The Chess Variant Pages
Custom Search

[ Help | Earliest Comments | Latest Comments ]
[ List All Subjects of Discussion | Create New Subject of Discussion ]
[ List Latest Comments Only For Pages | Games | Rated Pages | Rated Games | Subjects of Discussion ]

Comments/Ratings for a Single Item

Later Reverse Order EarlierEarliest
This item is a miscellaneous item
It belongs to categories: Orthodox chess, 
It was last modified on: 2011-01-08
 Author: Fergus  Duniho. On Designing Good Chess Variants. Design goals and design principles for creating Chess variants.[All Comments] [Add Comment or Rating]
Kevin Pacey wrote on 2018-09-19 UTC

To complete my examples of justifiable underpromotions in Seirawan Chess, here's a diagram where the move 1.Pf7-f8=A+ is the only way to win, underpromoting to an Archbishop. Note that if Black's king were initially on a8 (for example) then underpromoting to a Chancellor would be the only way to win.

Kevin Pacey wrote on 2018-09-18 UTC

Here's a possible Seirawan Chess position showing underpromotion to a Chancellor (best option, though not the only way to win). After 1.d7xc8=C+ ka7-b7 2.Cc8-e7+ White mates in two more moves. This appears to be faster than if White chose an Archbishop or R, while choosing a Q would stalemate Black.

Kevin Pacey wrote on 2018-09-17 UTC

@ Greg: I thought of a way to underpromote to a Chancellor just this morning, but felt it was too easy and not what you were initially doubting was feasible in the way of a justifiable underpromotion. I'll see if I can remember what I thought of.

Meanwhile, here's another Seirawan Chess possible position, this time with an underpromotion to a rook (I feel I thought of some other example position that was more crystal clear long ago, though); White's easiest way to win (if nothing else) is 1.Pe7xd8=R since if White chooses a B, Archbishop or a Q instead then 1...ac8-g4+ allows Black to at the least bother White for a while, since taking the archbishop stalemates Black. If White chooses a Chancellor instead then 1...ac8-b7+ may prolong the struggle considerably since taking the archbishop with either Chancellor stalemates Black.

Greg Strong wrote on 2018-09-17 UTC

Promote to a Chancellor

Aurelian Florea wrote on 2018-09-17 UTC

@Kevin Pacey

Nice one!

Kevin Pacey wrote on 2018-09-17 UTC

Here's a possible position from Seirawan Chess where White's best move is clearly 1.Pd7xc8=N, since promoting to a bishop or archbishop instead is meet by 1...ab8-d6+ and if the archbishop is taken by a White archbishop then stalemate occurs.

Kevin Pacey wrote on 2018-09-17 UTC

I think I could have replaced the White A/f8 with a N in the diagram of my previous post, and still 1.Pd7xc8=B would be the best move, though I chose to be 'safer' by using an A just in case it mattered (not only that, but I got to use one more fairy chess piece, too).

Whether the White piece on f8 is an A or a N, a question might be whether the initial position could ever arise after some logical previous move(s) (i.e. a 'prequel' of some sort). At least the initial position could have legally arisen, even if it had to be via bad play at least on Black's part. Perhaps putting White's K on a different square on the h3-c8 diagonal at the start might allow for a logical prequel somehow, if it's necessary/possible. [edit: if a White N is initially on f8, perhaps having a Black R on c8 instead of a Q could better help in this regard]

Fergus Duniho wrote on 2018-09-16 UTC

After White's pawn takes the Queen and promotes to an Archbishop, Black has only four moves available. Two of these moves would expose it to capture without any significant consequence for White, and another would allow White to move the Archbishop to b8. The remaining move is to go to d7, checking the King. If White responds by taking the checking piece with either Archbishop, the resulting position is a stalemate. If White responds by moving the King, Black can take either Archbishop en prise. If White responds with a block on e6 from the Archbishop on e6, Black can exchange Archbishops. This might still be favorable for White. If Black takes the Pawn at b5 instead, White can still probably force an advantage. So, it looks like White could still win if the Pawn promotes to an Archbishop. The advantage of promoting to a Bishop is that White can win quickly and decisively. If the Archbishop moves to d7 for check, the Bishop can take it without causing stalemate. If the Archbishop makes its only safe move to c7, then the Bishop can safely check the King from b7, forcing the King to move to b8. The Archbishop can then move to d7 for checkmate.

Aurelian Florea wrote on 2018-09-16 UTC

If your promote to a, after 1...ab8-d7 you can easily have 2.a c8-b7 with check and afterwards capture the enemy archbishop with your other one and then your are out of the woods :)!

Kevin Pacey wrote on 2018-09-15 UTC

Here's a possible position from Seirawan Chess where White's best move is clearly 1.Pd7xc8=B, since promoting to an Archbishop instead would allow a possibe stalemate after 1...ab8-d7+, if White captured the enemy archbishop either way. Similar positions for other variants such as 10x8 Capablanca Chess could be dreamt up.

Aurelian Florea wrote on 2018-09-15 UTC


In the era of computer chess promoting to captured pieces does not seem that relevant anymore. Your Gross chess idea on promotion is great. I had saw it once, but forgot the game and author and somehat use it in apothecary (the promoting to different things part not the captured pieces part). In the meantime I had read it again.

Fergus Duniho wrote on 2018-09-15 UTC

If promotion is only to captured pieces, as in Grand Chess, then underpromotion enables promotion when there are no captured compound pieces to promote to.

Aurelian Florea wrote on 2018-09-15 UTC

And being at the topic of promotion, I'd like to bring a discussion from this topic on. In CWDA I'm quite uncounfortable with pawns promoting to pieces from both starting armies (where the case). That is because well then the armies are less "different". The reason given by Betza is very sound. The pawns are then different and that difference should be accounted for. True. But for future CWDA if they are on larger boards (normally with more pieces) there is more room to optimize so different promotions for pawns would not be an issue !

Aurelian Florea wrote on 2018-09-15 UTC

I am encouraging promotion to weaker pieces on earlier ranks (technically not underpromotion I guess) but I am not sure why this principle does not catch steam :)! To me it seems extra choice and that is always good provided there are not clear inconviniences. And I see none :)!

Kevin Pacey wrote on 2018-09-15 UTC

I once figured out that it was okay to underpromote to a B or N instead of an Archbishop, if other type(s) of promotion caused a problem for some legal position(s) I dreamt up, in the case of my 10x10 Sac Chess variant, even, which has 10 possible piece types to promote to(!). Unfortunately, I didn't record the positions I thought of. If I think of examples of any of them later, I'll try to get around to posting them. In the case of avoiding a Chancellor promotion, for example, both a R or N promotion instead might still cause some sort of a problem that underpromoting to a B would avoid (though not if an Archbishop).

Greg Strong wrote on 2018-09-15 UTC

I agree with this principle and believe it is indeed sometimes violated.  Take all the Capablanca variants.  They typically allow promotion to any piece (Queen, Chancellor, Archbishop, Rook, Bishop, Knight.)  The last three are pointless, however.  You might under-promote to avoid causing a stalemate, but if the Rook-move causes the stalemate you promote to an Archbishop.  If the Bishop-move causes the stalemate, you prmote to a Chancellor.  Otherwise you promote to a Queen.  David Paulowich first pointed this out and that's why his games do not include pointless underpromotions (and hopefully mine don't either.)

Kevin Pacey wrote on 2018-09-15 UTC

It may be worth noting that in the case of variants where promotions occur, and then only on one rank (the last one!?), it would not be good game design to allow underpromotion to a given piece type, if there can never legally arise a position where the underpromotion would in some tangible way be a better move to play than promoting to a piece type of higher value. However, at the moment I cannot think of a game that would break this principle, assuming it's ever possible to do so.

Aurelian Florea wrote on 2017-04-01 UTC

Hello Kevin, Hello Fergus,

Actually I've noticed the camel issue on a 10x10 board in my own upcomming apothecary1 game, with the wizards being confident on exchanging favorably with black pieces. I currently don't know how to overcome this issue.Maybe enlarge the board.

Kevin Pacey wrote on 2017-03-31 UTC

Hi Fergus

One thing that's dissatisfied me a bit about a (small?) number of variants that I've invented is that I've discovered that the first player to move can threaten a checkmate (or alternatively, the second player's queen-like piece) immediately, while in at least one case (involving a variant with many pieces) a series of piece exchanges can be initiated immediately (apparently not favouring the first player, objectively). In the former instance this is due to a camel-like move on a 10x10 board (as a consolation to me, I've seen at least one other variant that has this 'flaw', but it actually happens to be a relatively popular game). This is not due to a pawn being unprotected in either side's camp in the setup, such as in the original setup for 10x8 Capablanca Chess, which is something your article does allude to.

Speaking of Capablanca Chess, just a reminder that I encountered a bug in the preset for that variant in a game with Carlos - see my somewhat dated Kibbitz Comment.

Fergus Duniho wrote on 2017-03-31 UTC

Regarding Rococo, it is currently the 14th most played game on Game Courier out of a total of 1141 games. In the past 90 days, it has been played five times, though every single one of those games was played between George Duke and someone literally called "?? ??". So that doesn't say much for its current popularity. Personally, I don't play it myself, because I'm not into Ultima style games. So, I never wrote up these design goals and principles with Rococo in mind as an example, and perhaps it is difficult for these kinds of games to not be overcomplicated, though I think David and Peter did try to make it less overcomplicated than Ultima.

Regarding games without rule enforcement, I usually avoid playing them myself, and I don't give them the same promotion as I give to some games with rule enforcement. So I do think it is a good idea to provide rule enforcement if you want people to play your games on Game Courier.

George Duke wrote on 2017-03-27 UTC

Contrary to Kevin, Rococo is the most played not the least played lately in recent finished Games (rhymes with Trump and crowd size). The article is too lengthy for having two inventors. My top CV of all 3000 in sixty words, enough to start playing immediately, the whole simplified rules:

All pieces have to move the same  like Queen but land on border square only if/in capturing. Pieces capture along Q lines like in  Abbott's Ultima and in Parton's classic CVs. See Advancer, Withdrawer, Chameleon, Swapper, Immobilizer capturing in video. King is f.i.d.e. and the only one moving and capturing the same. Cannon Pawn is specialized one- or two-stepper, also having unique capture.

Kevin Pacey wrote on 2017-03-22 UTC

On "Don't overcomplicate your game", I'm wondering about even the popular variant Rococo perhaps not quite satisfying this criteria. Just looking at how long the entire description of the various pieces' powers is for the game, it does seem there are many rules (plus arguably unfamiliar piece types) - and perhaps because I am not always inclined to learn a hard game to play, I haven't hurried to do so in this case either. Note that people can understand the basic flow of a physical sports game without knowing every rule, but it seems tougher to do so for even a game of chess, which seems less complex rule-wise than Rococo does - yet I suppose the latter is an enjoyable game once one fully learns it. Lately it does not seem much played on Game Courier, perhaps because there are a lot more relatively new people to CVP (like myself) playing nowadays.


On an unrelated matter, other than reviews or comments, it's a bit hard to tell if one has invented a chess variant that will ever be played or liked much at some point. So far I've submitted 17 variants, and 11 of these have Game Courier presets. Of the latter, only two games of mine have been played more than once (or in the case of many, if at all), so far. At one point I gave myself the objective of inventing at least one game per certain common types of board terrain that could be played on Game Courier, if there ever were a preset made for each game. Later I thought more about how to design a game that might have a strong chance of being popular, and Butterfly Chess had some success, though not as much as when I lucked out as a beginner to some extent with Sac Chess. One problem may be that that most of my presets are not rules enforcing, which may give a game a better chance of becoming popular; another thing that might make a game popular on Game Courier might be simply being an inventor and player familiar to many Game Courier players, ideally with a history of inventing a number of successful variants. Anyway, my modest 'success' with 2 out of 11 presets thus far is a success rate inventors in other fields of endeavour might not feel too bad about, especially if just starting out.

Butterfly Chess

Sac Chess

Jörg Knappen wrote on 2012-03-06 UTC
[25] There is Zugzwang: players with legal moves are obliged to move even if every legal move leads to defeat.

This is one of the most outstanding features of chess and its variants. Compare it to go, where no player is ever forced to deteriorate their position, they may just pass instead.

Jörg Knappen wrote on 2012-03-05 UTC
Here's my interpretation of A. Blacks criterion 17:

17. Pieces moves like chess pieces can move.

(a) Pieces move like leapers (true leapers or 'lame' leapers), riders, chinese or korean cannons, or combinations of those.

(b) Pieces have highly symmetric movement patterns (full 90 degrees rotational and reflectional symmetry for all non-pawn pieces, reflectional symmetry with different forward and backwards movement [like in the Shogi gold and silver generals] counts as a mild violation of this)

(c) Pieces move and capture the same way or their move and capture are at least 'similar' in some sense (I consider the pawn movement and capture similar because of forwardness and shortrangeness, also the pieces of separate realm chess or chinese cannons are similar in movement and capture. Frank Maus' knibis and bishight aren't). 

This allow much more pieces than just the traditional FIDEs ...

Jörg Knappen wrote on 2012-03-01 UTC
Back to the terrain question: a promotion zone does not constitute terrain for me, also the forward direction of pawns is not dictated by terrain.

Holes in the the board are somewhat strange to Chess and may constitute terrain. Barriers of all kind are certainly terrain.

Possible terrain effect are: Difficult terrain (mountains, swamps) slowing units (pieces) down or forbidding some kind of pieces (two heavy to move there ...) on that terrain, land/water distinction (land units need boats or bridges to cross the water, water units cannot move on land (but maybe shoot units down on land), air planes can operate both on land and water, but need to land after some time and need airports or carriers for this purpose), cities (providing supplies fo any kind, generating new units, allowing of repair of damaged units).

This leads to another chess criterion

[24] A chess piece is either fully functional or captured, there is no such thing like 'damage' or 'health' with consequences to the piece (slower motion, need of repair, easier capturability). Of course, a bad position (e.g. pinned) does not count as damage. In FIDE chess the only (very mild) violation of the no damage rule is the loss of castling rights.

25 comments displayed

Later Reverse Order EarlierEarliest

Permalink to the exact comments currently displayed.