# A question on chess analysis theory

I developed a chess variant. Being as it’s my own invention there’s no “book” on the opening moves. Being bored during a sixteen hour shift I was analyzing some of the possible opening moves and this led me to an issue of which chess moves are considered worth analysis. While my variant raised the issue to me personally, the issue is valid for regular chess.

Without getting into details, white has a possible opening move. Black can respond to this particular move with approximately thirteen different possible moves. One of these possible moves would lead to a checkmate for black which white could not avoid. So is there any point in analyzing the other possible moves? I see two ways of looking at it:

1. No, don’t bother. You should assume your opponent would make the best possible moves. Once you know he could make a move which would give him the win, you should assume he would make that move. Therefore you should never use that opening move.

2. Yes, explore the alternatives. Just because there is a possible winning move for black doesn’t mean he will automatically play it. Players make mistakes. Of the thirteen possible moves, one might be a definite win for black but the other twelve might be definite wins for white. So if the single winning response for your opponent isn’t obvious, the opening move might be a worthwhile risk.

Interesting question.

I wouldn’t play a move if I was just hoping for a ‘cheapo’ (i.e., hoping my opponent would blunder). So yeah, definitely assume your opponent will play the best move.

That said - if your opponent’s best response doesn’t leave you with a clearly inferior position, it might be a good move to play in the right circumstance. Personally, I’d probably be more likely to play such a move against an obviously superior opponent, simply because I’d be unlikely to beat him the longer the match goes on. But I’m just a woodpusher, so take my advice with a grain of salt <g>

That depends on how obvious the winning response is. But given that you were able to come up with it by yourself and so quickly, I’d say that it’s probably pretty obvious. And certainly, if there ever were to be a book of openings for your variant, that would be in it.

I think being checkmated would qualify as a clearly inferior position.

This particular move is pretty obvious so the opening in question should be avoided. But I was wondering about the general theory. I guess my question comes down to asking whether the “book” assumes knowledge of the book.

It certainly does. The jargon you’re looking for is minimax theory. In a game that has no random chance to it, your goal is a decision path that, even if the opponent makes all their best possible moves, still results in a victory for you, or at least doesn’t result in a loss. You’re not concerned with the possibility of the opponent making sub-optimal moves at one or more steps along this path. By definition, these moves will put you in at least as good of a position (or they wouldn’t be sub-optimal) and usually will be better.

The reason chess openings are so heavily analyzed and debated is not because of differences in game theory, but because of the difficulty of even judging which is the better of two chess positions in the first place. There’s such a vast branching of chess game states that the computing power does not currently exist to exhaustively check all future states to be able to decide with absolute certainty which path has the best outcomes. Since the real answer is out of our grasp, it comes down to human intuition to make estimates, and once you enter the world of opinion, you get differences of opinion too

I think known inferior lines shouldn’t be analyzed (with the exception of determining for yourself why they’re inferior).

Think of it this way. In regular chess there are many inferior openings that will still win or give you positional advantage because your opponent is unfamiliar with them and won’t understand how best to play. It’s true that against these opponents you may do well. But how do you know who these opponents are before you sit down? And if they’re that unfamiliar with the game, then you can just as easily beat them using a strategically sound opening.
And yes, you can come back with “but no one yet knows that this is inferior.” The key word is “yet.” They’ll figure it out eventually and when that happens you’ll have wasted time trying to master a position you already knew wasn’t that good to begin with and has now become completely useless.

This is why for a long long time, computers just weren’t up to the task of playing chess against a human. Computers would take the board in and examine all their moves by brute force, computing out every iteration to 9 moves and then ranking them. Humans can look at a position and say “moving my rook would be ridiculous, so i’m not going to examine that.” and poof! an entire useless branch of analysis vanishes instantly.

As a practical matter, there is value in considering objectively inferior or dubious moves as an element of a winning strategy. One famous example of such a strategy in the opening is the gambit in Karpov-Kasparov, Games 12 and 16 of the 1985 World Championship Match. Kasparov as Black played a somewhat dubious opening and “got away with it”, achieving a draw in Game 12 and a crushing win in Game 16.

This isn’t to say you should regularly play dodgy moves; adopting such strategies are highly situational. For instance, one often-taught example is that when you feel you have a superior position, you should aim to play solid and principled moves that steer the game towards a clear win ("…and the rest is a matter of technique"). However, when you have an inferior position, you should consider playing moves that complicate matters and offer your opponent a variety of choices and potential ways to go wrong, even if such a move is not the best move from a purely theoretical perspective (“When you are losing, you have nothing to lose”).

It has to; it would be far, far too long otherwise. In a typical position in a chess game, there are around 30 possible moves for each side. So if you were to go even four moves deep, you’d have to look at about a million different sequences of moves, a number clearly too large for any literal book, and starting to strain the resources of even a computerized “book”. A few moves past that, and you couldn’t even fit the entire book on the largest hard drives we have. Clearly, you have to trim that number down very aggressively, and you do that by completely discarding the bad moves.

I think you should drop the inferior lines. That is not the same thing, however, as dropping the suboptimal lines. Sometimes suboptimal is better for a human player. As an example, I just sac’d my queen for his last remaining pawn. Why? Because it left me with 2 pawns against nothing. That’s an easy win for anyone at my level. Was it the best line? No, but it was the easiest to see.

With my queen on the board, I could only see maybe 4 moves ahead. When I saw the queen sac, I saw victory 40 moves ahead with ease. So I think you should analyze the “valleys” if you can see the “peak” on the other side. But drop the lines where you find yourself falling off a cliff.

The difference is that in your life, you opponent can punish you by playing the best moves. In mine, I can play a bad move because the opponent doesn’t have the option of punishing me. He can only accept his fate.

Does any of that make sense? It’s hard abstractly discussing theory with someone outside my own head.

I guess the way I see it is that chess is not a thouroughly explored game. You cannot pre-determine the outcome of any given position on the board. There are some that might be proven to lead to a win for black or white but this is not true for most positions; the eventual victor cannot be determined.

So the players are not playing perfectly - they do not know which of them will win and they are not just going through the moves down to a foreordained conclusion. The game is played on the basis that neither player has a complete grasp of what possibilities exist and therefore may defeat his opponent by knowing something the other player does not or by superior play which leads to unforeseen opportunities.

And that’s where I see the gap arising. I can see it as a legitimate argument to think “I may know a way to defeat this move if I were in my opponent’s position but that doesn’t necessarily mean that he knows it. So maybe it’s a good move against my opponent even though it wouldn’t work against me.”

The book is written on the assumption that players are perfect and equal but the game is played on the assumption that players are not perfect or equal.

I’m far from a chess expert (I was hoping glee would have come in by now, but, come to think of it, I haven’t noticed him around much lately), but I used to read the columns on chesscafe.com, and one of them (called “Novice Nook”, written by an American chess instructor named Dan Heisman) used to repeat the principle that many beginners play what he called “hope chess”. This is defined as making a move and hoping that the opponent will not see the best response to it. He could not emphasise enough that that was not the way to play the game, and I agree. It follows that it is not normally sensible to analyse such lines.

As someone said upthread, it comes down to the fact that analysis is always trying to find the best line against any response. If the opponent plays anything less than the best response, then that should lead to a position that is at least as good (usually better) for you than if he made the correct response. If you play an inferior move, that implies that he will have a “better” response available.

One of the few times it can be worthwhile playing a questionable move that is less good than “best”, as is hinted above, would be if you are losing the game anyway and the questionable moves offers more opportunities for your opponent to make a mistake than the theoretically “best” move. The technical term for this is a (attempted) “swindle” (really). However, if your opponent sees through it and makes the correct response (as most good opponents will), you will end up worse off than before. The theory goes that that is not important if you are going to lose anyway.

In summary: yes, you do have to always assume that the opponent will play the best response to your move.

Of course, as Kasparov once said, the winner of a game of chess is the player who makes the second-to-last mistake, and there was recently a case where a grandmaster playing against a computer managed to completely overlook a mate-in-one situation and lost from what should have been a decent lead in position. So it’s definitely not unrealistic to think that your opponent might make some mistakes.

But that logic, if carried to its conclusion, says that I shouldn’t analyze 1.e4. King’s pawn openings have been defeated so they’re not “perfect” and should all be avoided.

Except that when you make 1.e4, you’re expecting your opponent to come up with a good response to it, and planning for it (or at least, you should be). The catch is that nobody’s sure what the best response is to that move.

Worst misspelling of “Tartakower” I ever saw.

But I’ll bet Kasparov did say it at some point.

Generally their ratings are good benchmarks for this.

Except midr-level amateurs have often memorized long variations (without necessarily understanding them) that can get them to the middlegame in roughly equal standing, where you should be able to beat them, but not “just as easily”. It’s not uncommon to see Masters in weekend swisses using generic modern-type openings to transpose their way later back into something familiar, or (at the other extreme) uncorking something really odd like the Grob.

I’m not much of a chess player, but I’m wondering what the position was where you could have a queen and two pawns against one pawn (with your experience level being able to know that two pawns against nothing is an easy win), and your opponent wasn’t conceding already.