Two Plus Two Forums
Chess with Randomness

10-28-2021 , 01:08 AM
This has probably been thought of already. But maybe not.

Every time it is your move you must specify two alternatives and then flip a coin to decide which one you use. Or maybe one of the moves is specified as 60% vs 40%. Or maybe you must specify three possible moves. Only exception would be if you are in check and only one (or two in the 3 choice game) moves gets you out of it.

If everyone started playing that game tomorrow how many of the top 100 in normal chess would also get into the top 100 of this randomized game the day after tomorrow? How long before a computer became the champ?
10-28-2021 , 01:18 AM
Notice that it need not be the two or three moves that would be best in a regular game that would be the best choices in this game.
10-29-2021 , 02:56 PM
Playing it doesn't appeal too much, though maybe you weren't suggesting that it would. Big problem is that sometimes there's only one move and everything else just loses, while in many quiet positions there are several playable moves with not much to choose between them.

A kind of silly but fun game that had a vogue about 50 years ago and has almost disappeared as far as I know is Vegas Fun Chess, utilizing two dice that determine which piece or pieces you can move, with two exceptions, one if you're in check, the other if your last move gave check. The latter exception allows you to carry out forced checkmates. In those cases you don't roll the dice. When you roll the dice if neither shows a piece you can legally move you lose your turn. You can play it with two ordinary dice of different colors, to allow for a slight difference in which pieces are on each die. In the good old days they made special dice with chess pieces engraved on the sides. These usually came in a plastic hemisphere attached to a spring attached to a suction cup which attached to the table so you could shake the dice by pulling back the spring and letting go. Here's a link to the rules. https://www.chessvariants.com/dice.d...sfunchess.html
10-29-2021 , 07:01 PM
Quote:
Originally Posted by gimmeabailout
Playing it doesn't appeal too much, though maybe you weren't suggesting that it would. Big problem is that sometimes there's only one move and everything else just loses.
So I would think the idea would be to avoid those situations. I bet a champion backgammon player rated 2300 would probably be a favorite over many chess grandmasters.
10-30-2021 , 08:58 PM
It would be so extremely frustrating any time you had one move available that is checkmate but had to pick some other non-winning move too and flip a coin to see whether you win or whether the game continues. I have trouble imagining anyone experienced in chess being willing to put up with that frustration long enough to put any real energy towards optimizing how to deal with it probabilistically speaking.

And common ideas like forced mate-in-three patterns that start with a sacrifice would be almost completely unplayable. 50/50 whether you make the sac, and then if you do and after your opponent makes their only legal response, you need to win two more coinflips in a row to convert the mate and if you lose either of them you end up in a dead lost position because of the sac.

I mean I do see that if you could get past all that, there are some interesting calculations to make about which moves to submit (including some moves that would be bad in normal chess but might be good here). Captures that would normally be an equal trade would be highly worthwhile to attempt (if your opponent only has one piece that could recapture) because there's a chance they lose their coinflip and have to play something OTHER than the recapture and you just win the material straight up. And moves that come with one-move threats that are easily parried in regular chess but can only actually be defended against in one way also become more effective because if they don't play the right defense...

I could imagine finding tactics where the goal is specifically to play a move that sets up checkmate in two different possible ways, rather than just one, so that if your opponent fails to defend it you can definitely convert. It's a marginally interesting exercise to spend a few minutes on. But playing a full game sounds interminable, much less actually trying to play games regularly and get good at the variant. Put me in the not-much-appeal camp.
10-31-2021 , 04:09 PM
Quote:
Originally Posted by BobJoeJim
It would be so extremely frustrating any time you had one move available that is checkmate but had to pick some other non-winning move too and flip a coin to see whether you win or whether the game continues. I have trouble imagining anyone experienced in chess being willing to put up with that frustration long enough to put any real energy towards optimizing how to deal with it probabilistically speaking.

And common ideas like forced mate-in-three patterns that start with a sacrifice would be almost completely unplayable. 50/50 whether you make the sac, and then if you do and after your opponent makes their only legal response, you need to win two more coinflips in a row to convert the mate and if you lose either of them you end up in a dead lost position because of the sac.

I mean I do see that if you could get past all that, there are some interesting calculations to make about which moves to submit (including some moves that would be bad in normal chess but might be good here). Captures that would normally be an equal trade would be highly worthwhile to attempt (if your opponent only has one piece that could recapture) because there's a chance they lose their coinflip and have to play something OTHER than the recapture and you just win the material straight up. And moves that come with one-move threats that are easily parried in regular chess but can only actually be defended against in one way also become more effective because if they don't play the right defense...

I could imagine finding tactics where the goal is specifically to play a move that sets up checkmate in two different possible ways, rather than just one, so that if your opponent fails to defend it you can definitely convert. It's a marginally interesting exercise to spend a few minutes on. But playing a full game sounds interminable, much less actually trying to play games regularly and get good at the variant. Put me in the not-much-appeal camp.
How about the coin flip is every fifth move?
11-19-2021 , 01:16 PM
heads najdorf, tails dragon seems like a fun flip
11-25-2021 , 11:22 PM
Quote:
Originally Posted by David Sklansky

If everyone started playing that game tomorrow how many of the top 100 in normal chess would also get into the top 100 of this randomized game the day after tomorrow? How long before a computer became the champ?
I think it would alter play so dramatically, that the subtle differences between the super-elite GMs and the other GMs wouldn't mean so much. Long calculations would mean almost nothing as you wouldn't be able to rely on playing all of those moves. Not to mention the meta of avoiding a coin flip that is deadly for yourself. So I think many of the top 100 might very well drop out.

m