Let's invent some very weird pieces


Let’s invent some very weird pieces that are nonetheless still playable and understandable in the context of chess-like variants.

I was thinking about:

Two pieces that move like rooks, but when and as long as those two pieces ‘see’ each other as rook they also have the knight’s move.

A piece that moves like a king, but if / as long as it is blocked by friendly pieces in all directions, only then it gets jumping abilities.

Two pieces that move like knights, but as soon as only one of them is captured the other will also get the queen’s move on top of that.


I see really new tactical patterns coming with them.


You offer a chess variant Chameleon: at general each piece has the move dependent from its own position. This will never do for our conservative GMs!


The weirdest piece I invented for one of my own Chess variants is the 'Werewolf' of Werewolf Chess. It slides like a Queen, but maximally upto 3 squares, but it can jump to the second square (but not the third) directly, even if the first square in that direction is occupied. And when it jumpsover an enemy piece that way it can (but doesn't have to) remove it in the process (like a Checker capture), even when it also captures something on its target square (so it is capable of double capture). But that is not yet the strangest property: every piece other than a King that captures it, will instantly promote to a Werewolf. (Which effectively prevents it is traded out of the game early.) That idea of a 'contageous' piece was taken from Maka Dai Dai Shogi, btw.

Another pretty weird piece, which I did not invent, but stole from Chu Shogi to feature it in my own variant Mighty-Lion Chess: it moves as a King, but twice per turn, and doesn't have to stop after a capture. (So it can do hit-and-run, and double captures, and rifle captures on adjacent squares.) And when it captures another Lion it becomes absolutely royal for one turn. (Implying you cannot capture an enemy Lion that is protected with it, unless it is through hit and run.)

Knight-relay Chess has pieces that acquire extra moves when 'seen'  by a friendly Knight; Chameleons acuire the moves of a piece that sees them, and can use them to capture that piece.

'Bifurcators' are sliders whose moves are 'deflected' by the presence of other pieces. E.g. one can start moving as a Rook, but when running into the first friendly piece change direction just before it, to continue sliding like a Bishop. (An enemy piece would siply be captured.) I used Bifurcating Rooks and Bishops in my variant 'Bifurcator Chess', but it was more to demonstrate a possibility, and I was not very happy with the result. (Too complex.)

The idea of having one piece promote if another one is captured is interesting. It would be another method to prevent a piece is easily traded out of a variant that would degenerate to orthodox Chess after the piece gets traded. You could for instance stipulate that the most left Pawn (the most backward one if it is doubled) turns into the piece when the original piece gets captured.

For testing purposes I sometimes use a piece I called (Lion) Cub: it moves and captures like a Knight, butin addition it can rifle-capture like a King. So when you end up next to it, you are cooked.

That reminds me of the Fire Demon of Tenjiku Shogi. Not only does it have a strange move (like Bishop or sideway Rook, but also up to 3 King steps in independently chosen directions, stopping at the first capture), but it 'burns' every enemy on squares next to the one it lands on. (Meaning it can capture up to 8 pieces per turn.) And this burning power is also passive: every enemy that ends next to a Fire Demon is incinerated, and instantly disappears. (They can capture it by landing on it, though.) Even an enemy Fire Demon is fried this way, before it can do its own roasting.

The 'jumping generals' of Tenjiku Shogi are also pretty weird: they move as normal sliders, but can in addition jump over an arbitrary number of pieces (friend or foe) to capture something on the square they land on. They cannot jump over each other, though, or capture a King that way. (In fact they are 'ranked' 1-3, and can jump over generals of lower rank or normal pieces.)


I would prefer not to emigrate to Heian Shrine. When in Rome do as the Romans do. Fortunately this fruitful idea had been already successfully realized at cards (poker, piquet and so on). Therefore my Waterloo can really help you.


Great thread!  I always think the king should be able to slay one of his own pieces - for example to avoid smothered mate as his survival is the absolute priority but perhaps people thought this would reduce the motivation of his troops!

The optionality of the knights was a great idea.  Would massively increase the complexity of the positions.


as far as i remember, Capablanca proposed a sensible extra piece. Perhaps you could look it up ??    it may be ok for online chess


@HGMuller, what is the general meaning of 'slider' in chess variants?


Strictly speaking a 'rider' is a piece that can repeat the same board step arbitrarily many times until it hits something, and a 'slider' is a 'rider' where the basic step goes to an adjacent square. But an additional specification (e.g. 'bent slider') can provide a context for a more liberal interpretation, which drops the requirement that every board step is the same (direction and length-wise). Then it becomes a piece that traverses a prescribed path sequentially, having to stop at the first obstacle. 'Bent' means it can switch step vector once (e.g. the Gryphon), 'crooked' that it switches between two step vectors on every step, so that it zig-zags ('Crooked Bishop'). 'Circular' means the step vector stays of the same type, but rotates in the same direction on every step. (E.g. the Rose is a circular rider.)

Note that this set of definitions basically applies to move types, and not to pieces: pieces can be compounds, where only some of its moves fit the definition. E.g. the Archbishop moves a Bishop or Knight, and only the Bishop moves qualify as sliding. So strictly speaking it is a slider-leaper compound. I would be inclined to call a piece a slider if it has some sliding moves, even if it also has other moves.


The 'Emperor'of Thai Shogi (also occurring in Maka Dai Dai Shogi as promoted version of the King) is also an interesting piece. It can move anywere in a single move. It can even capture anywhere. Except that you cannot capture the opponent's royal pieces when they are protected, even when it would be their last royal, so that capturing it would normally win you the game.

This amounts to a rather unusual priority of the winning conditions. In the first place (1) you win when you recapture an Emperor that just captured a royal. Only when you cannot do that, you lose (2) when you have no royal yourself. So capture of a royal other than that decsribed in (1) does not automatically end the game, as it would in most Chess variants. You would always have to play one more move before you can apply (2), to see if (1) was possible. (This is similar to the bare-King-loses rule of Shatranj, where you must play one more move to see if the capture that bared the King was legal, or can counter-bare. So there King baring is not an immediate loss, but can only be applied if your King is bare at the end of its own turn.)

One would expect the presence of Emperors to drastically affect the game, as you can suddenly gobble up anything the opponent leaves unprotected. But in practice, when both sides have an Emperor, there never isn't anything unprotected that could not have been captured in an 'ordinary' way (i.e. by another piece than your Emperor), as the opponent's Emperor always protect it. So during most of the game the Emperors neutralize each other, and you get quite normal tactics. You just must make sure your own Emperor stays protected by one of your own pieces, because when it is not, the opponent of course grabs it immediately. (E.g. if in Chess the Kings were Emperors, 1.e4 d5 2. exd5 Qxd5 would be a bad idea because of 3. Ee1xe8#)

I am not sure this kind of royalty is a good idea; it makes it virtually impossible to checkmate in the middle-game, as there almost always always will be a safe, protected square to which an attacked Emperor can teleport to get out of trouble. You can only win by either attacking every square on the board with non-royal pieces, or by reducing the opponent to so few pieces that he cannot keep his Emperor protected from yours. So every game must last to the late end-game. (Of course in Maka Dai Dai Shogi there are Hook Movers, which are Rooks that can make one 90-degree turn, and thus do cover the entire board just by themselves, which makes catching an Emperor a bit easier.)


There also are the 'Catapult pieces' invented by Mats Winther. These do not capture themselves, but they can 'sling' other pieces to a different location. E.g. a piece could move as a King, and then displace every piece standing next to it in the opposite direction as how it moved as far as possible (slider-like) in the direction which it moved. E.g. such a Catapult on f3 could sling a Knight that was standing on e2 to h5, by moving to g4.


Well, I suppose everyone invents pieces according to their own mental abilities...


The ring:

Any piece can carry it

You win by destroying it in the opponent's territory (final rank)

Capture is obligatory for a piece carrying the ring (which gives the opponent some opportunity to gain it :) )


This is a cute idea, but I think the obligation to capture is too much of a handicap. It also doesn't sound very natural in the context of the story. Perhaps it would be better to limit the piece that carries the Ring to a single forward or sideway step. If it moves a larger distance or backwards it would have to leave the Ring behind.


Nope, the idea is that a strong piece is a bad ring bearer because it will be trapped more easily by the enemy - simply because of the obligation to capture combined with more possibilities to capture.


On a very large board distance becomes a serious factor for knights. I would like to allow four of them to move like a queen as long as the squares in between are not occupied and if the four knights are in an exact 2x2 formation. And without capturing.


Would that really be useful? It takes many moves to assemble the group, and then disassemble it. The chances that the path will be unoccupied (or be left unoccupied by the opponent, when he sees you are assembling the group) seem close to zero.

HGMuller wrote:

Would that really be useful? It takes many moves to assemble the group, and then disassemble it. The chances that the path will be unoccupied (or be left unoccupied by the opponent, when he sees you are assembling the group) seem close to zero.

I was thinking of using relatively few pieces for such a large board, and relatively many knights so you really need to manage the presence of enough knights on the right place. And what about threatening to assemble a group forcing the opponent to do something about it.


The Cannibal Pawn: Before the game starts, each player writes on a piece of paper which one of their pawns (A-H) is the cannibal, ie. having the ability to promote into a bishop after capturing an enemy pawn, without showing that paper to their opponent until time comes to go cannibal. The Centaur: If you reach a queening square with your knight, that knight gets promoted to centaur which has the moves of both knight and queen. That means you can win even if you have only knight and king left. The Ghost: This piece moves at the outer files and ranks of an (imaginary) 10x10 chess board where the inner 8x8 squares are reserved for normal chess. On each turn, you make one normal move and one ghost move. The Ghost moves one square at a time in the rim but is also able to hop straight across the board (for example from F0 to F10) if there aren't any enemy pieces blocking it on that file or rank. The ghost only affects the enemy king and while unable to mate, if it manages to give check to the enemy king on the rim, your ghost disappears and your opponent loses a turn (both normal turn and ghost turn). The ghost doesn't have any effect on opponent's ghost and they can occupy the same square.


I have thought about it and i think the ghost variant should be reserved for mature audiences because of the sheer terror that it causes by ruining an otherwise beautiful game. However i think that by tweaking the rules it could allow some better strategic angle to the variant.

Teeer wrote:

The Centaur: If you reach a queening square with your knight, that knight gets promoted to centaur which has the moves of both knight and queen. That means you can win even if you have only knight and king left.

I think that it is too easy to do so and will lead to the board becoming full of centaurs. A variation is that the first time your knight reaches the end, it becomes a slight cenataur. the second time, a partial cenataur and the third time, a full cenataur. In my version, if a knight takes a cenataur (slight, partial or full), it goes up one level of cenataurness. In addition, if the cenataur is taken by a full cenataur, the cenataur can 'pass' the cenataurness onto a knight or slight/partial cenataur. This would take a turn. E.g if a knight reaches the end twice and becomes a partial cenataur, then is taken by a full cenataur, after the other player moves, the cenataur that took the piece can make anothr knight one level higher, even to a full cenataur.