Anatomy of a Mate in 18

In a post about a month ago, I alluded to a Mate in 17 position that was on my profile. (It's actually Mate in 18; I think the "Mate in 17" comes from that being technically what you'd announce (if you were so inclined, although I don't think it is the done thing these days) upon making the first move in the sequence). I said I wouldn't blog about it, because I didn't want to give away my opening preparation. This is partly true, but also it's because that position is antithetical to a maxim that I subscribe to: One's ability to "memorise" an opening directly correlates to how well one understands the ideas and themes of said opening - unless you have a photographic memory, straight memorisation is practically impossible. For reference, this is the position:

Yup, this is opening theory! But more to the point, I am initially presenting it as a position and not a puzzle. You can treat it as a puzzle and try and find the Mate in 18 yourself - in fact, I'll do that for you in a moment (below). But first, I thought it might be worthwhile looking at this position through the lens of memorisation v understanding. It's a tough ask (and lots of brainpower) to straight up remember the entire variation tree of something 18 moves long.

These are the first 20 moves in the Classical variation of the Caro Kann. I have had this exact position more than once in my life. For all intents and purposes, it could be said I have them memorised. In fact, I once sent a video to a friend reciting the 20 moves without prompts to make a point about how I hadn't proved anything by doing that. (Um... anyway...) The thing is: I have a terrible memory. It wouldn't ordinarily be possible for me to remember these lines. In fact, I played a game as White where I played 11.Bf4.

Not a bad move at all, but not part of my preparation, and I lost my thread and went on to lose after playing it. This game came between two instances of reaching the above Move 20 position. I'll forever remember 11.Bd2 now, as the best way to remember such a move - much like an anniversary - is to forget it once. (So I'm told; I've never forgotten). My point is: straight up trying to recall a variation, I wouldn't stand much chance. I remember the moves because I remember the ideas and plans. I know h4 and h5 get played, but not consecutively. Nf3 is the only move that makes sense to be played in between - that threatens Ne5, so the response to Nf3 is Nd7, and so on.

How does this distinction help with the Mate in 18? Well, for a start, don't look for Mate in 18 - just try to get Checkmate as fast as possible. The below is a plausible way that Black might defend, and you mate in 5 moves:

Mate in 5 seems much more manageable! Obviously, "Mate in 17" does not guarantee it will take as many as 17 more moves if you get this position against a human that doesn't know what the optimum defence is. In the puzzle above, Black only failed to act optimally on one move. The first move was almost completely forced (1...Ke8 is legal, but loses immediately to Nc7#), but on the second move, Black has options. 2...Ne7 keeps Black in the game for 10 moves, but is still not "optimal". Pick up the puzzle after the obvious 3.Qxe7+:

Like I said, Mate in 10, but for purposes of the puzzle (and lack of spoilers), I still haven't quite given Black the optimum defence - yet. But the starting position of the above (2nd) puzzle is worth another look. The first move in the solution is the only move that leads to a forced mate, yes, but it is also the only move that wins at all! White has a SLIGHT (0.5) advantage if Qc5, but otherwise, White is not even winning. White had sacrificed a lot of material to reach the starting position (albeit Black had given some of it back), so failure to force mate will fail to land a win at all.So far, Black has not lasted longer than 7 moves, when they could (apparently) have lasted a whole lot longer. But this is why I say the position is antithetical. If you're going to have learned enough that you can get to this Mate in 17 position, it makes sense that once you are there, you'd be sure to remember what to do, because otherwise you are just a rook down. It becomes easier if you understand the main mating patterns of the first few moves, because that's normally as far as you'll have to go. But for the next puzzle, let's switch sides and see how hard it is to find the optimum moves for Black:

This is the combination that keeps Black (just about) alive and kicking for longer. First, Black creates some luft in case the King is forced Kingside. But the 2nd move - it just looks like a way the computer likes to lose, doesn't it? Throw away material to delay the inevitable. But no; the computer's method is a genuine attempt from Black to turn this around! If White takes the Queen, they are winning, but they have lost the forced checkmate sequence. The fight continues. This gives White a dilemma: do they accept the Queen, and take the +6.6 evaluation? Or do they play for the forced win? And here's the rub: For the next three moves, assuming White does not capture the Queen, if White does not find the move that continues the forced Checkmate sequence, Black is winning (by a distance). Therefore, as White, you wouldn't want to play into this unless you were sure you knew the sequence, would you? Or you were sure you could figure it out at the table, but that takes some guts. Let's see that play out from the start:

They might not be the world's hardest to find moves, but you have still played yourself to a position where failing to complete the win will be embarrassing (particularly if you had the audacity to announce "Mate in 17" at the start of the sequence! ). Actually, even if you didn't, if this is part of your opening preparation, and you therefore know that you have a forced win, it's got to be all the more frustrating if you don't execute it. OR, you could be left to rue not taking the Queen, as one move further down the line, Black tries to trade them off:

Obviously, as White, you are not going to accept this! But again, you're going to lose the forced mating sequence if you don't know how to proceed from here. So, where does this leave us?

(a) White has to be familiar with the initial sequence, and then:

(b) White must decide if they are confident enough not to accept the free Queen when it is offered. If White does not know/remember the sequence, they should probably take the Queen, unless they back themselves to find the moves in the time they have available.

(c) Assuming White has not bailed out with Qxh2, they need to find the only moves that win for a few moves, before they emerge from the forest and...

(d) In the final stages, the stakes are not quite as high. Black will actually offer up the rook, so White has the choice of slow-rolling by capturing it, or going for the quicker win. However, the ending is not trivial - fxg2 is still a threat, so White can't afford to switch off.

By now, I suppose the whole thing seems kind of easy? Here it is, anyway:

I suppose it could be argued that every combination of combinations in this sequence could be understood as opposed to directly memorised, thus not being an example of an antithesis at all. However, that neglects one key factor: you cannot reach the Mate in 18 position without first having memorised the sequence that gets you there! What I've shown here might not be the most densely clustered woods, but this only all comes after sacrificing....

THE ROOOOOOOOOOOOOKS!

But since how to get there is genuinely a slog through counterintuitive (and not always optimal) moves, I think it is fair to say that part of the theory is outside the scope of this post, which means I get to keep it secret. I say secret... I'm sure I can't be the first person to have discovered this line. But in case anyone is doing their opposition research on me, I'm not just going to give it away. Next week (hopefully): anatomy of a lesson, where I actually will share some opening preparation, if I get the green light from one of my students.