There's this interesting anti-computer chess variant described on ChessBase news site:

http://en.chessbase.com/post/option-...by-paul-bonham

Sounds interesting, I tried the problem and it's amazing.

Hard to take the guy too seriously when one of the important rules is

I prefer this variant of that rule:

7. On their opponent's time, a player may move as far away from the chessboard as they want. After making a double move, if one's opponent is more than two arm lengths away from the chessboard, the player is required to shout "BOOMSHAKALAKA" at the top of their lungs in order to alert them.

The article as written is hard to take seriously because of too much discussion of the mechanics of how to play this variant over the board (rather than just a focus on the actual rules of the game) - particularly considering that it's meant as an anti-computer variant. And the tangent about possible ways that the tokens could be made (you could use a 3d printer!) was hilariously unnecessary.

All that said, the actual rules of play seemed somewhat interesting, and the problem was a nice demonstration of how the game would work. My concern is that 12 options seems like a lot, though. Yeah, conserving them to a degree will matter, but you can hammer out double moves from move 9 through 16 and still have four options left. If your opponent DOESN'T use options on all those moves, you'll probably build up a position overwhelming enough that your remaining four options can finish it off. And if your opponent does use his options in response, then it didn't really hurt you because you have just as many options left as he does.

I think starting the game with only like 3-5 options would make it more strategic in terms of figuring out when to use them. The more safely spammable the double moves are, the easier it will be for engines I think. It shouldn't be very hard to add a search function of "if I use all my remaining X options every move for the next X moves, is there a forced mate available in the sequence" to existing engines, with minimal modification, and when X can be as high as 12 the engines will find a lot of ridiculous crushing mates where they burn through all their options and it doesn't matter that they have none left because they have a forced win. So again, that would defeat the purpose.

I also worry that the existence of these double moves might possibly increase the first-move advantage to the point where (particularly with 12 options available) white could well have a forced win in the initial game state. Once again, that would be something that would make the game easier, not harder, for computers to play.

Of course the rules are interesting. But there are probably a million rulesets for 'option chess' that are somewhat like this and sound interesting, and only a few thousand of those that lead to a fun, balanced, playable game.

Making good games is hard. Any idiot can say 'wouldn't it be cool if this game had a move analogous to en passant', but it takes genius or play testing to realise whether this mechanism improves the game. The guy clearly hasn't play tested it and clearly isn't a genius, so the game is most likely crappy and unbalanced, with a chance it's outright broken.

The idea that computers would be bad at this game, without testing, is also laughable. Agree that with spammable options, there's a good chance computers only get better at finding forced mates in the middlegame. And with non-spammable options, why is a heuristic for valuing 'number of options left' any harder to come up with than one for, say, king safety?

Yeah, I agree with all that.

For clarification I should add that when I say the rules are "interesting", what I mean is that I think they're well conceived enough that it's worth spending some time to actually analyze why they wouldn't really "work" to achieve the anti-computer goal. Basically every one of these new "anti-computer" variants that someone invents tends to be something a computer could actually handle just fine, this one included. The only difference between them is that some variants are so bad it's not worth even taking seriously, while others are at least good enough to be worth looking at and taking some time to debunk. This one is in that second category for me.

"Interesting" definitely does NOT mean I'm a convert who will now begin proselytizing for the "option chess" variant

This guy understands the question of humans vs computers a lot better. Very politely points out why the previous writers are full of ****.

http://en.chessbase.com/post/compute...ess-variants/1

Yeah that was a good read and a voice of reason

Good article. I wonder how many chess positions there are nowadays from which the top grandmasters could beat the top computer programs playing either side. Of course that set is shrinking. Human versus computer matches aren't interesting anymore but what if the human was allowed to select the starting position, and then play it once as white once as black against the computer. That might still be interesting. Or, would it be so easy to select anti-computer chess positions that the match would be massively lopsided in the human's favor?

You can still find tons of positions where a GM (or even an idiot) can score 1.5/2 because it doesn't recognize some fortress and the alternative loss is slow enough. Winning both sides would be harder but I'm sure it's doable.