AlphaZero testing new chess rules

And though the video was there, the link to the video was lost, never to be seen by anyone.....

I’ve found more information on this. Apparently Kramnik himself is saying he thinks the self capture variant of chess plus giving pawns the ability to move 2 squares on any move increases the tactical capabilities of the game and offers a richer calculating environment for professional chess players. A word of warning to Lichess administrators and developers...Chess.com is looking to add these new variants to their platform. I suggest Lichess seriously reevaluate their reluctance to adding more chess variants and dive right into this to keep up. The self capture variant appears to be the most promising one right now.

Summary of the "paper".

p(πAdraw < πBdraw) =Z Z I

πAdraw < πBdrawp(πA|GA) p(πB|GB) dπA dπB .

It seems none of these variants lives up to the expectation of making the game livelier.

Oh really? According to a preliminary finding...eliminating the castling rule causes decisive results to go up to 85%! That’s with eliminating the castling privilege alone. Adding the ability to let pawns move 2 squares on any move plus giving it the ability to move laterally one square would make pawns a dangerous element in chess that was usually ignored before. And the self capture idea would allow blocked positions to be opened up allowing more tactical resources to getting at the enemy positions. Add in Fischer Random rules for setting up the pieces minus the castling rule would save the game from draw death. Anything that allows more decisive results should be applauded not reviled. Changes to the game of chess are coming whether we like it or not.

What about instead of changing the rules of how the game is played, change how the outcome is decided?

In competition combat sports including boxing and bjj, a point scoring system comes into play in the case where there’s no knockout or submission after a set number of rounds. In chess we normally call this a draw. But a point system for chess would be interesting to implement. Points would be awarded for a combination of aggression and dominant positions, just like in combat sports. Say, for example, a point for each piece attacked by the piece you just moved. This could also be a handy way of rewarding initiative in chess (if that initiative only amounts to a draw). Go also uses points to make games more decisive, and it’s lasted a very long time.

Why?

People are worried that as pros get better and better at chess, a draw will be the only expected result. Decisive results increase audience interest and engagement.

Imagine if there were draws in sports like basketball, American football, golf, tennis, soccer, boxing, mma, even poker, or esports. None of these highly popular sports feature drawing as a common or even possible result.

I thought it was more than self evident that the reason chess was so drawish is because of stalemate.

If the person getting stalemated lost, then King vs. King and Pawn is always winning. Queen vs. Bishop / Rook Pawn is always winning. 99% of all endings and their associated theory would change from being drawn to decisive.

It is such a basic, non detrimental rule to what chess is already that I don't understand why other "possibilities", which lets be honest no one will actually play (I don't really care much for a game other than chess, and even small changes like not allowing castling or letting pawns move two spaces leads to enormous play differences), are even being explored. When we all started out, we all had that situation where we are up three queens, the King is in the board, we see a check mate in two, move one of our queens and... then we all thought, "I am up three queens, how is it fair that we tie the game?!". Realistically, this change won't happen, but I think it is the prime contender to tackle "draw-death", which I wonder if it is really all that bad of a thing.

I think it is a non-issue with a simple answer. Draws are chances for lower rated players to climb the rating ladder, and to be avoided if you need a win. They are part of chess, and really just because at the highest level games end in draws, does not mean that is the case for the majority of chess players like us. I draw less than 2% of my games online.

The vast majority of players are still not as strong as the top grandmasters in the 60-70s, chess's "golden age", but the rules were exactly the same, the basic endgame theory hasn't evolved all that much, and they did not have a draw problem. It thus entails that players up to the level of these old grandmasters do not and cannot play for draws to such an extent as the top grandmasters do today.

With that being said, it is only natural for us to feel disappointed with all the draws at the top level. It is just the nature of the game, but one can't help but feel a bit empty after watching an hour long match just for there to be no winner, and this in the majority of games.

If the solution to all of chess’s problems was to simply allow stalemate = win was it would be missing the point of the investigations by Deep MInd and Kramnik’s study. They’re interesting in lessening the value of opening theory which is what is causing the glut of draw death. Because all you’d have to do is become super booked up in whatever opening line you choose and you’re off to the races but against someone equally booked up will only end up with a massive amount of drawn games. And if you’re trying to say most of the draws out there are stalemates...sorry. That’s not the experience I’m seeing when going through those games. These games are straight up draws no matter how you count them.

Self capturing pieces and allowing pawn moves laterally and backward moves increases the complexity of the game opening up new avenues of investigation while at the same time lessening the value of opening theory. Add a dash of Fischer Random rules and opening theory can be kicked to the curb. Then you can kiss the draw glut goodbye.