48
I am not an expert on stockfish source code, but my understanding is the following.
Humans:
It is true, that the 1 piece equals 3 pawns approach is pretty accurate, surprisingly so. However as you are probably aware, when evaluating a position, we consider many other aspects as well, such as piece activity, space, king safety, etc. The difference however ...
38
The answer by MikroDel gives the commonly-used "Reinfeld values" of pawn=1, bishop=knight=3, rook=5, and queen=9 (kings are essentially worth an infinite number of points, because the game ends if it is lost). While this is a good guide, chess is rarely that simple. Many books will give the value of bishops as 3.5 instead of 3, simply because they ...
37
I'm pretty sure Stockfish doesn't have explicit code that handles opposite-side castling. What it does have is:
Some kind of "menace" score for enemy pawns advancing against our king. The closer they get, the more dangerous Stockfish thinks they are.
Some kind of "pawn shield" score for friendly pawns in front of our king. The fewer ...
35
Why do chess engines fail to spot good moves in some positions?
The reason are called forward pruning techniques (see http://chessprogramming.wikispaces.com/Pruning)
with ProbCut in first line.
The move 15...Bf2 is pruned on shallow deeps because it includes the sacrifice 16.Rxf1 without regaining something concrete with evidence immediately after.
The ...
28
This is a fortress and a draw.
The black king can't advance as the rook, shuffling between h3 and f3, prevents it. The only way to break the fortress is to trade the queen for the rook and pawn, but the resulting endgame is still a draw.
27
When I said chess is a zero sum game in that answer, I wasn't referring to anything involving ratings. Obviously if we include ratings then chess isn't strictly a zero sum game, since the gains and losses aren't always balanced out. But different rating systems are arbitrary, and aren't part of the game of chess itself.
Your question should be "Is the FIDE ...
23
It's because 1. exf8=Q+ Kxh7 2. Rxd7?? would be stalemate. I think it is a even a theoretical draw after 1. exf8=Q+ Kxh7 2. Qg7+. Therefore 1. exf8=B+! (with check!) is better, since white then can keep an extra piece and win easily.
[FEN "5r2/3qPbkB/8/7P/8/8/8/1K1R4 w - - 0 1"]
1. exf8=Q+ (1. exf8=B+!) Kxh7 2. Rxd7? (2. Qg7+)
22
Individual pieces:
Pawn - 1 point
Knight - 3 points
Bishop - 3 points
Rook - 5 points
Queen - 9 points
Piece combinations:
Rook and Knight - 7.5 points
Rook and Bishop - 8 points
Pair of Rooks - 10 points
Three minor pieces - 10 points
Rook and two minor pieces - 11 points
N.B. The values may vary because of circumstances, but these are the basic values.
22
This position is a standard classical fortress. There is no way the Black color can go through as long as White just shuffle pieces, wait and do nothing silly.
Stockfish is a computer algorithm, it has no intelligence. It doesn't know it's a draw unless it searches for all the possibilities, but it's practically impossible.
Connect to a 6-piece tablebase, ...
21
There is a very easy way to detect whether King and Pawn endgames are drawn or not.
This method I use is a very easy to understand one from Karsten Mueller and Frank Lamprecht's excellent book Secrets of Pawn Endings
It concerns key squares and opposition
The rule states that if pawn has not reached or crossed the central line (5th rank for White and 4th ...
20
If the engine can choose between getting mated in 2 or mated in 3, it'll choose the line where it is mated in 3 (even though the mate in 2 might be 'more difficult' to spot for humans).
It can't really set traps, because it doesn't know what things might be difficult to spot for a human (or other engine) opponent. It just evaluates the position, without ...
19
Pawn - 1 point
Bishop, Knight - 3 Pawns
Rook - 5 Pawns
Queen - 9 Pawns
The evaluation depends on the position.
In some situation you will find it equal or good to give you Rook and Pawn (6 Pawns) for Bishop and Knight (6 Pawns). But it is also possible that two light pieces are more valuable than Rook and Pawn.
The value of pieces given to you will be a good ...
18
The Site ratings at slow time controls can be quite reliable for servers where strong players congregate (ICC, FICS to name a few) as the ratings VERY closely reflect your true playing strength if you've played enough games. For very standardized rating systems such as USCF and FIDE/ELO, you will notice that the different rating classes tend to point to the ...
18
Those aren't scores; there are just three possible scores in chess: White wins, Black wins, and draw.
Those are evaluations of the chess engine; a score of +0.62 means the engine thinks White's position is better by 0.62 pawns. A negative score would mean Black is better. On professional level, a score of 1.5 pawns is in most positions decisive, i.e. enough ...
17
Is there any established system for evaluating positions taking into account time for both players?
No, there isn't. This is for two main reasons:
There is no objective measure for how time affects a player's ability to play any given position.
Such evaluations would be useless. They would literally tell you nothing useful
Let me give an example from a ...
16
Interesting question. I think it depends on how much bottom-up intelligence the engine has. For example, AlphaZero was given no explicit heuristics, but was able to infer plenty of strategy by playing itself millions of times and learning that way.
An engine with explicit heuristics can also exhibit additional strategies you didn't program into it. As a non-...
14
There's a great analysis/article about this by GM Larry Kaufman available here.
To summarize:
Pawn = 1
Knight = Bishop = 3.25
Bishop Pair = 7(+0.25 for each bishop.)
Rook = 5
Queen = 9.75
There's also a lot more detail in the article about what situations favor which groups of pieces. For example, when B+N is better than R+P, or when Q+P is better than R+R,...
14
The best way to normalise a centipawn score to [-1,+1] range is using a sigmoid function, as that closely approximates the likelihood of a given centipawn advantage converting to a win, and avoids the need to identify a strict maximum or minimum. This is discussed here.
14
Chess is a zero sum game. There is 1 point available and it is divided over the players, what one gains the other loses. It is a simple concept with a simple answer. Tournament rules mention some very rare cases where the result is 0-0 or 0.5 - 0, but in my opinion those are "outside the game". That a mobile phone may ring is not really a matter for the game ...
14
[FEN "r4rk1/2p1qppp/1p3n2/p7/2B4B/4P3/PPQN1Pbb/2KRR3 w - - 0 1"]
1. f3 {traps the bishop} Rad8 2. Re2 Bxf3 (2...Bd6 Rxg2 Kh8 Rh1 {the Black king is not safe at all}) 3. Bxf6 Qxf6 (3...gxf6 Rxh2 {checkmate on h7 will follow}) 4. Rxh2 {threatening mate on h7} h6
5. Rf1 {the bishop is now pinned against the queen and will be won shortly} b5 6. Be2
...
13
I am trying this with Houdini 1.5 (the free version). At first 15...Bf2 doesn't show up in the top 5 moves, then when it reaches depth 15, it does, as #1. Takes about 19 seconds on my four year old computer, not that long.
So it doesn't fail to spot it, instead it only considers it the best move once depth 15 is reached. Apparently, as long as it only looks ...
12
A good chess engine won't stop after a predetermined number of moves, but will keep looking until the position is "quiescent", which roughly speaking means that there are no pending captures or checks. See Quiescence Search in chessprogramming wikispaces for a more detailed explanation.
11
White does indeed fork the king & rook after Nc7+ - but after Nxa8, the knight is not escaping, so Black is at worst even on material.
[FEN ""]
1. d4 c5 2. Nf3 cxd4 3. Nxd4 Nc6 4. Nc3 e5 5. Ndb5 Bb4 6. Bg5? Qxg5 7. Nc7+ Kd8 8. Nxa8
After something like ...b6 followed by Bb7, the knight's a goner. White can get at most one pawn for the knight (after ...
11
I agree with the other two answers, but I feel like I also need to comment on the beginning of the game since there was A LOT more important stuff there than just the answer to why Qxg5.
This is very similar to a Kalashnikov Sicilian by transposition with the exception that the knight is on c3 instead of a pawn on e4, but the ideas are very similar. In ...
11
Just because Stockfish thinks a move is best doesn't mean it is best. But let's see why Stockfish might like it.
When I look at Stockfish's static evaluation of that position, and move the king from c1 to b1, I see four sub-evaluations change. In order of the size of the change, those sub-evaluations are:
Psqt bonus. This gives a bonus to each piece based ...
10
If I understand correctly your x axis is the moves. Your script implied Anand and Prag were blundering every time they make a move,
Clearly you have a bug in your script. Stockfish engine always give you a score relative to the player making the move, NOT white. You need to multiply the score by minus one if it’s Black to move.
Please do this and you will ...
10
First, I don't think that there is any doubt that black has significant compensation, but clearly, black is still trying to hold this, not win it when talking two computers playing each other. I believe that between two humans, I would probably prefer black.
If you have an eval that says +.63 for white, but black is down two pawns, you have to realize that ...
10
Please note that those values are "abstract", later to be modified by the specifics of the position. For example, even though a knight appears 0.8 pawns less valuable than a bishop, it could be that bigger bonuses are awarded to well-placed knights than for well-placed bishops, turning the balance around.
It's also worth noting that the "3 pawns equal a ...
10
Unless you explicitly hard-code these endgames, or provide the computer with endgame tablebases, they will apply what they've always done throughout the game: search lines and return the static eval (which is basically the evaluation function of the engine, evaluated on the final position). The static eval isn't smart enough to realize that this game is a ...
9
Though one cannot trade one's king for other considerations -- and in this sense the king cannot be evaluated -- the king still has a practical strength as an attacking and defending piece in the many concrete positions in which no immediate mate is in view -- especially during the endgame. This strength can indeed be evaluated. World Champion Emanuel ...
Only top voted, non community-wiki answers of a minimum length are eligible
