alt.hn

3/30/2025 at 12:29:49 AM

Chess324 – A Chess Variant to Reduce Draws

 https://www.chessprogramming.org/Chess324

by Mr_Minderbinder

4/1/2025 at 6:28:19 PM

Chess and Go are very different in that one is for points and the other is for annihilation. However, Go solved the draw (and first player advantage) with 'komi'. Giving white (the second player) some extra points, I think it's usually around 6.5 points right now. The amount that komi should be is still up for debate and changing, though I think everyone agrees the half portion is good.

Though stronger players can no longer give 'presents', where you force a draw on a weaker opponent by ensuring both players end up with the same amount of points.

Is there amendment to chess that could work similarly? Nothing is coming to mind, but Chess is not my domain.

by jesterswilde

4/1/2025 at 6:54:09 PM

Actually komi in go solves the problem that black won nearly all the even games between strong players, except when white was an exceptionally strong player.

Hence the original 4.5 points komi about 100 years ago. The half point is the part that turns a draw into a win for white, that starts playing after black. But it's not strictly necessary: we could state in the rules that white wins draws.

Black started playing less conservatively and komi was soon adjusted to 5.5 points. It became 6.5 points at the turn of the century. It's 7.5 with Chinese and area rules.

This variant of chess seems to me potentially unfair to the player that gets a bad starting position, one that loses most games or that could even be winning but only with a narrow and difficult path to find. It could work in a tournament if all players have to play all the initial configurations. In thnk

by pmontra

4/1/2025 at 7:40:48 PM

Just for fun: it is worth noting that there were/are other solutions to the "black wins" problem, usually in the form of playing multiple games and alternating colors, and stuff like kadoban formats where if you win multiple times you start getting pushed to get progressively larger handicaps and see if you can win then.

by kadoban

4/1/2025 at 7:27:33 PM

Chess 960 was done to address memorized openings and,to a certain degree,the issue of draws.

But Chess960's castling mechanism is totally unintuitive. So, Chess 744 was born. The exact mechanic (rook moves to king and king swings behind it) from standard chess is used and all of chess 960's setups that had a rook and the king in a corner were removed. This website goes into detail. There is no better method, IMHO, of creating an intuitive chess variation that addresses this issue. It is very hard and unintuitive in Chess 960 to remember where to move your pieces to castle and whether or not it is even legal given the squares the king has to move through.

https://sites.google.com/view/chess-744/todays-744-game

by IvanChess744v3

4/2/2025 at 10:36:19 AM

Could you add a diagram showing how castling works in this?

by arunix

4/2/2025 at 11:14:41 AM

I can later (short on time this morning), and I plan to add diagrams to the website and a video discussing it.

Basically, the rook, assuming the spaces are open and he and the rook havent moved, either moves adjacent to the king (if he isnt already adjacent) and the king moves two spaces around him.

by IvanChess744v3

4/1/2025 at 7:37:48 PM

Go doesn't really have the draw problem. Scoring is sharper in go, so even with integer komi you don't see enough draws to really be a problem. Only reason you _really_ need non-integer komi is if you're running a tournament or rating system that can't accomodate them at all.

Even on 9x9, like look at GoQuest for example, they must have thousands of game a day at 7 komi and ~nobody complains if there's a jigo, it just means the game was really close.

In chess the problem is just high level play has _so_ very many draws, so it can be worth trying to reduce that.

In go you get a draw once in a while. In chess you get a draw by default and have to really work to avoid it (assuming you're at a level vastly vastly higher than I am).

by kadoban

4/1/2025 at 8:42:05 PM

I think the only solution is to just abolish draws like the Armageddon chess mode does, where a draw is counted as a win for black.

And then have either an imbalance in allowed time for each player or have them bid on it.

by golli

4/1/2025 at 6:31:30 PM

So, this website introduces Chess 744 and addresses the confusing castling issues of Chess 960. Basically, we removed the Chess 960 setups with rook and king in a corner and then used the exact mechanic from standard chess. Rool moves to king (if not already adjacent) and then king swings behind the rook. Here is the site - along with a designated setup for each calendar day.

https://sites.google.com/view/chess-744/todays-744-game

by IvanChess744

4/1/2025 at 7:25:58 PM

> the exact mechanic from standard chess

The mechanic for standard chess is:

(If Rook and King have not moved (and no funny business with pawns promoted to Rook), and none of the three spaces the King is involved with are attacked, and all the spaces between the King and the Rook are unoccupied) King moves two spaces, Rook moves to the space the king passed over. The order is important, at least in tournament play.

by toast0

4/2/2025 at 11:30:37 AM

Thanks for pointing that out. I will definitely have to reword a few things on our site. Maybe the same 'essential' mechanic, or something to that effect.

The king could still move the same two squares in the direction of the rook, but it might mean that the space is sometimes occupied by the rook or you may be simply jumping over the rook.

by IvanChess744v3

4/1/2025 at 8:55:08 PM

Small nitpick: you could replace “no funny business with pawns promoted to Rook” with “and are on their home squares” as there’s no way for a promoted pawn to get to a rook’s home square without moving ;)

by scott_w

4/1/2025 at 9:15:30 PM

Well... if you want to get into really funny business, white could promote to a black Rook on its home square (not valid with current FIDE rules, but older rule sets weren't always specific about the color of the piece promoted to). I don't think you should castle with that Rook, but maybe?

I guess something about being on the squares since the beginning of the game.

by toast0

4/2/2025 at 1:06:36 PM

Yeah but now you’re following a path the rules don’t allow so it’s not really necessary :-p

by scott_w

4/1/2025 at 4:33:56 PM

The idea of an asymmetric Chess starting position is very interesting, although it does introduce more risk of one side starting with a big advantage (perhaps this has been analyzed).

I also like that in this variant, castling works like normal -- that is one of the most unintuitive aspects of Chess960.

by ActivePattern

4/1/2025 at 6:41:18 PM

We fixed the castling issue with Chess 744. Thoughts?

https://sites.google.com/view/chess-744/todays-744-game

by IvanChess744v3

4/1/2025 at 7:30:33 PM

Yes, that appears to be another good solution to the castling trickiness! And probably how you assume castling works in Chess960 if you weren't given the rules.

by ActivePattern

4/2/2025 at 1:18:31 PM

But that's the problem with Chess 960 - that's not how you castle at all in that variant.

by IvanChess744v3

4/1/2025 at 8:45:29 PM

One option is bidding with points: player A looks at the position and bids X<0.5 points for the privilege of picking the color. Player B either accepts or raises the bid, the process repeats until one of them accepts or bids 0.5 points. The match is then played for the remaining 1-X points.

Or, more simply, player A is shown N random positions, picks one of them and lets player B pick the color.

by orthoxerox

4/2/2025 at 3:50:08 PM

I don't think that bidding system really works. If one side is strongly favored in the opening, the optimal bid would be essentially 0.4999999999... so that you can pick the color and win the game by a slim margin. Players then increase the bid with tiny steps ad infinitum.

The other idea works but is essentially just discarding all of the lopsided starting positions, in which case they might as well not be in the game.

by ActivePattern

4/1/2025 at 9:18:22 PM

Or you just run things like duplicate bridge. Everybody plays the same set of randomized boards.

by bsder

4/1/2025 at 5:13:42 PM

> one side starting with a big advantage

I have never played this variant of chess, but on the surface it seems that having both bishops on the same color would be a sizeable disadvantage.

The other randomized pieces (queens and knights) can get to any square, so having two knights start on dark squares, for instance, doesn't seem to really matter.

by philsnow

4/1/2025 at 5:19:40 PM

The bishops are required to be on different colors.

by anamexis

4/1/2025 at 4:53:28 PM

I don’t personally think reducing draws would improve the game at all

by seanhunter

4/1/2025 at 5:29:24 PM

This is primarily for computer chess, where the draw rate is even higher than in super-GM level chess. It's an alternative to having engines play from a preset position where one side starts with a significant advantage, which is what usually happens in computer chess tournaments. It's essentially the subset of Chess960 positions that don't need modifications to the castling rule, and thus should work on all engines without requiring any changes to the code.

by RUnconcerned

4/1/2025 at 6:22:13 PM

It's not quite a subset of Chess960 positions, because Chess324 allows different piece layouts between white and black and Chess960 does not.

Edit: A comment further down points out that only 18 of the 324 positions are symmetrical, so the vast majority of Chess324 positions could not happen in 960.

by tomku

4/1/2025 at 6:23:54 PM

An alternative is just to stop bending over backwards solely for the sake of supporting castling, which is an inelegant hack in the first place.

by kibwen

4/1/2025 at 5:53:12 PM

Isn't the argument something about how grandmasters are incentivized to play it safe to a draw much of the time, which results in less interesting games?

by bitshiftfaced

4/2/2025 at 4:46:49 AM

That is the argument that people make, yes, but if you watch top level chess you’d see that’s absolutely not the case. In the current age of computer prep a lot of games involve extremely sharp lines where one side will make some kind of positional or material sacrifice quite early to secure an imbalanced position based on computer preparation and then the whole game revolves around whether or not they can convert the subsequent advantage in initiative. Removing draws from the game would mean there is absolutely no incentive to ever sacrifice material, making adventurous attacking play a strictly losing strategy at the highest level, to the obvious detriment of the game in general.

Secondly, at other levels one of the most important things is always whether a player can show enough sustained technique to convert some advantage and there’s always a chance for the person who is behind to swindle some draw if the guy who’s ahead can’t do this. This sort of a change would make that conversion technique somewhat irrelevant. The person ahead could just turtle up and guarantee themselves a win.

I know for a lot of people having draws in a game is very unsatisfying to some spectators[1], but the existence of draws in a game changes the strategy a lot meaning a player who is ahead has to press their advantage in order to secure the win and not allow a late heroic rearguard defence to secure a draw. That is (in my view) generally good for the game.

[1] I am a big fan of cricket and this is a common criticism of the game from folks who don’t like it.

by seanhunter

4/1/2025 at 5:29:28 PM

I don't think this variant is for playing, but for practicing tactical games against a computer.

by legitster

4/1/2025 at 5:30:24 PM

Certainly not if it is done by randomly giving one player a substantial advantage.

by gweinberg

4/1/2025 at 5:16:39 PM

Does this "reduce draws" purely because it randomly gives one side a small (or big) advantage? If so, that doesn't seem to be an improvement at all.

It would be easy enough to test if one side having an advantage is common: have a chess engine play against itself many times using the same setup. The try a new setup. Find out what percentage of the time one side seems to have been given an advantage. (i.e. an advantage that's greater than the built-in one that white has.)

If there's a different reason this reduces draws, what is that reason? If it's simply reducing the reliance on book openings, that's fine, although there are other solutions for that. If there's something in a particular piece structure that reduces draws, what is it?

by SamBam

4/1/2025 at 5:26:12 PM

This was developed mostly for computer chess, where games are often played starting from preset positions, where one side already has a significant advantage. The top chess engines, when playing each other from the regular starting position, which is very equal, will almost always draw (especially in time controls where they have an increment of one second or more), so we humans have to force them to play lines that they would never go into so that the games aren't all draws (and while some games can be exciting and still drawn, that is not the case with most of the lines the engines will pick when playing from the start position).

This is essentially just a subset of the Chess960 positions that don't require any modification to the castling rule.

by RUnconcerned

4/1/2025 at 4:41:03 PM

With an asymmetrical setup there has to be cases where one side has a real advantage over the other.

How big is hard to know. Maybe it doesn't matter much.

Chess engines can figure that out empirically.

PS I checked, and this has several mentions before April 1!

by BurningFrog

4/1/2025 at 4:44:59 PM

It's by Larry Kaufman, so it's safe to assume it's for computer chess rather than humans. If that's the case, it's not a problem; engine matches are usually played as "game pairs" where each engine plays the same position as both White and Black.

Having unbalanced positions could even be better, as you'd be more likely to have a decisive game pair.

by qsort

4/1/2025 at 5:17:43 PM

I agree with the idea that rooks should be in their regular positions along with the king; This makes the castling rules much more intuitive than chess960. I think the asymmetric starting position is going to be pretty rough as this would lead to huge advantages. E.g. having 2 bishops next to each other is so much stronger than not.

by y-curious

4/1/2025 at 4:56:16 PM

How does this work? Do players secretly choose their starting positions, or are the starting positions randomly determined?

by slig

4/1/2025 at 8:58:43 PM

I think the idea is that you'd play with a chess engine that randomizes the positions for each player.

by CollinEMac

4/2/2025 at 5:22:36 AM

I suppose, if you do not have a computer (or do not want to use it) then you might use dice to make the decision. (You can use d20 good enough, that there are 18 possibilities, and if it is 19 or 20 then roll again. Opponent then does the same with their pieces, too.)

by zzo38computer

4/1/2025 at 4:33:29 PM

Fairly small change, just some changes to the initial piece placement.

“All Kings, Rooks, Pawns are in their original locations as chess but other pieces are placed randomly in their first and last ranks, with no symmetry requirement between two sides, with the only restriction for bishops of each side must be on different colored squares.”

by azhenley

4/1/2025 at 7:38:08 PM

so the knights and bishops can swap places?

by Suppafly

4/1/2025 at 4:41:21 PM

Has anyone thought/experimented of a variant that works a bit like the starting phase of Risk: pawns are as they are classically, but players take turns in the beginning phase placing their other pieces in their respective bottom/upper file.

by ak_111

4/1/2025 at 6:22:37 PM

I've heard this referred to as either Placement Chess or Bronstein's Chess (after the GM who reportedly suggested it). It's available to play on https://pychess.org, where it's called Placement Chess.

by autocorr

4/1/2025 at 8:30:16 PM

Yes: https://www.chessvariants.com/alphabet.html

My point not being that that is a link straight to the exact thing you've described, but that if you ask that site for all the variants it has with a blank search [1], the first 500 variants on file (give or take some administrative garbage) gets you the variants starting with a number and up to "Avalanche Chess" alphabetically... that is, you can't even get all the way through the As before the search bails out.

[1]: https://www.chessvariants.com/index/mainquery.php

by jerf

4/1/2025 at 5:32:26 PM

I've heard of that variant but I think it is very seldom played. I don't know if it has a name.

by gweinberg

4/1/2025 at 5:42:14 PM

Chess960[0] aka Fischer Random aka "Freestyle Chess" (the last is a name chess.com has been using") is a fairly popular variant with 960 possible starting positions.

The way I'm interpreting this, 324 is different in that the rooks & kings are fixed, but the position is asymmetric.

Is it called 324 because there are 324 possible starting positions? Also, I'm inferring that this is popular in computer chess specifically, is there a reason for that?

[0]: https://en.wikipedia.org/wiki/Chess960

edit: there's a quote from the inventor in the wiki that answers my first question and sort of answers the second:

> This version has huge advantages over chess960. First, no special castling rules, any engine or GUI or human can play with no instruction after seeing the initial position. Second, since all but 18 of the 324 positions are asymmetrical, opening play should be much more interesting and complex. Third, the normal positioning of the rooks and kings and normal castling makes the game feel closer to normal chess. Fourth, matches of up to 648 games can be played with no repeat positions, generally enough for most purposes. Most important, no matter how many cores or how much time the engines get, there should be plenty of decisive games for the foreseeable future since many positions are at least not too far from the win/draw line.

by cowpig

4/1/2025 at 6:36:15 PM

> Is it called 324 because there are 324 possible starting positions?

Yes, each side has 18 possible setups. The bishops have 3 x 2 possible opposite-color placements among the 5 squares, and then the queen and 2 knights have 3 ways to occupy the remaining 3 squares. 18 x 18 = 324.

by tromp