.Qa8+ Kh7 26.Qe4+ Kg8 27.f4 Bxa2 28.Qa8+ Kh7 29.Qe4+ Kg8 30.Qa8+

(draw by three fold repetition)

<2. = (0.11): 23...h6 24.Rxe8+ Qxe8> 25.Qe3 Qd7 26.f3 Bxa2 27.g4 f6 28.Kg2 Kf7 29.Bc1 Qc6 30.h4 Be6 31.Qd3 Bc4 32.Qc2 Bd5 33.Qf5 Qe6 34.Qxe6+ Kxe6 35.Kg3 Bc4 36.Kf2 Kd5 37.h5 Bd3 38.Bf4 Kc4

Final Position:

184611: Chandler,M - Nikolic,P, Leningrad 1987


click for larger view

Draw by three fold repetition

Analysis by Houdini 1.5a w32 depth 24:

1. = (0.00): 55.Qf2 Qe5+ 56.Qe3 Qxd5 57.Qe8+ Kh7 58.Qh5+ Kg8 59.Qe8+ Kh7 60.Qh5+

2. = (0.00): 55.Ke2 Qh2+ 56.Ke3 Qg1+ 57.Ke2 Qh2+ 58.Ke3

Mar-15-12
Premium Chessgames Member
  AylerKupp: (Part 1 of 2) <<SChesshevsky> I guess while some machines might consider 22. h4 even, I can't believe they would consider after 23...h6 even.>

I was thinking of sliding forward to see at what point the engines determined that White was at a definite disadvantage. So here is how some of them evaluated the position after 23...h6, including a "second chance" for Houdini. I forgot to record the times but I let each engine calculate for several hours.


click for larger view

Engine: Critter Houdini Komodo Rybka Spike Stockfish
Depth: d=26 d=28 d=27 d=25 d=30 d=34

24.Rxe8 [0.00] [-0.10] [0.00] [0.00] [0.00] [0.00]

24.Kh2 [-0.05] [0.00] [0.00] [0.00] [+0.13] [+0.08]

24.f3 [-0.24] [-0.26] [-0.40] [-0.16] [0.00] [0.00]

24.c4 [-0.54] [-0.45] [-0.55] [-0.45] [-0.49] [-0.48]

24.Bxh6 [-1.99] [-2.06] [-1.28] [-2.19] - [-2.70]

24.Be3 - - - - [-1.54] -

So some of the engines still see the position as approximately even after 24.Rxe8 (as you suggested), 24.Kh2 (as Houdini played in the game and which it still prefers at d=28), and some even after 24.f3. But it is indicative of the state of White's deteriorating game that the engines are starting to run out of good moves for White so that their 4th and 5th best moves are of the "desperation" variety.

Here are each engine's PV. Spike and Stockfish, which evaluated the position after 21...Qd7 as somewhat favorable for White ([+0.49] and [+0.52] respectively) still consider the position as very slightly in White's favor, although after 24.Kh2 (as Houdini played) instead of after 24.Rxe8. But 24.Rxe8 seems to force a draw since White can give up its 2 extra pawns in return for mating threats of its own. The attacking possibilities of BOC work equally well for both sides!

Critter 1.2: Critter finds nothing better for Black than a draw by repetition since he's in danger of getting mated also. 24.Rxe8 Rxe8 25.Re1 Rxe1+ 26.Bxe1 Bxa2 27.Bd2 Bd5 28.f3 f5 29.Qg6 Bxf3 30.Kf2 Bg4 31.Bf4 Qd5 32.Be5 (one good mate threat deserves another!) Qf3+ 33.Kg1 Qd1+ 34.Kh2 Qc2+ 35.Kg1 Qc1+ 36.Kh2 Qb2+ 37.Kg1 Qa1+ 38.Kh2 Qa2+ 39.Kg1 Qa1+


click for larger view

Houdini 1.5a: With more time to calculate, Houdini switches to 25.Rad1 instead of 25.Be3 and reaches a drawn position with BOC fairly quickly. 24.Kh2 Bb7 25.Rad1 (the game continuation was 25.Be3 which allowed White to double its rooks on the e-file) 25...Qc6 26.d5 Qxd5 27.Qxd5 Bxd5 28.Rxe8 Rxe8 29.Be3 Bxa2 30.Rd6 Re6 31.Rd8+ Kh7 32.g4 Re4 33.Kg3 Be6 34.Rd4 Rxd4 35.cxd4 g6 36.g5 hxg5


click for larger view

FWIW and at least in this analysis run, Houdini preferred 24.Rxe8 over 24.Kh2 from d=13 to d=17. It might have been able to find the draw if the rooks had been exchanged.

Mar-15-12
Premium Chessgames Member
  AylerKupp: (Part 2 of 2) <<SChesshevsky> I guess while some machines might consider 22. h4 even, I can't believe they would consider after 23...h6 even.>

Komodo 3.0: Komodo also finds nothing better for Black than a draw by repetition. 24.Rxe8 Rxe8 25.Re1 Rxe1+ 26.Bxe1 Bxa2 (maybe this isn't best, removing the bishop from it's good post just to capture a meaningless pawn, but what else is there for Black?) 27.Bd2 Bd5 28.f3 f6 (a different attempt to prevent White's queen from establishing itself on e5) 29.Kf2 Bc4 30.Qg4 Qe7 31.Bxh6 (forcing the draw) 31...Qe2+ 32.Kg1 Qf1+ 33.Kh2 Qf2+ 34.Kh1 Qe1+ 35.Kh2 Qe2+ 36.Kg1


click for larger view

Rybka 4.1: Rybka finds an interesting continuation and in the end Black's king is just as exposed (or even more so) than White's. 24.Rxe8 Rxe8 25.Re1 Rxe1+ 26.Bxe1 Bxa2 27.Bd2 Bd5 28.f3 Qe6 29.Kf2 f5 (this move order prevents White's queen from establishing itself on g6, byt Rybka finds a good defense/offense for White) 30.g4 Bc4 31.Kg3 Qe2 32.Bxh6 Qe1+ 33.Kf4 Qxc3 34.Qe8+ Kh7 35.Bg5 Qxd4+ 36.Kxf5 Qd3+ 37.Kf4 Qxa3 38.Qe4+ Kg8 39.Kg3


click for larger view

Spike 1.4: Spike drifts into a clearly drawn position. 24.Kh2 Bb7 25.a4 b4 26.Qc5 bxc3 27.Bxc3 Qg4 28.d5 Rxe1 29.Rxe1 Rc8 30.Qd4 Qxd4 31.Bxd4 Bxd5 32.a5 Bxa2 (the draw is now pretty much inevitable, with BOC and symmetrical pawn position) 33.Be3 Bb3 34.Rb1 Bc4 35.Rb6 Bb5 36.Rb7 h5 37.Re7 f6 38.Kg2 Kh7 39.Kf3 Kg6 40.Bb6 Rc4 41.Ke3


click for larger view

Stockfish: Stockfish also drifts into a clearly drawn position after having Black go pawn-grabbing) 24.Kh2 Bb7 25.a4 b4 26.d5 Rxe1 (this and Black's next move effectively blunt the k-side attack but what other choice does Black have besides 26...Qxd5 exchanging queens or 26...Bxd5?) 27.Rxe1 Qxa4 28.cxb4 Qxa2 29.Re2 Qxd5 30.Qxd5 Bxd5 31.Re5 Rd8 32.Be3 f6 33.Re7 Bf3 34.Ra7 Ra8 35.Rxa8+ Bxa8 (any doubt that this is a draw?) 36.g4 Kf7 37.h5 f5 38.Kg3 fxg4 39.Kxg4 Be4 40.Kf4 Bd3 41.Bc5


click for larger view

Mar-16-12  SChesshevsky: <<AylerKupp>>

Thanks for the great analysis.

I really don't know much about computer engine chess. Forgive me if this isn't the right forum for these questions but..

When the engine shows a numerical advantage like [+.50], what is the scale of significance?

Is [+1.00] meaningfully better than [+.50] or does the number have to go much higher like [+2.00] or more?

and

Does a computer engine ever agree to draw or will it always play out for an advantage?

Mar-16-12
Premium Chessgames Member
  AylerKupp: <SChesshevsky> This is probably not the right forum for general computer engine questions but I will be glad to answer them.

1. The engines' evaluations is expressed in centipawns (hundredths of a pawn) and are typically shown as positive when White is considered to have an advantage and negative when Black is considered to have an advantage. So an engine's eval of [+1.00] indicates that the engine considers White to be the equivalent of a pawn ahead and an eval of [-0.50] indicates that the engine considers Black to be the equivalent of half a pawn ahead. I say "equivalent" because many factors go into an engine's evaluation besides material (mobility, passed pawns, backward pawns, etc.) so it may be possible for White to be 2 pawns ahead but for Black to have the superior position so that the engine considers that Black has the equivalent of one pawn's compensation for his material deficit so White is only ahead the equivalent of a pawn. This game is a good example of that since after 23...h6 the engines evaluate the position to be almost even though White is 2 pawns ahead.

For a example of some of the factors that Stockfish uses to evaluate a position (it's evaluation function), see Morozevich vs D Fridman, 2012.

2. What someone considers to be a meaningful numerical advantage depends on that person's degree of skepticism of an engine's evaluation. I personally don't consider any advantage less than [ 0.50] to be meaningful and don't consider any advantage less than [ 2.00] to be winning, but not only is that just my opinion but it greatly depends on the position. A KNN vs. K endgame will likely be drawn even though White has a 2 knight or approximately a [+6.00] advantage.

This is probably a good time to consider what an engine's eval of [0.00] or something close to zero means. It doesn't mean that the game will necessarily end in a draw (although that's probably what will happen most of the time) but that both sides have equal chances to win. The example that I cite time and time again is the Sicilian Defense, Yugoslav Attack. The engine may give a position an eval of [0.00] but the game is not likely to end in a draw!

3. An engine will typically "agree" to a draw by forcing a draw by repetition or allowing itself to be put in a position where its opponent has a forced draw by repetition since it obviously can't offer a draw by itself. And it will definitely try to force a draw by repetition if it can when it evaluates its position as substantially inferior since a draw is obviously better than a loss. Some engines allow this behavior to be modified somewhat by user-specified parameters. Houdini, for example, has a parameter called "Analysis Contempt" or just "Contempt". Its default value is 1 but if you make it 2 then Houdini will try to avoid a draw as much as it can, unless it considers that all alternate moves will lead to a loss. Conversely, if you set Contempt = 0 then Houdini will try to play for a draw unless its opponent is so weak that Houdini finds a win in spite of itself.

If you have any general computer engine-related questions please feel free to post them in my forum and I'll gladly try to answer them as best I can. I'm somewhat of a nerd when it comes to computer chess engines so I enjoy learning about them an trying to figure out how they work. And I am retired so I have a lot of free time on my hands. The most important thing about using chess engines is that you shouldn't take their evaluations as gospel; you need to review all their suggested moves and convince yourself that some of their unchosen alternative moves are indeed not better.

Mar-16-12  The Rocket: ;"I said it was a "handicap" resulting from Houdini not having access to an opening book">

you are talking in circles.. it's a handicapp not having access to an opening book because it did not have access to an opening book.

I am simply questioning how on earth it's a handicap to let an engine think for itself in a middlegame(positions they are programmed to play- unlike openings).

Mar-17-12
Premium Chessgames Member
  Golden Executive:

<you are talking in circles.. it's a handicapp not having access to an opening book because it did not have access to an opening book.>

No, he is definitely not talking in circles, please read again carefully <AylerKupp>s posts.

<I am simply questioning how on earth it's a handicap to let an engine think for itself in a middlegame(positions they are programmed to play- unlike openings).>

Now that is a new question....which was previously answered:

<What is an opening book? It's really nothing more than a historical record of which moves over time have worked well in certain positions and which moves haven't. Early Ruy Lopez, Marshall Attack games featured the move Bxd5 relatively frequently, but it was gradually discovered that Black gets adequate compensation for his pawn deficit in his two bishops and light square pressure, particularly after White has weakened the light squares around its king by g2-g3. As a result, White typically avoids Bxd5 unless forced.

Without access to an opening book Houdini would not know this, and Bxd5 on the surface looks like a good move, exchanging Black's well-posted knight for White's bishop which, at the time, seems to have limited prospects. So I would call Houdini's unforced 15.Bxd5 a positional blunder, one whose long term consequences would not be obvious and which would not be detectable by today's chess engines without ultra-powerful hardware at the search depths they can reach under the time control used in the tournament>

Of course is not handicap to let an engine think for itself in a middlegame, but what about if the engine entered the middlegame after being handicapped?

p.s. sorry for my bad english, is not my native language

Mar-17-12
Premium Chessgames Member
  AylerKupp: <<The Rocket> you are talking in circles.. it's a handicapp not having access to an opening book because it did not have access to an opening book.>

I didn't say that, but I obviously didn't make myself clear. So let me try again. A human has an opening book in his memory where he knows which moves have been best over time in a given position. That's all that an opening book is. This opening book is the result of many hours of analysis by many strong players, plus the results of many games played. If a computer does not have access to an opening book then it has to try to find the best moves OTB in a limited amount of time. So it's clearly handicapped against a human that has an opening book memorized and the benefit of all that earlier analysis so not only does he know the best moves but doesn't have to take any time figuring them out.

<I am simply questioning how on earth it's a handicap to let an engine think for itself in a middlegame(positions they are programmed to play- unlike openings).>

You originally asked <How can it ever be unfair when a program gets to decide for itself which move to play in a middlegame?>. You are now asking < how on earth it's a handicap to let an engine think for itself in a middlegame>. I said that "unfairness" and "handicaps" are not the same thing and that handicaps are not necessarily unfair. And I never said that the computer is handicapped in the middle game, just in the opening if it doesn't have access to an opening book. And you're right, it's neither a handicap nor unfair to ask the engine to think for itself in the middlegame.

And, in case you were going to ask, the computer is also handicapped in the endgame by not having access to tablebases. The human knows the principles of endgame play such as KNN vs. K is not a forced win; a KBP vs. K is not a win if the pawn is an a-pawn or an h-pawn, the lone king is in front of the pawn, and the bishop does not control the queening square; a KRP vs. KR is usually not a win, nor is it a win if there are additional pawns all on the same side of the board, and many other situations. Without access to tablebases the computer has to once again figure all these positions OTB in a limited amount of time, so it's also handicapped in the endgame. But that's not unfair either.

Mar-17-12  The Rocket: <"So it's clearly handicapped against a human that has an opening book memorized and the benefit of all that earlier analysis so not only does he know the best moves but doesn't have to take any time figuring them out.">

OH apart from the millions of positions already programmed into chess engines? they have tons of knowledge similiar to ours of bad and good positions regardless of a specfic opening book.

Mar-17-12
Premium Chessgames Member
  AylerKupp: <The Rocket> You may have a misunderstanding of how chess engines work. If you eliminate their use of opening books and tablebases then chess engines don't have ANY positions already programmed into them. They must evaluate each and every position based on the factors they use their evaluation function, and they have no idea whether any position is good or bad (or how good and how bad) until they evaluate it. The only "knowledge" they have is by storing the evaluation of each position in their hash table (or at least as many as will fit) so that if during their transversing of the search tree that they form, the don't have to reevaluate a position that they have already evaluated once. This speeds up the search but nothing more.

But, as I mentioned to <SChesshevsky> above, this page is probably not the proper forum to discuss how chess engines work in general, it should be reserved for discussions of this particular game. If you would like to continue this discussion on how chess engines work, I would suggest that we continue it in my forum. I'll gladly answer whatever questions you might have, and I'll be the first to admit that I don't know everything (or even a substantial amount) of how chess engines work. But I will be glad to do some research on a specific detail if I don't know the answer to some of your questions.

Mar-17-12  The Rocket: <"If you eliminate their use of opening books and tablebases then chess engines don't have ANY positions already programmed into them">

some engines do indeed have grandmaster games programmed in in their database.

Such a case was the Deep blue chess engine and I suspect Rybka surely has this as well.

Mar-17-12
Premium Chessgames Member
  AylerKupp: <The Rocket> See my forum at AylerKupp chessforum for a response to this post.
Dec-08-12
Premium Chessgames Member
  Open Defence: <"handicap" resulting from Houdini not having access to an opening book"> the biggest one is time since instead of playing instantly, the engine has to spend time analysing right from move 1 onwards which allowed the human team to discuss plans and variations in advance

if we were playing an engine armed with an opening book it would play the moves instantly leaving the human team in severe time pressue

Mar-13-13  copablanco: So now computer chess programs are giving humans "out of book" odds or "out of database" odds. Chess programs have no foresight. Chess programers replaced that with an evaluation function, which considers a move based on "polling" various board positions , and assigning them a score, then chooses the one with the highest points. The whole thing started with Charles Babbage in the 1800's, and Alan Turing in 1950's who thought that games like chess were ideal models for automated "thinking" or A.I.(Artificial intelligence).
Mar-13-13
Premium Chessgames Member
  AylerKupp: <copablanco> Turing may still have been right in thinking that games like chess were ideal models, or perhaps "candidates" is a better word, for AI. Many think that chess programs and humans approach chess very differently but I disagree; I think that it's just a matter of degree. Chess programs look at, search, and evaluate an immense number of moves and positions and humans are much more selective in the moves that they consider and the positions that they evaluate. But current chess programs must, of necessity, also be selective in the moves they consider and the positions they evaluate since current computer technology (i.e. speed) would not allow then to get to the necessary search depths in a reasonable amount of time. This is where the chess programs heuristics come in, ways of pruning the search tree to get time savings greater than the alpha-beta algorithm alone could provide. So I think that what differentiates humans and computers are the quality of their search elimination heuristics and the quality of their evaluation functions.

At one time it was thought that in order for a computer to challenge a human master it would be necessary to consider many factors in their evaluation function in order to simulate human cognitive processes. In practice it turned out that evaluation functions of medium complexity were adequate, when combined with heuristic searches, to provide dramatic increases in computer program playing strengths. I still feel that in order to have computer chess playing approaches converge with human chess playing approaches (although I'm not sure that this is as much of a good thing as a realization of the limitations of human chess playing approaches) that computer programs' evaluation functions must become more complex and adaptive and that heuristic searches must improve. That would allow more definitive evaluation of positions in the same amount of time by virtue of the smaller number of positions that would need to be evaluated. But, of course, that's just my opinion and I have no way to back it up.

Apr-06-13  zebekias: For the record, Houdini 3 instantly prefers 25. Ra1-b1 and sees it as a draw.

At a ply depth of 23 it evaluates 25. Be3 as -0.17, which ranks as the 8th choice!! lol Chess software has gotten a lot better.

Apr-20-13  Conrad93: 15. Bxd5 practically refuted the entire attack.
Apr-20-13  Conrad93: By the way, Schiller's combination loses by a pawn and a half.
Apr-20-13
Premium Chessgames Member
  AylerKupp: <Conrad93> As usual, you fail to support your assertions. How was it that "15.Bxd5 practically refuted the entire attack?" And why do you say that "Schiller's combination loses by a pawn and a half."

Your credibility would be much enhanced if you provided analysis, either yours alone or supported by one or more engines, resulting positions, etc. to justify your statements.

Apr-20-13
Premium Chessgames Member
  AylerKupp: <zebekias> I don't know at what ply Houdini 1.5a searched before choosing 25.Be3. I remember that <Golden Executive>'s computer at the time was a 2.0MHz, 2-core, 32-bit laptop so it is quite possible that Houdini did not reach d=23 before making its move. Maybe <Golden Executive> still has it in his notes among all his "flying papers" (an inside joke).

Although perhaps not relevant, I tried Critter 1.6a in the position after 24...Bh7. At d=25 here is what it came up with for White's top 10 moves:

(1) 25.Rad1 [0.00]
(2) 25.a4 [-0.01]
(3) 25.Rxe8 [-0.08]
(4) 25.Rab1 [-0.08]
(5) 25.Be3 [-0.10]
(6) 25.Rac1 [-0.17]
(7) 25.Re3 [-0.17]
(8) 25.f3 [-0.20]
(9) 25.Rf1 [-0.32]
(10) 25.Rg1 [-0.39]

So, per Critter, 25.Be3 is not that bad of a move and overall, it considers the position nearly even. And, at lower plies, (d=16 to d=18) Critter considered 25.Be3 no worse than White's third best move with evals no worse than [-0.05].

Equally non-relevant, I tried Komodo 5 in the same position and it was much more critical of 25.Be3. It went in and out of its top 10 list, ranked as high as 2nd best ([-0.10]) at d=20, dropped to 6th ([-0.13]) at d=21, and dropped out completely from the top 10 at d=22, where the lowest evaluation was for 25.Qxb5 (!) at [-15.83]! And then I think that Komodo crashed, since it stopped counting nodes. So I'm not sure that chess software has gotten a <lot> better. :-)

Jump to page #    (enter # from 1 to 4)
search thread:   
< Earlier Kibitzing  · PAGE 4 OF 4 ·  Later Kibitzing>

NOTE: Create an account today to post replies and access other powerful features which are available only to registered users. Becoming a member is free, anonymous, and takes less than 1 minute! If you already have a username, then simply login login under your username now to join the discussion.

Please observe our posting guidelines:

  1. No obscene, racist, sexist, or profane language.
  2. No spamming, advertising, duplicate, or gibberish posts.
  3. No vitriolic or systematic personal attacks against other members.
  4. Nothing in violation of United States law.
  5. No cyberstalking or malicious posting of negative or private information (doxing/doxxing) of members.
  6. No trolling.
  7. The use of "sock puppet" accounts to circumvent disciplinary action taken by moderators, create a false impression of consensus or support, or stage conversations, is prohibited.

Please try to maintain a semblance of civility at all times.

Blow the Whistle

See something that violates our rules? Blow the whistle and inform a moderator.


NOTE: Please keep all discussion on-topic. This forum is for this specific game only. To discuss chess or this site in general, visit the Kibitzer's Café.

Messages posted by Chessgames members do not necessarily represent the views of Chessgames.com, its employees, or sponsors.
All moderator actions taken are ultimately at the sole discretion of the administration.

This game is type: CLASSICAL. Please report incorrect or missing information by submitting a correction slip to help us improve the quality of our content.

Featured in the Following Game Collections[what is this?]
maestro37's favorite games C89 Marshall
by nakul1964
maestro37's favorite games C89 Marshall
by nakul1964
maestro37's favorite games C89 Marshall
by maestro37
Spanish, Marshall Attk. Modern ML (C89) 0-1 Computer Engines
from motel swim pool splashed Fredthebear C50s-C90s by fredthebear
Spanish, Marshall Attk. Modern ML (C89) 0-1 Computer Engines
from Comp Chomps, Chumps & Challenges CageFredthebear by fredthebear
Three amigos against the Maschine
from 94_-> Middlegames with opposite-coloured Bi~ by whiteshark
Direct attacks 2
by TheDestruktor
C89 Spanish: Marshall Attack [Black]
by chess.master

Home | About | Login | Logout | F.A.Q. | Profile | Preferences | Premium Membership | Kibitzer's Café | Biographer's Bistro | New Kibitzing | Chessforums | Tournament Index | Player Directory | Notable Games | World Chess Championships | Opening Explorer | Guess the Move | Game Collections | ChessBookie Game | Chessgames Challenge | Store | Privacy Notice | Contact Us

Copyright 2001-2021, Chessgames Services LLC