When chess moves are logical but don’t make sense

My brother was once playing chess against a computer. He was winning the game, though both sides had only a few pieces left.

The computer had managed to get one of its pawns almost to the other end of the board — hoping, of course, to send it all the way and get it promoted to a queen. But my brother had gathered his forces, which had now gone and got the pawn cornered.

The pawn was one step away from the last square — but it was too late. If it moved forward, it would get captured. If it didn’t move forward, it would get captured anyway.

“Might as well give it a chance,” the computer thought. So it sent the pawn forward…and promoted it to a bishop.

A bishop? Yes, a bishop. Not a queen, but a bishop.

Why would anyone ever do that?

Computers don’t always think the way people think. They think the way people tell them to think — and that’s not always the same, because people don’t always know how they think (even if they think they know).

Computer-games — that is, games designed specially for the computer — aren’t symmetrical. The computer plays differently and the players play differently. They don’t have to do exactly the same thing. Computer-games have their own challenges, of course, but thinking like a human is not one of them.

Board-games are different: you have to take an already existing game, and tell the computer how to play it. Not just what to play, but how to play it. That’s the difficult part, which can end up with playing strategies quite different from what humans would do.

How do you play chess?

I don’t mean the rules about how the pieces move, or how to set up the board. Those are relatively simple to explain. No, what I man is the strategy of the game: how you actually think while playing it.

I mean the way you look at a bishop, and see diagonal lines radiating out. The times you look at your opponent’s pieces, and notice a knight-fork just waiting to happen. Moments when you recognise a strong pawn-formation, or notice it’s in danger of being broken. And, the general feeling of the way the game is moving.

When I first learnt to play chess, all my moves were along the lines of “Where can this piece go? Where can this piece go? Are there any immediate threats? Okay, I’m moving this piece and hoping for the best…”

The time when I actually “saw” how to play chess was much later, when someone was showing me how to checkmate with only a king and a rook, when the other side has only a king.

It’s annoying if you don’t know the strategy, because you keep moving your pieces towards the enemy king, hoping to corner it, and then suddenly His Majesty escapes through a gap and there you have to go all over again.

The actual strategy, as I later learnt, is to focus not on where the king can go but where the king can’t go. (It reminds me of playing volleyball: you’ve got to aim the ball not at the other team’s players, but at the empty spaces between them. That way, they have to run around more and can’t get the ball by simply standing in the same place).

In this case, the places where the enemy king can’t go is nothing but the places your own pieces can go. So, if we take your rook, for example…

…then the highlighted lines are the enemy king’s “no-go” zone.

Similarly, your king creates a “no-go” zone of all the squares directly next to it: a big square surrounding the little one.

All you’ve got to do is manipulate those “no-go” zones until the enemy king is completely trapped.

That’s basically all there is to it. (If you want to know the full strategy, leave me a comment and I’ll let you know).

But the most important thing I learnt in this was the ability to visualise “no-go” zones — or, more generally, the places every piece can move to. Now, I have to but look at a rook, to instantly see horizontal lines radiating out — and stopping, of course, when a piece comes in the way. The same goes for bishops, pawns, and, indeed, any other piece. (Knights, as you can guess, need a bit more practice).

Now, I can plan my strategies a bit better. I don’t need to ask “Where can this piece move? Where can this piece move?” because I can actually see where they move.

The rules of chess are simple. Just different moves for each piece, and a few special ones like castling, promotion, and en passant.

But, out of those rules come many more patterns and strategies; many curiosities that were never designed, only discovered.

The Skewer, where one piece is under attack, but cannot move because moving will expose the other, more valuable, piece behind it. The Knight Fork, where a knight attacks two pieces at once: in the next move, one of them will have to go. The Discovered Check, where the king is attacked not by the piece that moved, but by the piece behind it whose path was blocked until the current moment.

And I haven’t even mentioned the openings, gambits, formations, and recognised move sequences, or the combined attacks and pieces combinations, all of which let people play with their own individual style. Some people may prefer sending their queen out in front; others push forth their two bishops, to weave their way through the battle as a team.

Computers, poor things, can never “see” how to play. They need to always ask “Where can this piece go? Where can this piece go?” every single time.

The quickest way to write a chess-playing program is to make it look at every single move possible, and every single move that can come after that, and go on until the end of the game. Then, they can see which move has the best ending for them, and move their piece accordingly.

Of course, that’ll take too long, even for a computer: to make their first move, they’d have to analyse every possible chess game that ever has been, will be, and can be played! That’s why most chess programs only think a few steps ahead — maybe three or five moves into the future.

The computers give each move a “score”, depending on things like the position of the board, how many pieces are under attack, and how many have got captured. (When calculating their opponents’ moves, they assume the opponents will play the best possible moves too — and a high “score” for the opponent is counted as a low “score” for them).

Nowadays, some modern computers probably use “machine learning” algorithms to update and fine-tune the exact way the score is calculated.

Be that as it may, what finally happens is that the computer picks the move with the best “score” and plays it. There’s no “feeling” or “seeing” or “hoping” or “noticing”. Just a plain, logical, “Which move puts me in the best position?”

Usually, it’s very effective. Computers nowadays can play as well as human players, if not better. But even though they play the same moves, the ways they choose those moves are very different. And that difference can show up in unusual ways.

Why did the computer promote its pawn to a bishop instead of to a queen? We may never know the real answer, but we can make a guess.

My guess is that the logic went something like this: “My pawn will most probably get captured anyway. But if it doesn’t get captured in the next move, then a promoted piece will be more able to escape, so I’d better promote it anyway.”

So far, so good. But then comes the point-calculation: “What should I promote the pawn to? After my move, it will most probably get captured. So if I promote it to a queen, then I’ll lose a queen — but if I promote it to a bishop, I’ll only lose a bishop!”

And so, the pawn was promoted to a bishop.

Have something to say? At Snipette, we encourage questions, comments, corrections and clarifications — even if they are something that can be easily Googled! Or you can simply click on the ‘👏 clap’ button, to tell us how much you liked reading this.