Slight change to rules of standard chess

[Mike Sherwin]

A simple rule change would breathe new life into chess. Allow the king to castle out of check. It would allow much more aggression to the game.

[Cornfed]

You have stats to back this up?
I think you may be wrong. I mean, to me it would largely serve to help nullify early aggression for the attacker which has perhaps taken some risks to put the King in check.

[Madeleine Birchfield]

Ask Deepmind to train a net for AlphaZero on your rule change and publish another paper to see if your claims are true or not.

[mvanthoor]

A simple rule change would breathe new life into chess. Allow the king to castle out of check. It would allow much more aggression to the game.

Why do people still want to change chess?

Things such as opening preparation down to move 35 is only going to happen on the very highest levels of the game. You only need to play a game and then walk through it with any 2500+ Elo engine to learn how much you're still not able to see. Even the fact that an engine such as Stockfish can -still- lose games against neural network engines shows us that there are lots of things in chess we still can't fathom ourselves.

Same goes for Go. It has so many 'proverbs' (rules of thumb) fuseki (openings) and joseki (standard move sequences) that many people feel 'they're playing the same game over and over again'. Look at the latest Go engines to see that they are turning Go theory upside down; variations that were 'analyzed dead' and seen as 'unplayable' are suddenly played by an engine, and even grand masters are unable to refute them despite the fact that the engine is doing something 'unplayable.'

If chess isn't dead yet for a 3500 Elo engine that has 20 move opening books and 6-piece table bases, it's certainly not dead for humans.

If someone wants something else beside standard chess, play Fisher Random, or one of the 5000 available variants. There are variants with so many pieces that move in so many different ways that I even have trouble remembering all of the possible rules...

[Guenther]

A simple rule change would breathe new life into chess. Allow the king to castle out of check. It would allow much more aggression to the game.

I am surprised about the other answers, as they don't even address the biggest flaw, probably no chessplayers in this thread.

The possibility to castle out of check (if allowed) in real games would be so tiny anyway, that it defeats the purpose of a new variant.
I guess not more than once in several hundreds of games. You would also need to allow castling over covered squares and other pieces
to make sense.

[mwyoung]

A simple rule change would breathe new life into chess. Allow the king to castle out of check. It would allow much more aggression to the game.

Why do people still want to change chess?

Things such as opening preparation down to move 35 is only going to happen on the very highest levels of the game. You only need to play a game and then walk through it with any 2500+ Elo engine to learn how much you're still not able to see. Even the fact that an engine such as Stockfish can -still- lose games against neural network engines shows us that there are lots of things in chess we still can't fathom ourselves.

Same goes for Go. It has so many 'proverbs' (rules of thumb) fuseki (openings) and joseki (standard move sequences) that many people feel 'they're playing the same game over and over again'. Look at the latest Go engines to see that they are turning Go theory upside down; variations that were 'analyzed dead' and seen as 'unplayable' are suddenly played by an engine, and even grand masters are unable to refute them despite the fact that the engine is doing something 'unplayable.'



If someone wants something else beside standard chess, play Fisher Random, or one of the 5000 available variants. There are variants with so many pieces that move in so many different ways that I even have trouble remembering all of the possible rules...

Shocking

You mean the stronger player still wins in go and chess. Exactly!

"If chess isn't dead yet for a 3500 Elo engine that has 20 move opening books and 6-piece table bases, it's certainly not dead for humans."

[mmt]

Things such as opening preparation down to move 35 is only going to happen on the very highest levels of the game.

The impact of play at the top level is huge. That's what people watch and that's where there is at least some money involved. Look at other competitions - if Nadal vs Djokovic had just a few important rallies in their matches and the rest was just about not making a huge careless error and it was like this for years, everyone would be asking for changes.

If someone wants something else beside standard chess, play Fisher Random, or one of the 5000 available variants. There are variants with so many pieces that move in so many different ways that I even have trouble remembering all of the possible rules...

Yeah, Fisher Random is not a bad option.

[lkaufman]

A simple rule change would breathe new life into chess. Allow the king to castle out of check. It would allow much more aggression to the game.

It would have just one significant consequence. The Italian (1.e4 e5 2.Nf3 Nc6 3.Bc4) would have to be met by the two knight's defense, 3...Nf6, since 3...Bc5? would be dubious due to 4.c3 Nf6 5.d4 exd4 6.cxd4 Bb4+ 7.0-0! with a big advantage. But the two knight's is sound, so this wouldn't really make White's advantage much greater. Any rule change should have significant benefits to be worth fighting for. The most significant simple change would be simply to forbid any repetition of position; some tests with Komodo show that this would cut draws to well under half the current percentage in bullet play.

[Madeleine Birchfield]

It would have just one significant consequence. The Italian (1.e4 e5 2.Nf3 Nc6 3.Bc4) would have to be met by the two knight's defense, 3...Nf6, since 3...Bc5? would be dubious due to 4.c3 Nf6 5.d4 exd4 6.cxd4 Bb4+ 7.0-0! with a big advantage. But the two knight's is sound, so this wouldn't really make White's advantage much greater. Any rule change should have significant benefits to be worth fighting for. The most significant simple change would be simply to forbid any repetition of position; some tests with Komodo show that this would cut draws to well under half the current percentage in bullet play.

And if the two knignt's defense wasn't sound, then black would ditch 1...e5 for 1...c5 and still end up about even after playing the Najdorf.

Mike's rule change doesn't even allow castling through check, only castling while in check.

[mvanthoor]

The impact of play at the top level is huge. That's what people watch and that's where there is at least some money involved. Look at other competitions - if Nadal vs Djokovic had just a few important rallies in their matches and the rest was just about not making a huge careless error and it was like this for years, everyone would be asking for changes.

That comes with being an almost perfect master at something. Two people become so good at it, that a single mistake in the game can determine the winner. With chess, trying to emulate grand masters and memorizing openings down to move 35 is stupid. If you do that, the game will become work instead of a game, and 0F COURSE it's no fun anymore at that point.

Study opening principles.
Study mid-game principles, tactics and strategies.
Study endgame principles.
Don't try to memorize openings by the hundreds.

If you do that, chess can be fun for a life-time, at least for an amateur player.

The computer is a problem with regard to this. It can -help- to analyze openings down to move 20, and then they become an opening book... and then move 25, 30, and 35... and end-game databases are getting bigger and bigger. The computer is going to kill chess, just like it has killed Checkers (Checkers is basically solved.) At some point, it'll kill Go. (The reason why this took longer is because Go's search space is bigger, and chess has had much more research through the years.)

Fischer Random Chess SEEMS an option right now, but it's not played a lot in tournaments on the highest levels; if it is, it's more like a curiosity. If it would become the standard, then the best players would make a list of all 960 starting positions, and the world would start to build opening books. At some point (in 10, or 20 years) most strong players would remember a few openings for each of the 960 positions (some of the positions will probably have openings that are very similar; so there will probably be new opening principles).

The point is that games such as Checkers, Chess and Go are finite, and only a small part of the moves are actually 'good'. The computer is able to assist in finding those good moves, and if too many people know too many of the good moves (or have a way of finding them with a 95% chance of success) the game becomes exhausted.

That a game is exhausted for a 64-core computer running a 35-move opening book and a 7 piece table base, doesn't mean the same game is exhausted for the average human/chess player. It might be -almost- exhausted for the grand masters, but I don't feel the need to tell them they have to change. THEY have to play the game at that level, so THEY will have to decide about that.

Even Arimaa was cracked, even though it was -designed- to have a search space as huge as possible. You can make 1-4 moves when it is your turn, split over different pieces or not, and it has several features such as pieces that MUST stay defended or they'll 'die', and traps in the board, and probably other things I don't remember off the top of my head. Computers are now stronger at Arimaa than humans... and THUS they can be used to analyze it. In the first few years already principles and openings emerged on the internet, such as the "camel hostage" technique.

In the Netherlands, I've seen many people play the game Santorini. It's chess-like in setup: "You can move your pieces such and such, under these conditions, and if you achieve that result, you win." I'm -SURE- a simple Alpha-Beta searcher with a transposition table and a few covers for gotcha's in the evaluation function would destroy any human player; at least in the version where "powerup cards" are not used. In that version, each player has a number of cards (and can get more) that give him special powers, and your opponent doesn't know about them. The game becomes half chess-like, half power-like at that point, so it's harder to solve.

Any game with a finite board, set of pieces, and conditions will eventually end up like checkers: solved for computers, and almost solved for the very best humans.

I keep chess interesting for myself by doing what I posted before: studying the general principles for the opening, mid-game and end-game, and I only play against computer opponents around my own level who I give an 8 move opening book and no end-game tablebases. I don't play humans in a chess club anymore, because it's always the same (mostly older) people "who have been playing chess for 60 years" (and telling it to anybody who wants to hear it), trying to win games with "their" pet opening. I see no reason to play the same people over and over again and get stuck in the same types of games year after year.

edit: whatever anyone else does, I don't really mind; but if you change the rules, don't call the game "Chess". Many older chess engines are now already playing outside the rules, because (if I remember correctly) the 50 move rule is now the 75 move rule. If rules change and GUI's go along with those changes without supporting the older rules, it'll break all previous chess engines in those GUI's.