Talk:3139: Chess Variant

I should add support for playing this to my WIP chess library. 73.164.32.149 21:43, 8 September 2025 (UTC)

I feel nerdsniped by this one, and I'm not even into chess. Should you either slide a tile or move a piece in your turn, or should you do both, or should you move a piece on your turn and slide a tile on your opponent's turn? Also, should it disallow sliding the board back to its immediately previous state, to avoid the back-and-forth situation on the title text (but would still allow moving in circles)? Which would be more fair, and reduce the chance of draws? So many questions... 185.81.126.164 22:00, 8 September 2025 (UTC)

I have another, but related, question about the sliding, on the assumption of it being "move or slide", each turn: Do you have to fully move a tile, or may you half-move it? It would give the opponent only the chance to either complete or reverse the 'slide', if they found it tactically more advantageous to do either (would depend upon which pieces, of either side, were 'loaded' on the 2x2 as it moved; and/or perhaps which through-paths were enabled/disconnected for each slider-position; and a half-move completed by the opposing player is effectively a 'free slide'-then-move for the original player, if not accountsd for in other ways). But, if the opposing player chooses (or is forced; perhaps from 3-repeat/5-repeat consequences, or even due to potential "discovered check" exposures?) not to complete/reverse the half-slide, then does the loading/unloading any bishop upon the half-moved 2x2 (with the black/white chequerboard temporarily misaligned in the vicinity) give it an opportunity to change which shade of diagonals are its 'home'?

Yes, it'd be simpler (relatively!) to just require it to be whole-tile (2x2-square) sliding, but it'd be more interesting to consider the (otherwise valid) half-disjointed positions. Especially insofar as it works for the combined bishopry on the board! 82.132.238.63 08:29, 9 September 2025 (UTC)

If the opponent is allowed any way to make the bishop leave its designated color then I am allowed to blast them with holy water for their sins46.144.8.194 11:18, 9 September 2025 (UTC)

Hooooookay. If 1. e4 e5 2. Nf3 Nc6 3. d4 , it's Black's move in canonical chess. For the position shown in this variant, White would have to be allowed to move a piece and a sliding section: 3. d4 & ef34>gh34, Black to move. Why White would use the extra move to double down on surrendering the center of the board, sacrificing the gambit pawn for no apparent benefit, is beyond me. At best, this is taking hypermodernism to incomprehensible extreFOOOOOOOOM. Ow ... 2605:59C8:160:DB08:816E:805:44F1:B553 00:45, 9 September 2025 (UTC)

A clarification on moving in circles. The draw by repetition rule causes a draw when a position on the board is repeated 3 times at any point during the game. Thus, moving tiles in circles would cause a draw. BobcatInABox (talk) 12:49, 9 September 2025 (UTC)

Couldn't this be black beginning their turn by moving a tile? 2603:8001:0:46DC:117A:208A:40D2:9A5F 06:54, 9 September 2025 (UTC)

What is the starting position of the tiles? Can ranged pieces pass over the “gap”? Can you slide a row of tiles at once or only one? 2A02:AB88:7815:7E80:E212:8D1:BCBB:7DD8 01:03, 9 September 2025 (UTC)

Can that black pawn advance straight to the tile in front of the king or is it stuck until the gap is filled? 2603:8001:0:46DC:117A:208A:40D2:9A5F 06:54, 9 September 2025 (UTC)

If the sliding number puzzle idea was taken to its logical(!) conclusion, you would set up the board first and then randomise the tiles, which would result in some... interesting starting positions. If you did this, would it be possible for either, or both, of the players to be in checkmate at the start of the game? 82.13.184.33 08:51, 9 September 2025 (UTC)

Perhaps the most balanced starting position would be a complete board, with removing an unoccupied tile (once only) being one possible move.196.245.54.177 05:32, 9 September 2025 (UTC)

How long do you think it will be until someone actually implements this? 206.193.5.5 23:04, 8 September 2025 (UTC)

Buddy I am working on it 73.164.32.149 03:28, 9 September 2025 (UTC)

Hope you make a version that allows En Passant with a pawn that had the tile slide --2604:3D09:84:4000:9CC4:9BCE:BD01:C811 05:28, 9 September 2025 (UTC)

Do we specify positions for the transcript or do we just say "scotch opening"Mathmaster (talk)

Could you just slide yourself out of ladder checks then? 108.211.178.78 (talk) 03:06, 9 September 2025 (UTC) (please sign your comments with ~~~~)

This reminds me of the Twist and slide cube. Fabian42 (talk) 04:12, 9 September 2025 (UTC)

For anyone interested, a pretty awesome board game somewhat similar to this concept already exists called The Amazing Labyrinth.
Listing on BoardGameGeek / Listing on Ravensburger website / Wikipedia page. — Lheydon (talk) 05:08, 9 September 2025 (UTC)

This seems highly reminiscent of something that Ralph Betza of Chess Variants Dot Org would have created. I can basically outline the rule set that he would have used for games like this:

1. A move that undoes what the other player just did is not allowed.
2. A move of that does not alter anything other than the position of the hole is not allowed.
3. Riders (R, B, Q) cannot cross the gap. Leapers (N) may jump across the gap. A piece moving diagonally may cross a vertex if there are three or more actually present squares touching that vertex.
4. Pawns are simply stuck if they are not on rank 8 and there is no square in front of them.
5. Moving the hole does not move the piece, so castling rights of a Rook is not affected by moving.
6. Moving the hole counts as a move, so if you had en passant rights beforehand you lose it.
7. Notation would be to pretend the hole is a piece, perhaps using the fake piece H and taking the square closest to a8 as its nominal square, so for the move depicted you'd write, e.g. 1. Hg4.

I cannot find something exactly identical to this idea, but I certainly recall a "Subway chess" of some sort (not written by Betza) where there is a blob of squares in the middle of the board and you can move the train left and right in lieu of moving a piece. And Betza did write Earthquake Chess, which is this game but with a different thing to slide. Isoraqathedh (talk) 13:39, 9 September 2025 (UTC)

When I was considering the "riders and leapers" issue, I couldn't thing of a possible instance of "Leapers (N) may jump across the gap" where the move of two in one orthogonal and one in the other (where the mid-move 'landed' on 'space') could not be performed as first the one then the two (the mid-move being on a non-'space'). Just as it doesn't matter normally whether either, both or none mid-move squares are occupied by other pieces, and it affects neither Knight or mid-spottee piece in any way, though a 'long knight'-type fairy-piece ({±3x ±1y}, or vice-versa) might be affected by the same 'initial/final ride' being restricted. Similarly thinking of it as moving one square orthogonally then (continuing, rather than reversing) one square diagonally can be also one diagonal then an orthogonal (being just one of the components of the diagonal).

Of course, some people may think of knights as always 2-then-1 or orth-then-diag (or vice-versa, in either case) and handwave away the possible 'mid-jump' occupant this might involve. But mid-jump 'barriers' are still always ignored (even if on any/all 'tracks' that a moving knight might use), so unavoidable mid-jump voids would need explicitly to be defined as different to the ability to ignore the more mundane sitation of a piece being 'inconveniently' where any knight needs to transition via. (It wouldn't even need the edge-skimming diagonality rule, which I appreciate, but might be more simply defined that any other square which forms part or all of the movement in that direction must be free. Or, for the destination, an opposing piece which may be captured.)

But whether there are prior "fairy boards" with holes in, that specify explicit limits to a knight's leaping-movs over the gaps, I don't know.

I had thought to take a look at 3036: Chess Zoo and derive Randall's philosophy regarding 'knight-permeable' barriers, there. But there seem no obviously intentional selectively-blocking 'cage walls' (i.e. a knight could jump from A to B, with B as a mere token open square, but travelling from B to A, with rotated "L"-route, would involve a token blocking-square instead). There's at least one knight who could make a single-move transition that would be impossibly in reverse (or would be impossible, but have been possible in reverse), given a certain limitation assumption. But it can progress between the two by multiple other (openly valid) leaps and so there's no obviously intended limitation that could be assumed to be invoked.

Apart from that, I like the castling 'allowance', but does it mean that a castle and/or king that has slid-but-not-moved must then invoke it by the king moving two steps from its (possibly slid-to) initial position towards the castle's (possibly slid-to) initial position, by whatever orthogonal and/or diagonal steps are deemed necessary and valid to do so, and then the castle placed upon the intermediate king-step? Trivial for fully orthogonal (presuming that kings aren't asked to castle 'rankways', either way, when the castle involved is actually displaced directly 'filewards'), possibly trivial for fully diagonal (assuming that it's not further that it's "the major orthogonal separation" that the king's steps are constrained to, or preferentially by rank when it could be either?), but gets a bit more complicated if it's a more uneven diagonal with no 'integer' solution to each {X,Y}-step (first king-step must be the move that least deviates from the diagonal, second king-step must be to deviate least either from the original diagonal or the new one or the two identical steps taken must be the two-step (between the 2*orth and 2*diag transitions) that deviates least from the initial diagonal?).

Also wondering what if the sliding-squares bring them unusually close. An example (sticking to just the home rank, for the sake of description, except for any necessary slides that transitioned us there) would be to go from R...KB.R =by sliding=> R.KB...R, say, then =by castling=> KR.B...R; or even via a slid rearrangement, KBR....R basic setup, but then moving the bishop to allow .RK....R, or also sending away the 'near'-rook to allow a slightly different .RK..... result.

With the exception of a combination 'knight-move' by the king (in the manner of a diagonal displacement forcing an orthogonal+diagonal or diagonal+orthogonal two-step), the rules for no attack being upon any of the king's positions probably don't need considering further than merely checking the sole path over which the king must clearly shuffle. But if the square over which the king first needs to move is a choice between two intermediates, it could be a choice between one attacked/occupied/void square and another that is clear and not missing. Then the only correct choice (and the spot upon which the castle shall be transfered to) must be compatible with the diagonality-rule/-freedom for establishing the precise motions made during the shuffle.

Obviously a lot of open questions there. Perhaps unnecessarily so, once some of the earlier issues are answered and thus provide obvious precedent and disambiguation to much of the remaining indecisive uncertainty. ;) 82.132.238.27 15:48, 9 September 2025 (UTC)

Mr Betza has a bunch of helpful pointers for these as well. He's quite prolific and most importantly all his games hang together well, which is why I take to him as basically the authority in how Chess-like games should behave.

• With Castlingmost Chess we interpret a castling piece to be moved in such a way that the King moves to the midpoint of the locations of the starting two pieces – rounding away from its starting point if that's between squares – and then moving the Rook to the other side. Along with FIDE rules stating that Rooks may only castle if they are on the same rank as the King this basically handles all cases. In particular, a castling move with an adjacent K and R would simply have the two pieces trade places.
• The rules of FIDE Chess specify that Knights move to one of the up to eight closest squares that is not the same rank, file or diagonal as its starting square. This sidesteps all concerns about whether or not a Knight passes through any square or not and renders any idea of "jumping" moot. In fairy chess though it is generally understood to be orthogonal first, then diagonal, I presume due to the analogous piece in Chinese chess being hobbled at this square specifically. In any case a leaper being able to jump over spaces where there is no square is a frequent feature of Chess variants and so putting in the rule just sets it in easily. Balance-wise it gives Knights a chance at navigating tight spaces without being trapped. UX-wise it just preëmpts any awkward questions about what is blocking a Knight's path and having to relitigate it every time someone makes a slightly controversial move. Isoraqathedh (talk) 16:27, 9 September 2025 (UTC)

How would one handle the board gap and pieces jumping into it/moving through it? Which side has advantage based on gap location? I think slide and move (order to be determined). Move/slide as a combined step would be interesting for rook/bishop/queen/horse calculaitons. 163.116.254.45 14:43, 9 September 2025 (UTC) psyllix

Tile grid and algebraic notation - Proposal for the notation:

• Tile grid:

   --------------------
 s| Ws | Xs | Ys | Zs |
   --------------------
 t| Wt | Xt | Yt | Zt |
   --------------------
 u| Wu | Xu | Yu | Zu |
   --------------------
 v| Wv | Xv | Yv | Zv |
   --------------------
    W    X    Y    Z

• Algebraic notation for moving tiles:

 (from)>(to)
 
 Ex: Yu>Zu

(also, note: logged into my account of User/AverseABFun, yes I forgot to mention this in the thing so if there's a way for me to prove it's me lemme know) 73.164.32.149 20:13, 9 September 2025 (UTC)

My only suggestion, is that you only need the "(from)". Or call it "(from)›", perhaps. Piece-moving notation often omits details that can be infered, so the sole/only queen that can land on d4 just needs "Qd4". But it could need the crucial starting file and/or rank, such as "Q6d4", if there are multiple queens that could get to d4.

Every proper board-slide must have the "(to)" that was the last slide's "(from)", or the initial choice of starting-hole, so you just need to know which of the 2-4 neighbours, of the gap, you're moving into it (and that will become the new gap).

And, with the suggestion (up there) that the first gap is initially created by only removing from a complete board any unoccupied 2x2 candidate, you could notate that instead as "(from)»" to describe the removal. Unless there's only one "uncamped" sub-board available, when "»" might be considered sufficient record, with the other 15 locations being irremovable at that point. 82.132.238.27 03:22, 10 September 2025 (UTC)

That sounds good, I agree that probably just "(from)>" is good for the tile, and "(from)»" for the initial removed tile. Maybe the equivalent FEN notation could be "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1 Zu" (which is my suggestion for the starting position and seems to be what Randall has as the starting position). AverseABFun (talk) 18:14, 10 September 2025 (UTC)

So how would this move be rated? Is it a blunder, is it brilliant, or is it something inbetween? Can we even really know right now, without any history of this specific variant being played? Maplestrip (talk) 06:41, 10 September 2025 (UTC)

I'm looking into it and it would be possible to set up Elo ratings for this style of chess as Elo ratings can be created for any zero-sum game. For figuring out blunders and others, that's subjective and when calculated is calculated through engines (commonly stockfish) so until someone adds support for slide chess (that's what I'm calling it as well as calling it X3139 chess) to stockfish or another engine we won't know. -- AverseABFun (talk) 19:10, 10 September 2025 (please sign your comments with ~~~~)

Okay, I have finished my initial version. Note that it does not have any sort of move validation. To move a tile, select any square in a tile in which you have at least one piece and then select any part of the empty tile. https://github.com/Aversefun/x3139-chess https://x3139.trustworthysources.xyz. -- AverseABFun (talk) 19:01, 10 September 2025 (please sign your comments with ~~~~)

It's your implementation, your rules (and I haven't yet tried your version out for myself, to check), but I didn't get the impression from the comic that there was any limitation to which slidy piece you could slide based upon whether it held one of your pieces or not (which seems to be what you're considering).

But maybe I'm just reading it too openly. My idea of the tactics involved being to keep the 'hole' away from the vicinity of any defensive formation you may have, lest your opponent gets to move your pieces out of formation and then procedes to further reshuffle your end of the board with the "portable hole" in ways are also actively disruptive to your placement. Conversely: make them think they're doing that, but actually letting them bring your pieces out, en mass, in a more valuable offensive arrangement. ;) 82.132.245.43 22:33, 10 September 2025 (UTC)

True. Could be fun to add a mode switch so you can pick between different rulesets. AverseABFun (talk) 01:19, 11 September 2025 (UTC)

Alright, added a mode select. Should be up soon. It has four different modes. Now I'm going to see about implementing rules for allowed moves. See you in ten years I guess. AverseABFun (talk) 01:51, 11 September 2025 (UTC)

Okay it has many a bug and I need to figure out how to block off paths when it hits something because it is too lenient but basic rules for what's allowed are in place AverseABFun (talk) 02:31, 11 September 2025 (UTC)