Fair Chess and Simultaneous Games | Hacker News Reader

Fair Chess and Simultaneous Games

asvarga.github.io

70 pointsmlavrent43 comments

People get hung up on the unfair White advantage in chess, but actually it's not large enough.

In tennis, on every point there's a big advantage to the server. Not because he "gets to go first" but because he gets a second chance in some situations. You can prove this: in high level mens tennis the server wins ~70% of points [0] but on a second serve - equivalent to playing without the second chance - he wins almost exactly 50% [1]

This creates a tension in every game where one player is attacking and expected to win, the other player needs to "break" him at least once or twice in the course of the match to win the overall contest.

Chess is similar, but worse because of the possibility of draws. 60-90% of top level chess games end in draws.

Computer chess is even worse again! 95%+ of top computer play ends in draws. Organisers of engine tournaments have solved this: they let the computers play from positions considered advantageous to white, usually where they expect White to score ~75%. They play each position with both White and Black. [2]

This wouldn't be a popular or practical change for human play. But that's not the point, letting White take back his moves à la tennis wouldn't be a change people would accept either. The point is that chess isn't in need of evening out the first-move advantage.

[0] https://www.ultimatetennisstatistics.com/statsLeaders [1] https://www.braingametennis.com/the-art-of-winning-2nd-serve... [2] https://tcec-chess.com/articles/TCEC_Openings_FAQ.html

> This wouldn't be a popular or practical change for human play.

What about Fischer random chess? That could help, no? I heard that even Magnus Carlsen played in some tournament with it.

Right, that's a good example: it's the most mainstream of variants, gets support from FIDE and rich sponsors and world champions, and improves on perhaps the biggest complaint people have about chess - and it still only attracts a few percent of the audience and the participation of the standard game.

Fischer random chess is interesting because some starting configurations give a pretty big advantage to white (more than a half pawn advantage) whereas other starting configurations give less of an advantage to white than standard chess.

The main thing is that people have invested so much in getting good at the game as it is, that they don't want to change it, making so much of their hard work useless. That's understandable, and OK. As long as everyone understands that this is the reason things are as they are.

What's more mystifying to me is that in a weird variant like Duck Chess which has oddly gained a lot of popularity online, where so much of the hard work is invalidated anyway, they don't take the opportunity to fix the starting imbalance. In particular, in that variant it would be so easy: just let the first move be duck only!

> In particular, in that variant it would be so easy: just let the first move be duck only!

I don't see why that would make a significant theoretical difference. Black plays first and moves the duck, he can block either e4 or d4 (or something offbeat like g6 if what he wants is to get a Modern Defence at all costs) but not the other one. I suppose there's a practical difference where White couldn't specialise in e4- or d4- openings, and Black gets to choose which to face depending on the opponent.

(And in any case, as above, I disagree with the thesis that the imbalance should be "fixed").

I honestly really want to see a TCEC style rapid chess tournament. The Capablanca variant tournament is somewhat similar, and I really enjoyed that, but the idea where we pick obvious imbalances and see how many can be saved is fascinating to me.

> People get hung up on the unfair White advantage in chess, but actually it's not large enough.

It's fairly significant given the 'theoretical' assumption that perfect play always ends in a draw. However, in practice, whites wins significantly more than black.

> In tennis, on every point there's a big advantage to the server. Not because he "gets to go first" but because he gets a second chance in some situations.

Yes. It allows servers to take chances and go all out on their first serve. It's extremely beneficial to power servers. But you don't get 'take backs' in chess. So your analogy doesn't apply. You can't gamble with white and if black knows the opening, then start all over with another opening.

> The point is that chess isn't in need of evening out the first-move advantage.

As long as both players get equal chances to play with white. No player would agree to a tournament where you get black 10 times and your opponents gets white. Just like no tennis player will agree to a tournament where his opponents gets to serve all game.

As you computer chess stats show, it seems like better play leads to more draw. And the assumption that if chess is solved, then perfect chess is always a draw. But that's not how it works in the real world. White has a distinct advantage. Whether it is due to human psychology or something else altogether is up for debate.

The statement still applies even if you found one thing that isn't literally the same. The point wasn't "you can do takebacks in chess so this is exactly the same!!!" but that the advantage for moving first in chess is much smaller than is already accepted in other sports.

> The point wasn't "you can do takebacks in chess so this is exactly the same!!!"

That wasn't my point either. My point is that if chess allowed white takebacks ( as in tennis ), the white advantage would skyrocket to 70%+. The analogy was pointless because it wasn't comparing likes with likes.

> but that the advantage for moving first in chess is much smaller than is already accepted in other sports.

Yes. But in most sports, you get equal or close to equal chances to 'move first'. In tennis, you alternate serves with each game. In football, you take turns playing offense. In chess, it's not necessarily like that. Some tournaments you get to play the same opponent with black and with white in equal amounts. But in many tournaments, that's not the case.

So in tennis, you are pretty much guaranteed an equal game since both sides get to serve. But in chess, even though the first move advantage is less, it's still more significant than in tennis many tournaments don't have equal games of white and black.

Isn't this game with the given conflict resolution rules just broken? But forcing conflicts feels like a way to cause a game to become a draw. Imagine I move my queen next to the opponent's king (with no defense). My opponent's only legal move is to capture it in the next move. Let's say I don't like my chances of winning, so I just keep on trying to move the queen away from that square. Every such attempt results in a conflict, eventually leading the game to be drawn.

Yes, conflict blocking moves makes no sense in this variant, since you can block checkmate indefinitely unless the opponent angled their pieces in a way which prevents all king moves and intervention from other pieces which would be very rare.

Yeah, I’ve been thinking about this and there’s two problems you’ve illustrated: there’s the repeating bluff problem. That can be solved by simply saying you can’t submit the same move that has already been blocked this round.

But that makes the “force draw” problem even worse. I guess it’s a lot more fair to make draws easy to achieve. I don’t know if it’s fun.

The proposed resolution scenarios on conflicting moves would make the "moves stuck" to often. My proposal would be to split move into two phases: - simultaneous selection of moves (players select action at the same time) - non-simultaneous execution of moves (with circular queue of order rotating each turn) It's not a perfect solution, because in case of the conflict on the first move it gives advantage to first player in execution order, however many complex games (Chess, Go) allow to perform multiple moves before first conflicting move may happen. Some other (Tic Tac To, Othello) allow to make conflicting moves from start. Keeping an eye on the execution order would give additional strategy layer for planning moves. I like also idea on resolution of conflict depending on the set "strength" of the piece, or annihilation of both when it's equal. This would also bring interesting complexity for decision making.

My solution to fair chess:

1. White makes their opening moves - they can move more than one piece, and even the same piece more than once, but: all moves must be legal, and no captures.

2. When White is done, Black has the option to change sides, taking over the white pieces.

3. Regardless of step 2, the player with the black pieces makes the next move.

4. A draw counts as a win for the player with the black pieces.

Thus there are no longer drawn results, and the start must be relatively equal (between a white win and a black win or draw) in White’s estimation.

That's an interesting approach. But it still allows opening preparation by White. They could setup positions with deeply hidden threats, that Black would at least need to spend a lot of time on to try uncover.

So I propose that after White made all their opening moves, Black likewise makes their arbitrary legal moves, and then it's White who has the choice of swapping.

That should not only make the game fair, and optionally guarantee wins with your draw rule, but also eliminate all opening preparation.

There should probably be a limit on the number of moves that each player makes in their opening; perhaps around a dozen.

> That's an interesting approach

Thanks! I was put off by how much the article's solution changes the nature of the game, saw someone's comment about chess being too even and draws already being a problem, and it hit me in a flash how to largely solve both problems at once.

I agree that White potentially has an advantage because they can play a deceptively strong or weak opening, hoping to fool Black into a bad choice, but I don't see that as too much of a problem, but maybe I'm wrong -- I'm no expert at chess.

If the point of this exercise is to create a game that removes first move advantage, while keeping the game as unchanged as possible, this is an utter failure.

Turning a game of perfect information into a bluffing and anticipation game is not a minor change. It's a fundamental change of the essence of what the game is.

I think the point of the article was to explore and create a rough system for evaluating rules and creating a "fair" (in the terms of the article fair) system where simultaneous moves are allowed.

I don't think anywhere its stated that the author wanted to leave a game unchanged. Or as unchanged as possible.

In this perspective I think this article/exercise is not "an utter failure". Its pretty decent.

Chess already has an element of bluffing and anticipation, it just happens over the course of many moves.

I am watching the Speed Chess Championship on Twitch as we speak and the commentators along with interviews with GMs and the players themselves are nonstop talking about bluffing and anticipation.

Magnus for example talks about the importance of purposely making a sub-optimal move as a way to bluff your opponent. Your opponent likely has memorized all the optimal moves and so you make a sub-optimal move to leave your opponent guessing whether you made a genuine mistake that you can be punished for, or whether you made that sub-optimal move on purpose because you studied it extensively whereas your opponent has not and you know if your opponent doesn't play it absolutely perfectly you can trap them.

I mean seriously you can watch the championship for free on Twitch right now with GM Naroditsky and GM Rozman, and they don't really talk that much about what the theoretical best moves are, they talk about the psychology, about players going on tilt, about making aggressive moves to throw your opponent off. It's fun and fascinating.

Sorry to be pedantic but Rozman isn't a GM

Worthwhile correction, thank you. He is an International Master.

Yes, I agree with you.

Successful moves don’t always commute. Black has just moved his pawn from g7 to g5. White submits what he thinks is an en passant capture, fxg6. Black submits a knight move, Ng6. One order removes a Black pawn, the other a Black knight.

This interested me enough to convince me to click through to the article itself. Is your observation allayed by this rule?

> Rule 2: a. Moves are tried in both orders, and only moves that are legal in both orders are merged. b. If both moves are legal in both orders but a different game state is reached in each order, neither move is merged.

GP is in response to

> All of these scenarios illustrate rule 2a. Rule 2b is in fact irrelevant for chess, because successful moves always commute.

in the article.

Thanks, I overlooked that

Apparently you can notate "fxg6" as "fxg6 e.p.", so enforcing that would fix this?

"Players enter a boxing ring and fight for the square." - I LOLed :) Great stuff!

For anyone unaware, this is a real thing! https://en.wikipedia.org/wiki/Chess_boxing

Not quite… in the most common version of chessboxing it’s more a case of alternating between the chess & the boxing via, say, a 3 minute timer from the start of each “round”. You do need to be very careful moving the chess board in/out of the ring between rounds so as not to upset any of the pieces but otherwise play just continues round by round until checkmate or knockout (or TKO or a clock falls etc.)

Source: I fought in (and won!) a one-off chessboxing exhibition match in London back in 2012

In human play, white has somewhere between 0.25 - 0.5 pawn advantage. There's evidence that at higher ELO's, the advantage declines but it's not clear what the lower bound is.

So for top players (and even more-so for top chess engines), the white advantage isn't enough to translate to a win.

In human play, the top players often need to play suboptimal moves to convert a win. Magnus Carlsen is probably the most famous for doing this. The point is to break away from the well studied lines, and play something that other pros aren't familiar with.

Basically: It's not clear that white's small advantage actually counts for much, at least at the very top tiers of chess.

Changing the game in such a way that white and black odds become even more balanced, would just lead to more draws - which I personally would think makes the game less interesting.

I think this could be a very interesting way to reinvent chess which is in my opinion less interesting with AI, it would add a strategy facing the type of player but as my colleagues said the risks are conflicts like queen next to the king (and a lot of other situations).

We should have to create many rules to avoid this.

Nice post :)

One game I've played and thought of as similar to some sort of simultaneous chess was Toribash. It's a turn-based fighting game where you control your character via contracting/extending muscle pairs on a simplified skeleton. Your opponent chooses which joints to move at the same time as you, then the time advances a few frames. The game gives you a preview of what would happen if your opponent did nothing that turn. "Conflicts" are resolved via physics (e.g. a punch is stopped by an arm blocking it).

Have you seen Yomi hustle? it's similar but more of a modern fighting game.

Interesting to read the various[1,2] discussions on this chess variant

1 - http://www.hexenspiel.de/engl/synchronous-chess/ 2 - https://www.chess.com/forum/view/general/synchronous-chess

Makes more sense to slightly buff black's opening move. E.g. allow two pawns to move one square, or select a few pawns that can move three squares. The flexibility will make white harder to play than black.

This is how esports are balanced, and how a game like Starcraft was (at least when I played) more fair than Chess even though there were three "colors" involved (Zerg, Protoss, Terran) and way more "pieces" and complexity.

When we consider sets of rules that show emergent properties,very small details like the conflict resolutions discussed here can have a tremendous impact on how the general strategies would pan out at high level play. That really makes me think of how rules for "game-of-life" type of cellular automata rules can be designed. Small tweaks decide if the automata has a trivial computational power or is Turing-complete. Within the Turing-complete class there's a wide range of behaviors that can be obtained if we look at the structure that can spontaneously appear or statistical properties (e.g. what is the probability that the grid ends up entirely empty).

These observations are interesting as they give a test suite, or 'local properties', that can be run against any given simultaneous ruleset to characterize it. It would be fascinating to be able to run an AI to have an idea of how 'optimal strategies' (global properties) would look like in each case and see what relations we can draw from it.

(unfortunately I would assume it is still unreasonably costly to do something like this?)

I was expecting: The players play two games against each other concurrently. Each plays white on one board and black on the other.

I believe a tactical move blocking strategy could evolve from these rule changes. Imagine we're in a game that's more complex than what's depicted in fig 6 but the rooks are positioned identically. Additionally, there's a white pawn at c5. Wouldn't it then be advantageous for black to make the same move as shown in fig 6? If this move results in an illegal state nothing is lost since the pawn remains blocked. Perhaps this isn't the best example as my chess knowledge is somewhat limited, but do you see what I mean? Wouldn't this tactic allow players to maintain control over two positions simultaneously in certain scenarios?

The simultaneous move requests sound like Diplomacy. I'm not convinced this would improve the game in any way.

Reply on news.ycombinator.com