I've been involved in a discussion on another site about the meaning/s of the word "sacrifice". Let me start by giving you my conclusion so you can start thinking about it. I think there's an ambiguity in that when people use the generic term "sacrifice", they mean "giving up material", but when they stick an adjective onto it (Queen sacrifice, Rook Sacrifice, Piece sacrifice), they mean "gave up my adjective, and didn't get the other guy's adjective).

The conversation began over a Tal game in which he gave up his Queen as the first move of a Mate in 3 combination. Is giving up material for an immediate win a sacrifice? Or no? And if so, is there any kind of distinction about what kind of sacrifice it is?


[Date "1965.??.??"]
[Result "1-0"]
[White "Mikhail Tal"]
[Black "Alexander Koblents"]
[ECO "B82"]
[PlyCount "43"]

1.e4 c5 2.Nf3 e6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 d6
6.Be3 Nc6 7.f4 Be7 8.Qf3 O-O 9.Be2 Bd7 10.O-O-O Nxd4
11.Bxd4 Bc6 12.g4 Qa5 13.g5 Nd7 14.Rhg1 b5 15.Qh5 b4
16.Rd3 bxc3 17.Bxc3 Qxa2 18.Rh3 Bxe4 19.g6 Bxg6 20.Qxh7+ Bxh7
21.Rxg7+ Kh8 22.Rhxh7# 1-0


One guy called 20 Qxh7+ a "decoy sacrifice", because it was meant to draw a minor piece away from a square it was covering. Another one argued that since the capture of the queen was forced, that it wasn't a "decoy", it was just a plain old "sacrifice". A third guy said that it was a "sham sacrifice" because it lead to a forced win, and so didn't really give up anything at all. Others objected to the term "sham sacrifice", calling it a personal quirk, not a part of chess terminology.

That Reinfeld game I posted the other day, in which Fred wrote that Black "sacrificed his Queen", for a mate in two illustrates the ambiguity too. Black did "sacrifice" his Queen (he gave up his Queen, and didn't get Whites). But did he sacrifice "material" overall? This is where terminology gets vague. If I give up my Bishop to win a Rook, nobody would say that I "sacrificed" my Bishop (even though in one sense I did). Because people think about a sacrifice as giving up something, AND losing material in the process.

But with a major piece, people tend to forget that. One guy gave an example in which White gave up his Queen for two Rooks, and a back rank mate. He said sure, you give up your Queen, but mate follows immediately, so no sacrifice. One thing he overlooked though, is that, even without a checkmate, in his example, White gives up his Queen (9 points), and gets two Rooks (10 points). So even if no victory were in sight, why should that be called a sacrifice? But yeah, a lot of people would think of that as a Queen sacrifice (because you gave up your Queen without making your opponent do the same). It's not a "material sacrifice", but it is a "Queen sacrifice" (confusing, isn't it?) I'm sure that giving up a Rook for a Bishop would be called a Rook Sacrifice, if a special term (Exchange Sacrifice) hadn't been invented specifically to cover that situation.

Now, imagine a situation in which someone gives up a minor piece for no material at all. People would call that a piece sacrifice. If someone gives up a piece for a pawn, most people would still call that a piece sacrifice, even though he's given up less than a full piece this time. If he gives up a piece for two pawns, many annotators would still call it a piece sacrifice. It's ambiguous, isn't it? In the one sense he sacrificed a piece without getting a piece. But he didn't sacrifice an entire piece worth of value, if he got two pawns for it. So in one sense it's a piece sacrifice, and in the other sense it isn't.

Take Fischer's so-called "Game of the Century". People talk about that as a Queen sacrifice. In one sense it was. He gave up his Queen, but never did manage to get Byrne's queen. It's still on the board in the final position. But in the other sense it wasn't. When he gave up the Queen, he had a forced draw by repetition for it. So, did it become a Queen sacrifice when he gave it up, or when he abandoned his ability to force the repetition? And even in that game, Fischer didn't sacrifice a full Queen. He got two minors and a pawn's worth of immediate compensation, so even thinking materially, he only sacrificed 2 pawns worth of value. But most people would still call it a Queen Sacrifice.

Now, what about when a sacrifice leads to an immediate mate? Like in the original Tal game. Or better yet, to make it really clear, here's a game I played myself yesterday:


[Date "2011.12.16"]
[Result "1-0"]
[White "Petrosianic"]
[Black "Guest"]

1. e4 Nf6 2. e5 Nd5 3. c4 Nb6 4. d4 d5 5. c5 N6d7
6. Bd3 Nc6 7. e6 fxe6 8. Qh5+ g6 9. Qxg6+ hxg6 10. Bxg6++ 1-0


What kind of sacrifice did I make here, if any? I did sacrifice my Queen, in the sense that I gave it up without winning Black's. But did I sacrifice "material"? That depends on what you think the value of a King is. I've seen some books that assign it a value of 0. If that's true, then I sacrificed material here. But I've seen other books that assigned it a value of infinity. If that's true, then I didn't sacrifice anything in any way, shape or form, I just won material, by giving up 9 points (the Queen) to win infinity points. That's not a sacrifice (a material loss). It's a material gain. But still, I think most annotators would call it a "Queen Sacrifice", the same way Fred Reinfeld used that term in the other game. Because it is one in the sense of giving up a Queen without getting one.

On the other hand, would you call it a Queen sacrifice if I gave up a Queen, and got two Rooks, a minor, and three pawns (but no immediate checkmate) out if it?


Now, a side question, about my game. Was 10. Qxg6+ a less-than-best move in any sense of the word? I gave up my Queen as the first move of a mate in 2. But I didn't have to. I could have won with 10. Bxg6+ hxg6 11. Qxg6++ without giving up the Queen. So, should that move be be considered dubious (10. Qxg6+?!)?

Well, let's go back to the idea that a King is worth infinite points. With Qxg6+ I give up 9 points, to win a King (infinite), and a Pawn (1). So the net result is infinity minus 8. With Bxg6+, I give up 3 points to win a King and Pawn (net: infinity minus 2). So, which move is better? Well, I believe (but am not entirely certain), that a mathematician would consider that "Infinity - 8" and "Infinity - 2" are the same number. If that's the case, then the two moves are exactly equal even though they don't appear to be.

I ran it through Fritz, and there's no eval, of course. It just lists both moves as "#2". But it does list Qxg6+ as its first choice. I'm not sure what that means. I don't see any way to argue that Qxg6 is objectively BETTER than Bxg6+. Any superiority it has is in the form of style and panache, which a computer wouldn't take into account.

Does anybody else have any opinions about any of this?

Not really, but I will say that Qh7+ was not a sacrifice, but part of a combination. And a combination is a combination of more than one tactical ideas.

Qxh7+ is definitely part of a combination. I'd agree with that completely. But if someone wrote "Tal sacrificed his queen to deliver checkmate", I don't think I'd disagree.

But if a King is worth infinite points, then the phrase "sacrificed his queen to deliver checkmate" is a contradiction in terms, if we assume that "sacrifice" means "a combination that loses material for some greater gain". I do think of Tal's move here as being a "Queen Sacrifice". But it's not a speculative sacrifice, it's part of a combination that leads to an easily foreseeable mate.

Here's an example of a completely speculative Queen sacrifice:

Byrne-Spassky. 1974

I really LOVE this game. This is Game 3 of the Spassky-Byrne Quarterfinal Candidates Match. The first two games were drawn. Spassky senses Byrne's nervousness, and takes adantage of it with a Queen sacrifice with 19...exf3 that sets Byrne problems that he just isn't ready to try and solve.

In the actual game, Byrne returned the Queen with 19. bxa6. Did he do the right thing? Or was he just not ready to meet the challenge Spassky offered him? I'm still not sure. But it's a great game. Spassky at his best.

But not many games are this good. When Reinfeld writes that Black "sacrificed his Queen" when he did it for a forced mate, it's hard to disagree, because he's describing an extremely counter-intuitive move.

A big question here is whether the King is worth 0 points or infinite points. If it's worth 0 points, then it makes perfect sense to say that giving up the Queen for a forced checkmate is a sacrifice.

But maybe it makes sense to say that it's a sacrifice even if a King is worth infinite points. Chess began as a war game. Even though it's value as a simulation is nearly zero, we still think of it that way. And if it's like a war, then we feel that we'd like to not only win, but bring our army home as intact as possible. In a real war, how often would you have to give up your most valuable division in order to win the war? You might send them into a battle of attrition that would chew them down to nothing just the same, but you wouldn't just allow them to be destroyed.

I think the problem is that the terminology is inadequate. I think of both Tal's move and Spassky's move as being Queen sacrifices. But they're not the same kind of sacrifice. We've seen four kind of Queen Sacrifices:

1. Tal-Koblents: Giving up the Queen for a forced mate.

2. R. Byrne-Spassky: Completely speculative Queen sacrifice.

3. D. Byrne-Fischer: Giving up the Queen for a clear positional advantage.

4. Petrosianic-Guest: Giving up the queen for a forced mate, but unnecessarily, since a Bishop sacrifice would have done the job just as well.


I'd maintain that all four of these are Queen Sacrifices of a sort. But not of the same sort.

It strikes me that writing about chess, many authors see chess as a metaphor for battle. In the spirit of that metaphor any piece that leaves the board "dies" in a way. If that "death" leads to victory, I guess the author has the poetic license to call that a "sacrifice."

It also seems that some of these writing date back to times before the computers came into chess. In those times, I am guessing that any analysis carried a shade of doubt that it might just be wrong. For that reason, combinations that were a bit longer and not so forced were probably more difficult to evaluate (a lot of Tal's games fall in this category) than the absolute certainty of computer analysis that we get today. In the spirit of that uncertainty, some moves may have been dubbed "sacrifices" where today we will call them "white wins" or "white messes up."

Just a couple of random thoughts on the word

Yeah, I think the war metaphor has something to do with it. Generally speaking (pun intended), a general wants to bring his army back home as intact as possible, and not give up his best brigade even in victory. But the terminology is still fuzzy. In the war analogy, exchanging your queen for the other guy's queen could be called a sacrifice (couldn't you have won without giving up that unit?), but no chess player would call an exchange a sacrifice.

For my money, all of the above examples are types of Queen sacrifices, I just think they need some kind of sub-group modifiers. I've seen of games like Tal-Koblents, where annotators said something like "White sacrificed his Queen for checkmate", and never before heard anyone say that that was a contradiction in terms.

The question is what modifiers to use. The other guys used terms like "sham sacrifice" and "decoy sacrifice", and I'm not wild about either one. "Decoy" describes the purpose of the sac, but not really its nature. As for "sham", that seems to imply that it's not really a sac because the sacrificer gets something for it. But that's the point of all sacrifices, isn't it? To come out ahead in the end?

In the Byrne-Spassky game, I'd call that a Positional Sacrifice. Black gives up the Queen, purely for positional compensation, with no checkmate or regaining of material in sight. I'd maybe call Fischer's sacrifice in Byrne-Fischer the same thing. Queen sac for positional compensation (and a forced draw if he'd wanted it). In neither case was the opponent absolutely required to even take the Queen at all.

Tal's sac in Tal-Koblents, I'd just call a plain old Queen sac. Maybe you could argue that if the King is worth infinite points, he gains material, but the King is never actually captured in chess, is it?

If the term "Sham Sacrifice" should be used at all, it would be for my game. I gave up the Queen for checkmate, but did it purely for showmanship value, since the Bishop sac would have done exactly the same thing. Tal HAD to give up his queen to force that checkmate, but I didn't.

Speaking of my game, there's a well-known trap in Bird's Opening with a similar motif, but where Black mates:

1. f4 e5 2. fxe5 d6 3. exd6 Bxd6 4. Nc3?? Qh4+ 5. g3 Qxg3+ 6. hxg3 Bxg3++

Spielmann made a distinction between sham sacrifices and real sacrifices. Shams led to a foreseeable gain, real sacs were unclear. If the solution is somewhere beyond the horizon then it must be real.

I'm not sure computer analysis enters into it. Lots of real sacrifices lead to positions where one player is up material but his position is worse, so his losing chances are greater. I think computer analysis only would make a difference with sham sacrifices.

The problem I have with the term "sham sacrifice" is similar to the problem I have with using "Dwarf Planet" to describe a non-planet. Is a sham sacrifice a kind of sacrifice called a "Sham sacrifice"? Is it a non-sacrifice that looks like a sacrifice?

The kind of sac that you say Spielmann considers genuine, are what I'd call Positional Sacrifices. A type of sacrifice, but not the only kind. Maybe I'm wrong, I've just read too many annotators who said that so-and-so sacrificed something to deliver checkmate to be able to abandon the term entirely.

But if a King is worth 0 points of material, that works.

I think the problem is that there is no handy term to describe the planned or willful surrender of material. "Gambit" applies only to pawns in the opening. When you give up a piece for some pawns, some pressure, or maybe some attacking chances, what do you call it? Likewise, if you play the BxRP ch stock sac and it is mate at the end, what else can you call it but 'sacrifice'? Language is letting us down here.

There is a term where you lose material for nothing, and that is 'blunder'.

Sacrifice = "The surrender or destruction of something prized or desirable for the sake of something considered as having a higher or more pressing claim." (from dictionary.com)

From this definition it seems that you can call "sacrifice" just about any act of giving up material for something more "prized" (e.g. mate or other material) or something more "desirable" (positional advantage, making your opponent think, messing around, going home early, etc.). This second part seems pretty much arbitrary since only you know what is desirable.

It is unclear what is the difference in this definition between sacrifice and "trade" - I would think that there needs to be some element of uncertainty in the sacrifice for it not to be a "trade." Which brings us back where we started.

Speaking of that, and on a similar note, you've probably read books that say that the Queen's Gambit isn't a true gambit, because White can win the pawn right back if he wants to. And I have problems calling it a gambit at all for that reason. It's only called that because it's the mirror image of the King's Gambit, which really is a gambit (i.e. a positional sacrifice).

It's not academic, because not only can White win the pawn right back, he does in fact win it back in a move or two in all the main Queen's Gambit Accepted lines. The fact that he doesn't recover it on the very next move doesn't make it a gambit for me. That's a delayed exchange.

Have you ever noticed that, since the Queen's "Gambit" is usually declined, that the books describe the opening as "Queen's Gambit Accepted", when it's accepted. The King's Gambit, on the other hand, which is usually accepted, is usually just called "King's Gambit" (with the "Accepted" left out).


I wonder why the word "Gambit" is ONLY used to describe opening lines? That might be interesting, if queen sacs like the one in Tal-Koblents were called sacrifices, while positional Queen sacs like the ones in Byrne-Fischer, and Byrne Spassky were called gambits. But nobody uses the terms that way.

I think we can say that a "sacrifice" is giving up something in hand for something not in hand, whether it is a piece for "compensation" or a bull for the favour of Mithra, or any other deity or supernatural power. A "trade" is an equal exchange. A "gambit" must be a special subset of sacrifices, referring to sacrifices made in the opening.

It is worth noting that I.A. Horowitz coined the term 'grand gambit' for piece sacrifices in the opening.

This got me interested in where the word "gambit" comes from. Man! I am really hitting the dictionary tonight. Here is what it says "Italian gambetto, literally, act of tripping someone, from gamba leg, from Late Latin gamba"

I guess that at least the origins of the word focus more on the "trick" aspect of the gambit opening than the act of sacrifice.

Also interesting is the question of when the opening is "over" for the purposes of "gambit" being the term rather than sacrifice. There are certainly gambits sacing a piece - like that line in the Petroff or the Pierce Gambit in the King's Gambit

Yeah, I've heard that it was an Italian wrestling term or something, involving tripping or leg holds, or something. But I haven't the slightest idea why something like that should find its way into chess.

But here's something interesting on thefreedictionary.com




I've heard the word use that way. A "conversational gambit" is an icebreaker of some kind, with no concept of sacrifice implied. That usage probably derives vaguely from chess terms. There seems to be something in parlance where all strategy is compared to chess (football commentators saying that it's been a real chess game).

From this definition though, it looks like "gambit" means "a positional sacrifice that's in the book." It would be nice if it could be used to mean all positional sacrifices.

I've never heard the term Grand Gambit, but there aren't many of them, and most of the ones I know are King's Gambit lines: The Muzio Gambit, the Allgaier Gambit, the Rice Gambit. The only one I can think of that isn't in the KG is the Cochrane Gambit in the Petroff. Oh yeah, and the Irish Gambit (1. e4 e5 2. Nf3 Nc6 3. Nc3 Nc6 4. Nxe5). If we were going to use a term like Grand Gambit, I'd want it to include all gambits larger than one pawn (so the Danish Gambit would qualify).

I'm not familiar with the Pierce Gambit, at least not by that name.


I gather that the rule is that if it's in the book, it's a Gambit, if not, it's a sacrifice.

The Pierce Gambit is a fun way of harassing the king in the King's Gambit. I came across it in a book on the Vienna game and seemed fun for 5- minute games but I have not been able to talk opponents into pushing the g-pawn so early and the two or three times when I played it, I lost due to general suckiness.

http://www.chessvideos.tv/chess-opening-database/search/Pierce-gambit

Oh okay, that's similar to the Muzio. Another one of those KGA piece sac lines.

The Fried Liver Attack is another piece gambit that isn't a King's Gambit Line.

I wrote a monograph on three Grand Gambits in the King's Gambit: the MacDonnell, the Lolli, and the Ghulam Kassim. Fun lines, all of them. Not very good though.

MacDonnell: 1. P-K4 P-K4, 2. P-KB4 PxP, 3. N-KB3 P-KN4, 4. B-B4 P-N5, 5. N-B3.

Lolli: 5. BxP ch

Ghulam Kassim 5. P-Q4.

The last one has the best name.

The Jerome Gambit is another Grand Gambit: 1.P-K4 P-K4, 2. N-KB3 N-QB3, 3. B-B4 B-B4, 4. BxP ch KxB, 5. NxP ch. Fun! I've played it about thirty five times.