If I were a chess variant enthusiast who invents and proposes his own variants, I might propose some of the following. The nicest are maybe §2 (weakened crowned bishops in particular), §8 (with sergeants), §9, §10, and §13. -- As far as already existing variants are concerned, my favourite ones are listed in the homepage of my chess web space.
§1 Paradigm dragon Seirawan
Seirawan Chess, with the following change: the special pieces are a bishop-xiangqihorse hybrid (aka Paradigm dragon) and a rook-xiangqihorse hybrid. (One can propose an equivalent change for Illescas Dragon Chess as well, or for any kind of Capablanca/Gothic Chess. For other similar variants, where the knight-queen hybrid is present too, this piece can be replaced by a queen-xiangqihorse hybrid. -- You can try some sort of Paradigm dragon Seirawan, as a custom variant, on www.chess.com/variants/custom: with two Paradigm dragons, since the other hybrid is not available.)
§2 Modified versions of Paradigm Chess30
I love Paradigm Chess30. Nevertheless, sometimes I think those modified bishops are too strong. Is that fair? Can rooks still be called major pieces, in Paradigm Chess30? And, poor weak knights! On the other hand, the fans of Seirawan Chess or Capablanca/Gothic Chess will find that Paradigm dragons are not powerful enough. For all these reasons, here's a list of suggestions for modified versions of Paradigm.
[Replace the rooks too.] Standard chess, with the following change: each player has bishop-xiangqihorse hybrids (aka Paradigm dragons) and rook-xiangqihorse hybrids, in the place of the bishops and the rooks, respectively. Castling as in standard chess. Pawns can promote to any of the pieces that are present in this variant: no rooks, no bishops. This variant can be seen as a modified Paradigm Chess30, and can be played with the same type(s) of randomization -- either 30 or 1230, see section §5 of my web page on Paradigm Chess30. (By the way, I think that standard chess starting position is always a valid alternative, even for Paradigm Chess and for all these modified versions.)
[Replace the rooks and the queen too.] The same as above, but the queen is replaced too: here's a queen-xiangqihorse hybrid in the place of the queen. Pawn promotion, same principle as above. Randomization(s) as above.
[Very powerful piece replacements: leaping hybrids.] Paradigm Chess, or any of the above Paradigm subvariants, with knight hybrids precisely in the place of xiangqi horse hybrids: so, knight-bishop hybrids instead of Paradigm dragons, etc. Pawn promotion, same principle as above. Randomization(s) as above.
[Put away the dragon wings and crown your bishops!] Standard chess, with the following change: each player has non-royal king-bishop hybrids (aka wazir-bishop hybrids or crowned bishops or missionaries, in fairy chess terms) in the place of the bishops. Pawn promotion, same principle as above. Randomization(s) as above. -- Further subvariants: you could crown the rooks too (and/or the knights, for which see the next point).
[Replace the knights as well.] Paradigm Chess, or any of the above subvariants, with modified knights: (a) non-royal king-knight hybrids (aka crowned knights) in the place of the knights, or (b) non-royal king-xiangqihorse hybrids (aka crowned xiangqi horses) in the place of the knights. Pawn promotion, same principle as above. Randomization(s) as above. -- This simple idea, crowning the knights/horses (or the rooks, or both!), can be applied to standard chess too; no Paradigm-style randomization in this case (but 18 and 960 are possible, of course). Cfr. Betza's Augmented Knights / Different Knights, and the good simplified versions to be played on greenchess.net, links here and here; if you want crowned knights, try this.
[Peaceful crowning subvariants.] For each of the above subvariants involving non-royal king hybrids (aka crowned pieces), we can propose a further subvariant with peaceful crowning: the king move of peaceful-crowned pieces cannot be a capture -- not even a potential king-capturing move, for check and checkmate. (The diagonal moves of crowned bishops never count as king moves; the same is valid for the orthogonal moves of crowned rooks. -- My idea of peaceful crowning, probably not new, was inspired by the rules of the Spartan Chess piece called lieutenant or tagmatarchos. -- About peaceful-movement rules and conventions, see section §5 of my page On player-friendly piece design, and section §2 of my page Pictures.)
[Weakened peaceful-crowned bishops: horizontal-only or vertical-only peaceful movements. This is maybe the most basic modification to the rules of the bishops, in order to allow Paradigm-style randomization, either 30 or 1230.] (1) Standard chess, with the following change: the bishops can also move (but not capture) one square horizontally, like Spartan Chess lieutenants. (2) Exactly the same, but vertically instead of horizontally (cfr. the proposal 10a in section §1, Some possible variations on the rules of Spartan Mirror, of my page on Spartan Chess).
[Peaceful knighting/horsing too?] Peaceful knighting follows the same principle as peaceful crowning: the knight move of peaceful-knighted pieces cannot be a capture -- not even a potential king-capturing move, for check and checkmate. The same can be proposed for xiangqi horse hybrids too. For example, a bishop-xiangqihorse hybrid with peaceful horsing will be a very interesting weakened version of a Paradigm dragon: one should try a Paradigm Chess subvariant with this change. The principle of peaceful knighting/horsing can be applied to any (sub)variants where knight hybrids or xiangqi horse hybrids are present -- consider peaceful-knighted or peaceful-horsed kings too, either royal or non-royal!
[Update, 30 December 2024. Bishop adjustment/conversion rules.] Reading about variants on 9x9 and 9x8 boards (see my page on Spartan Chess, section §3), I found out that several variants adopt the Bishop Adjustment Rule (chessvariants.org link here) or the Bishops Conversion Rule (chessvariants.org link here). They are ways to turn a dark-squared bishop into a light-squared one, or vice versa, only once in a game for each player. If you like Paradigm-style randomization (either 30 or 1230) but want to keep the standard chess bishops, on a standard 8x8 board, you can introduce (and customize to your preference) some version of the bishop adjustment/conversion rule.
§3 Turn pawns into xiangqi horses, or into other minor pieces, with no need of reaching the eighth rank --not so interesting
[Let's try to make it short: one default subvariant only.] In addition to the standard chess rules, each pawn can turn into a xiangqi horse, this way: make a legal non-promoting pawn move, capturing or not, and then change the same pawn to a xiangqi horse, in the same turn. This change is always irreversible. [Further subvariants below.]
Standard chess, adding the following rule: the pawn-to-xiangqihorse change. Pawns follow the standard rules, including the two-square advance, en passant, and promotion. But, in addition to that, each pawn can turn into a xiangqi horse, according to either pattern (1) or pattern (2). (1) Make a legal non-promoting pawn move, capturing or not, and then change the same pawn to a xiangqi horse, in the same turn; (2) change a pawn to a xiangqi horse, and then make a legal move with that same horse, capturing or not, in the same turn. The pawn-to-xiangqihorse change is always irreversible. The check and checkmate rules remain unchanged: you can give check or checkmate with this special move and, on the other hand, it can be a get-out-of-check move. (The first half of the special move is not relevant or valid for check and checkmate.) Promotion remains unchanged too: no promotion to xiangqi horse (if it were allowed it would be a senseless choice, since a horse is even weaker than knights and bishops). Two further subvariants are possibile, with fixed order: move first and then change to xiangqi horse; or, inversely, change to xiangqi horse first and then move. Move-and-then-change is similar to promotion: therefore, it may be considered more elegant and chess-like, and that's one of the reasons why I think this fixed-order subvariant should be the default choice. Note that, if change-and-then-move is allowed, one must pay attention to heterodox (distance) pawn attacks, including checks and checkmates!
The same as above, but each pawn-to-xiangqihorse change takes one full turn, with no movement or capture. You are always allowed to make this special move, for any of the pawns you have -- unless you are in check, of course. (This subvariant is simpler, but probably less interesting.)
The same as above, in any subvariant (free order / move and then change / change and then move / change only), but pawns may turn into different minor pieces: (a) xiangqi horses only, as already seen; (b) wazirs; (c) knights; (d) bishops; (e) knights and bishops; (f) non-royal kings; (g) non-royal kings and knights; (h) non-royal kings and bishops; (i) non-royal kings and knights and bishops. In all cases, I think promotion would better remain unchanged: pawns promote to standard chess non-royal pieces only (most likely a queen, of course).
Set a limit! How many times can each player make the special move in a game? The starting idea was setting no limit (i.e. eight times, since the pawns are eight), but it's probably more sensible to establish a maximum such as three, two, or only one, also depending on the individual subvariants. So, choose any number, from one to eight.
§4 Variants with a more powerful king
Standard chess, with the following change: each player has a royal king-xiangqihorse hybrid in the place of the king.
Standard chess, with the following change: each player has a royal king-knight hybrid in the place of the king. (I think this variant already exists. For sure it can be set as a custom variant on www.chess.com/variants/custom. I have played several games and like it.)
Postscript. See section §13 for variants with much more powerful royal pieces. Even for the variants above, skilled players might want some anti-draw correction (promotion to a queen-knight hybrid, etc. -- see §13 again).
§5 Other xiangqi-inspired and Synochess-inspired variants: cannon or non-royal king instead of queen --not so interesting
Standard chess, with the following change: each player has a cannon in the place of the queen. (The same change applies to pawn promotion: no queen is allowed -- you can promote to a cannon, but a rook will usually be better of course.)
Exactly the same as above, with a non-royal king in the place of the queen. (I suppose this variant may already exist. I have tried it, as a custom variant, on www.chess.com/variants/custom, but I can't bear the Chess Dot Com symbol convention for royal vs. non-royal kings, when both are present.)
§6 A multifaceted variant with weakened pieces, inspired by ChinKor chess (that is xiangqi and janggi) --not so interesting
This "multifaceted variant" is probably not so interesting, not so well-designed. You may skip and ignore what I proposed as "default choices", and jump to the suggestions for a basic subvariant, at the end of this section §6.
Standard chess, with at least the following changes: xiangqi horses (aka ChinKor horses) replace the knights, and ChinKor elephants replace the bishops. There is no palace and no river: all pieces can move throughout the board.
Elephant subvariants. The elephants can be: (a) Chinese elephants; or (b) Korean elephants; or (c) Chinese-plus-Korean elephants, meaning that both the xiangqi move and the janggi move are allowed; (d) non-leaping ferz-alfil hybrids, aka non-leaping ferfils/falafels (1-or-2-square diagonal move); (e) leaping Chinese elephants, aka alfils/alpils/pils; (f) leaping Korean elephants; (g) leaping Chinese-plus-Korean elephants; (h) leaping counterpart of (d), that is ferz-alfil hybrids, aka ferfils/falafels or modern elephants or some further silly and useless names. The default choice should be (d), because that is a reasonably weakened (short-range) bishop, and is also very Chinese-flavoured, since it can be seen as a perfect advisor-elephant hybrid, in xiangqi terms; but (c-g-h) are very stimulating possibilities too. Hybrid versions of Korean elephants could be proposed as well.
Queen subvariants. (a) A cannon replaces the queen (cfr. section §5); or (b) a non-royal king replaces the queen, as in Synochess (cfr. section §5 again); or (c) a weakened short-range queen replaces the queen (1-or-2-square move: a non-leaping ferz-alfil-wazir-dabbaba hybrid in fairy chess terms -- also called a bodyguard!); or (d) leaping counterpart of (c); or (e) a rook-xiangqielephant hybrid replaces the queen; or (f) a rook-janggielephant hybrid replaces the queen; or (g) a rook-elephant hybrid replaces the queen, based on the very same elephant type that has been chosen as a bishop replacement (see Elephant subvariants above); alternatively, (h) the queen remains as in standard chess, which is the least ChinKor-like option. The default is (c), but cannon enthusiasts don't have to be sad. See Rook subvariants and Further cannon subvariants and rules.
Rook subvariants. A rook is a rook, with no variation of its rules. But, in the default subvariant, (a) each player has only one rook, the other being replaced by a cannon. You may castle either with the rook or with the cannon, following the same rules as in standard chess, if castling is allowed (see Castling subvariants). There is a setup choice before the game starts, similarly to the horse-elephant switch in janggi. White decides the position of his/her own rook and cannon (squares a1, h1); after that, based on White's decision, Black does the same on his/her side. Alternatively, (b) each player has two rooks, in the usual corner positions.
Further cannon subvariants and rules. (1) If each player has two rooks (see Rook subvariants above) and the pawns start on ranks 3 and 6 (see Pawn subvariants below), a white cannon and a black cannon will be inserted on ranks 2 and 7 respectively. In this case, too, there is a setup choice before the game starts: White places the cannon on one of the eight empty squares and, after that, Black decides the position of the cannon on his/her own side. (2) Xiangqi or janggi cannons? (a) The Chinese-rule cannon is default in this chess variant, but (b) the Korean-rule one is allowed too: see the general notes towards the end of this web page. (3) In any subvariant where cannons are present, either of the players cannot move his/her cannons in the first turn.
King subvariants. The face-off or facing-each-other rule: (a) does not exist; or (b) is the same as in xiangqi, that is vertical only; or (c) is the same as in Synochess, that is on files and ranks too; or (d) is queen-style, that is even on diagonals. The default is (b). In addition, the Korean variant of the rule could also be taken into consideration. (Non-royal kings are never affected by these rules. Note that the face-off rule regards the usual restrictions on castling too, if your subvariant allows castling.)
Castling subvariants. (a) The same as in standard chess; or (b) opposite castling, meaning that same-side castling is forbidden; or (c) no castling at all, which is the default.
Pawn subvariants. (a) The same as in standard chess, which is the default; or (b) standard chess pawns, but no two-square advance, and consequently no en passant; or (c) the same as (b) but with a change in the initial setup: the pawns start on ranks 3 and 6.
Promotion subvariants. (a) Pawns promote to non-royal kings only (default choice -- see also section §7); alternatively, (b) pawns promote to the pieces that are present in the initial setup.
Short suggestions for a basic subvariant. Standard chess, with the following changes: xiangqi horses replace the knights; short-range bishops replace the bishops; a short-range queen replaces the queen; pawns promote to non-royal kings only. (Short-range means: non-leaping, 1-or-2-square move.)
Other basic subvariants. The same as above, but with a different queen replacement: for example, a non-royal king. (You can also consider Pawn subvariants.)
§7 Some weakened-promotion variants, approximately arranged from highest to lowest weakness
Standard chess, with the following change: pawns promote only to ferzes (... as in the old days!).
Standard chess, with the following change: pawns promote only to wazirs.
Standard chess, with the following change: pawns promote only to xiangqi horses.
Standard chess, with the following change: pawns promote only to bishops and knights. (I suppose this variant may already exist.)
Standard chess, with the following change: pawns promote only to cannons.
Standard chess, with the following change: pawns promote only to bishops, knights, and cannons.
Standard chess, with the following change: pawns promote only to non-royal kings. (I suppose this variant may already exist.)
Standard chess, with the following change: pawns promote only to nrK-xiangqihorse hybrids.
Standard chess, with the following change: pawns promote only to nrK-knight hybrids.
Standard chess, with the following change: pawns cannot promote to queens (so that rooks will be the most common choice).
§8 Replace the pawns
Each player has eight non-royal kings in the place of the pawns. (1) No promotion, or (2) the nrK's are promoted on reaching the eighth rank like pawns in standard chess, or (3) see section §7 above, Some weakened-promotion variants, excluding the weakest options. (a) No two-square move and no en passant, or (b) one or another adaptation of the two-square move and en passant.
Some further subvariants. Other weak minor pieces in the place of the pawns:
eight nrK's with no backward moves;
eight ferzes;
eight ferzes with no backward moves;
eight wazirs;
eight wazirs with no backward moves;
eight xiangqi horses;
[update, 6 December 2024] eight Chess Dot Com sergeants in the place of the pawns. (The sergeants, that is standard pawn plus Berolina pawn hybrids, are very aggressive pawn-like chessmen, with the two-square move and en passant, and promotion of course. This can be set as a custom variant on www.chess.com/variants/custom. I like it very much.)
§9 See my page on Spartan Chess and Spartan Mirror, and other variants with more than one king per player
On that page I propose some variants such as:
Spartan Mirror with reverse symmetry aka rotational symmetry (so as to discourage early crowned rook exchange).
Spartan Mirror crazyhouse, Spartan Mirror bughouse; Spartan Chess crazyhouse, Spartan Chess bughouse.
Subvariants of greenchess.net Weak Dual Chess with both bishops: white's baseline RNBKKBNR (two regular rooks and no queen) or, probably better, QNBKKBNR (with a potentially castling-involved queen) -- either with or without the duple check rule.
Adaptations of greenchess.net Weak Dual Chess on a 9x8 board: RNQBKKBNR should be the best choice -- either with or without the duple check rule.
Weak Dual Knightmate, that is Knightmate combined with greenchess.net Weak Dual Chess.
§10 More variants with fairy pieces, based on some variants I have tried on greenchess.net (January 2025)
Replace the queen with another strong piece. Everything stays as in standard chess, except that the queen is replaced by one or another fairy chess piece. (Promotion to the pieces that are present in the initial setup, except the standard king.) This kind of variant is already present on greenchess.net, but only with Q+N and R+N pieces, under the names of Superqueen Chess and Almost Chess. (Almost Chess was invented by Ralph Betza in 1977. Almost Chess is a childish and silly name. Tens of chess variants are kinds of "almost chess", that is "almost standard chess", since only one piece or rule is modified.) I would find it better to have Superqueen Chess and Almost Chess as setup choices of one single (or two) macro-variant(s), as with Augmented Knight on greenchess.net (link here). The collective name could be something like Strong Replacement for Queen. Beside (a) Q+N and (b) R+N, the other choices would be: (c) B+N; the threefold hybrids (d) R+N+K and (e) B+N+K; at last, much less strong, the (twofold) crowned hybrids (f) R+K, (g) B+K and (h) N+K. (In all these variants, the only royal piece is the standard king. For very weak queen replacements, see §5 above.)
Crowned Perfect Chess. One of the nicest greenchess.net variants is Perfect Chess (link here), a "chess variant containing all possible piece combinations of rook, bishop and knight [...]. All piece combinations, including the basic pieces (rook, bishop and knight), occur exactly once on each side". That could be renamed Perfect Chess with Knight Hybrids, or Knighted Perfect Chess. Another similar variant (my idea) would be Crowned Perfect Chess, with (non-royal) king hybrids. (a) White's baseline: R+K, N+K, B+K, Q, K, B, N, R; or (b) the same, with threefold R+N+K or (c) B+N+K in the place of the Q. (As usual: promotion to the pieces that are present in the initial setup, except the standard king. Queenside castling involves the R+K piece. In all these variants, again, the only royal piece is the standard king. -- As for vocabulary, I never say or write chancellor, archbishop, hawk, centaur etc.: see my web page On usable and sensible names and identities for fairy chess pieces.)
[New, 9 April 2025.] Knighted Perfect Chess and Crowned Perfect Chess on a 9x8 board, keeping both standard bishops. A bishop is a bishop, but a light-squared bishop is not a dark-squared bishop. Let's save the pair! One (good and reasonable) idea is to put away the Q or the Q+N piece, on the regular 8x8 board. But, instead, let's move to a slightly widened board, 9x8, as in section §3 (Multiple kings on nine-column boards) of my page On Spartan Chess and Spartan Mirror, and other variants with more than one king per player. We'll stick to the model of the greenchess.net 9x8 variants, Active Chess and Superqueen Chess II, with a Q on the i-file, and the same castling patterns. 9x8 Knighted Perfect Chess, White's baseline: R+N, B+N, B, Q+N, K, B, N, R, Q. 9x8 Crowned Perfect Chess, White's baseline: R+K, N+K, B, B+K, K, B, N, R, Q (or the same with a threefold hybrid in the place of the Q -- see above). [... Some months later, let me add a little note, regarding both types of 9x8 Perfect Chess. I think that the i-file piece should better be a leaper, because of the after-castling position. So, White's baseline could be: ... Q, K, B, N, R, Q+N. Allow choosing among more than one initial setup, especially for 9x8 Crowned Perfect Chess -- not excluding some knighted-crowned combinations, both 8x8 and 9x8, such as (9x8) R+K, B+N, B, R+N, K, B, N, R, Q+N, or R+N, N+K, B, Q+N, K, B, N, R, B+N, etc..] [Yet another unmissable postscript, 6 April 2026. This new subvariant should be named (Emanuele Saiu's) Ultimate (Knighted) Perfect Chess! In italiano, scacchi perfetti (cavallizzati/equini) definitivi. As I was saying above, a good and reasonable idea is to keep the bishop pair and to put away the Q or the Q+N piece, on the regular 8x8 board. If we do so, choosing not to use the Q+N, we return exactly to the variant called Chess 2000 (by Gerhard Josten)… which is arguably superior to Perfect Chess (by Köksal Karakus), but the latter is more familiar to me since I can play it on greenchess.net. More than 25 years have passed since both these variants were invented. (And, for that matter, were they really new even back then?) Coming to the point… How will I make my "Ultimate Perfect Chess" truly ultimate? I have decided to draw inspiration from my recent experience with the amazon promotions of section §13! Therefore: exactly as in Josten's Chess 2000, we will have R+N, B+N, B, Q, K, B, N, R. On top of that, in promotion it will also be possible to choose the Q+N: a most special bonus for deserving pawns. Thus, and only thus, is the variant truly perfect. Isn't it?]
Crowned hybrids in basic variants modelled on Augmented Knight. Another of the nicest and simplest greenchess.net variants is Augmented Knight (link here), plus Asymmetric Augmented Knight (link here). It's an adjustable chess variant, with four different kinds of modified knights. They are very interesting. But a question arises: why not to include N+K too? The N+K hybrid would be stronger than any of those four augmented knights, but could well be included among the setup choices of the non-asymmetric subvariant. On greenchess.net we have the four augmented knights, and even the "short rook" (moving at most four squares), but unfortunately almost no trace of crowned hybrids, that is (non-royal) king hybrids. The N+K is present only in a 10x8 variant, Centaur Chess (link here). Adding new variants with N+K, but also B+K and R+K, would be a great improvement. I mean simple and basic variants, similar to Augmented Knight; perhaps with combinations as well, but simpler is better. (See other B/N/R replacement ideas in §2 above; and see the entries above, Replace the queen with another strong piece and Crowned Perfect Chess, with several non-royal king hybrid proposals.) [Update, 15 August 2025. I have played some games of "Normal chess, but non-royal N+K hybrids instead of knights", as a custom variant on Chess Dot Com. I like it! Unfortunately, Chess Dot Com doesn't have B+K's or R+K's either.]
[New, 19 August 2025.] Some variations on Knighted Perfect Chess as custom variants on Chess Dot Com (standard 8x8 board), www.chess.com/variants/custom. Unfortunately, Chess Dot Com doesn't allow you to change the rules of castling: for example, you can't castle with the king and the rook+knight hybrid (which is the rule of queenside castling in Knighted Perfect Chess). Anyway, you can design some nice variations on Knighted Perfect Chess; and the very good thing on Chess Dot Com is that you'll easily find opponents for playing your variants in real time -- some sort of opponents, even if not always good players... Now, among my saved game templates (Custom Rules), I have added the following three. They are all without the standard Q, but with the Q+N, R+N, and B+N hybrids. (a) Two knights and no bishops. White's baseline: R, N, R+N, Q+N, K, B+N, N, R. (b) Two bishops and no knights. R, R+N, B, Q+N, K, B, B+N, R. (c) Two bishops and one knight, and only one rook (kingside castling only!). R+N, B+N, B, Q+N, K, B, N, R. Promotions are different in each subvariant: you can always promote to the pieces that are present in the initial setup, except the standard king. -- I enjoy playing these variations.
§11 Variations on Mansindam aka Pantheon Tale: Custom Simplified Mansindam
Mansindam (English pronunciation /mænsɪn'dæm/, if the Youtube people say it right) is a nice and difficult variant, invented by a person from Korea (Daphne Snowmoon), and playable on PyChess since Summer 2023. It's based on shōgi (Japanese chess) and on heterodox western chess with knighted and crowned fairy pieces.
Read section §9 (About Mansindam aka Pantheon Tale) of my page On player-friendly piece design, link here.
However, even if an entirely self-explanatory set of pieces for Mansindam were made available on PyChess (which would be so easy, replacing the ugly unitary symbols for a few hybrids -- only four out of many), still I couldn't play it. I would try some games, or maybe a lot, with lowest-level AI, but it's too complicated: a large board, the drop rule, weak pawns, strong and superstrong pieces... For the time being, February 2025, I'm still a bad beginner at shōgi. (How embarrassing!)
Mansindam is attractive, but too hard for me. Too hard, but attractive. That's why I'd like to propose some simplified subvariants of Mansindam.
a. Board dimensions: 1. 8x8 as in western chess, or 2. 9x8, or 3. 9x9 as in shōgi and Mansindam.
For the 8x8 board (a.1.), we have to decide on the initial setup:
Eliminating one N per player is the most natural and advisable option;
If you want to make it simpler, each player will keep both N's but have no Q+N piece.
(In either case, two or three different setups might be available, but it seems to me that we'd better eliminate the first N or the second N or the Q+N piece, with no other change.)
b. 1. Mirror symmetry as in western chess, or 2. reverse symmetry aka rotational symmetry as in shōgi and Mansindam.
c. This is the most important variable: 1. No drop rule (a captured piece disappears forever), or 2. drop rule as in shōgi, Mansindam and crazyhouse.
d. The pawns start on 1. the second rank as in western chess, or 2. the third rank as in shōgi and Mansindam.
e. We may have 1. western chess pawns or 2. shōgi pawns.
In the case of western chess pawns (e.1.) starting on the second rank (d.1.), there is a further choice:
two-square move and en passant, or
no two-square move and no en passant.
With the two-square move and en passant, if the drop rule (c.2.) is active, then dropping a P on the first rank is prohibited, as in crazyhouse.
If we choose drop rule (c.2.) and shōgi pawns (e.2.),
giving checkmate by dropping a P may be allowed as in Mansindam, or
prohibited as in shōgi;
and
two or more unpromoted P's on the same file may be allowed, or
prohibited as in shōgi and Mansindam.
f. Promotion 1. on last rank only, as in western chess, or 2. on last two ranks, or 3. on last three ranks, as in shōgi and Mansindam. (Promotion is always compulsory, as in both western chess and Mansindam. The general rules of promotion are those of Mansindam: with the crowned pieces, of course.)
g. Castling? 1. Castling is allowed as in western chess, or 2. opposite castling only is allowed (meaning that you can't castle same-side if your opponent has already castled), or 3. castling doesn't exist, as in shōgi and Mansindam.
On the 8x8 board (a.1.), with castling allowed (g.1. or g.2.), both long and short castling are possible, depending on which side, with the K always moving two squares, as in western chess. Instead, on 9x8 and 9x9 boards (a.2. and a.3.), with castling allowed (g.1. or g.2.), we have two options:
the castling K always moves two squares, or
the castling K always moves three squares.
h. Campmate? 1. No campmate rule, or 2. campmate rule as in Mansindam, or (why not?) 3. hillmate rule as in King of the Hill. (On the 9x9 board, a.3., the hill is the central 3x3 area. On the 9x8 board, a.2., it is 3x2: 3 horizontal, 2 vertical.)
i. Being stalemated is 1. a draw as in western chess, or 2. a loss as in shōgi and Mansindam.
j. Threefold repetition 1. leads to a draw, if claimed, as in western chess, or 2. is prohibited as in Mansindam.
k. Fifty-move rule? 1. Fifty-move draw rule as in western chess, or 2. no fifty-move rule.
l. Draw by agreement is allowed. In shōgi and Mansindam, it doesn't exist! -- Well, ok: 1. It is allowed, or 2. it is not.
How many combinations. Shall I list my favourite ones? Maybe. I have to think over that.
The name... Mansindam is a nice abracadabra name. Isn't it? It means something? Is a man in a dam? I don't think so, I don't know. In any case, as far as my customizable proposal is concerned, I'd like to call it maybe Mansindames (pron. /mænsɪn'dæmɪs/), that is Mansindam + E.S.! But that's a joke. A more serious and neutral alternative is Custom Mansindam, or Custom Simplified Mansindam.
I might make Mansindam cardboard tiles, with exactly the same principles of my cardboard set for shōgi. This way, playing standard Mansindam or any kind of Custom Simplified Mansindam over the board will be possible and comfortable, providing opponents are available -- while waiting for the arrival of Custom Simplified Mansindam on some LiChess-like platform, perhaps a LiMansindam. :-)
§12 About shōgi with four- and eight-directional knights
See my pages on shōgi (shogi), aka Japanese chess: link here in English (whole page), collegamento qui in italiano (sezione §6), and shōgi photographs. On the page written in English, go to sections §6 (Some shōgi variants) and, especially, §7 (Some ideas about eight-directional knights).
§13 Standard-chess-based variants with a royal queen or another major piece in the royal role [new, 23 August 2025]
This could be an extension or a generalization of section §4 (Variants with a more powerful king). But, here, we consider much more powerful royal pieces.
Some days ago, I was trying to invent and test some custom variants on Chess Dot Com (see §10 above). On Chess Dot Com you can easily find opponents for playing your variants in real time, and that's exciting and stimulating.
How to exploit the available fairy pieces? For example, we have the Q+N (aka amazon or superqueen), which is not present in the most important and widespread variants. (Unfortunately, Chess Dot Com still lacks very interesting hybrids such as R+K, B+K, R+N+K, B+N+K...)
At a certain point, I decided to concentrate on the royal piece. The basic idea is to have a queen or some other queen-like piece as the royal piece (in the place of the king), while all the rest stays as in standard chess (including the non-royal queen, of course). I soon realized and verified that the result is probably too drawish, because it's difficult to checkmate a very mobile piece. So, we should probably add some sort of correction.
Three variables: a. which piece as the royal piece; b. which special promotions, if any; c. bare royal piece rule or not, in its proper form or in a modified one.
a. I tried with 1. the royal Q and 2. the royal Q+N. Any other queen-like piece could be tested too (unfortunately, only some are available on Chess Dot Com, such as the R+N).
b. Pawn promotions: 1. as in standard chess; 2. you can promote to a Q+N too (non-royal, of course).
c. Bare king rule, as in old Arabic chess? (That is: an additional way of winning a game is capturing your opponent's last remaining non-royal chessman -- unless your opponent does the same in their next half-move, in which case it is a draw.) 1. No bare piece rule, as in standard chess; 2. bare piece rule; 3. bare piece rule, but the strong side wins the game only if they have at least 3.1. one (non-royal) queen-like piece, or 3.2. two (non-royal) queen-like pieces, or 3.3. three (non-royal) queen-like pieces, or 3.4. one (non-royal) Q+N, or 3.5. two (non-royal) Q+N's, or 3.6. three (non-royal) Q+N's, or 3.7. two (non-royal) queen-like pieces of which at least one is a Q+N, or 3.8. three (non-royal) queen-like pieces of which at least two are Q+N's.
(Chess Dot Com allows you to set the bare piece rule, but only in its proper form.)
I find that a.1. b.2. c.1. is a very enjoyable subvariant, but more skilled players may prefer some further anti-draw corrections, such as those suggested in point c.3.
The option a.2. (royal amazons!) is also very interesting, but the anti-draw corrections are even more necessary in that case.
Postscript. On 21 March 2026, I invented the following two variants. They work well! I am very proud of them.
Royal archbishop (B+N) in the place of the K, with another B+N in the place of the Q; pawns can be promoted to amazons (Q+N).
Royal rook (R) in the place of the K, with a R+N in the place of the Q; pawns can be promoted to amazons (Q+N).
Consequently, I have changed the title of this section to: Standard-chess-based variants with a royal queen or another major piece in the royal role (the previous version was Standard-chess-based variants with a royal queen or a royal amazon).
Furthermore, I have added the Appendix on my personal Chess Dot Com Custom Variants at the end of this web page.
§14 Historical and geographical variants of European chess, from the Middle Ages to the early modern period and possibly beyond [new, 3 April 2026]
(This is not really my own invention. I invent nothing: I merely express a wish.)
It would be very nice if some platform such as PyChess allowed one to play medieval European chess, and Renaissance European chess, etc., according to the many rulesets that are attested. (Read the extraordinary book by H.J.R. Murray, 1913!)
On PyChess, and also on Chess Dot Com, one can play chaturanga/shatranj, and this is extremely interesting. But why not also play European chess of the 14th or early and mid 15th century -- still very similar to chaturanga/shatranj! --, or Italian chess of the 16th century according to the rules of Florence or Rome or Naples, or the French and English chess of Philidor (18th century), and so on?
One day, perhaps, it will be possible.
§15 Variations on castling [new, 3 April 2026]
(For those who have no time to waste! Go towards the end of this section §15: "In brief. The idea is this", and then "But enough talk".)
Variations on castling (introduction)
Western chess is rather castling-centric. Castling is one of its defining features, together with the presence of the very powerful queen. In modern standard chess, and in 960, we have only one kingside castling and one queenside castling. But the history of castling is complex and multifarious, and goes back to the time of medieval chess, when the queen was not yet the queen (the formidable and extraordinary queen we know so well). Our distant ancestors, in the early centuries of European chess, introduced the two-square pawn advance, and also the king's leap and the queen's leap (it's about the medieval queen, which was a ferz). Castling, as we still have it today, is a transformation and a crystallized survival of the old king's leap. Before settling into the modern forms we know, the king's leap and castling had many other possible forms, which varied according to place and period. (Again, read the extraordinary book by H.J.R. Murray, 1913: A History of Chess.)
As we were saying: one single type of kingside castling, and one single type of queenside castling. Even in the vast majority of modern heterodox chess variants (I mean variants based on standard chess), including those with larger boards, the situation is the same: on each side, only one type of castling is possible, with fixed and predetermined moves. The same situation also extends to many heterodox variants which perhaps would not need castling, or would not need it in the same way: for example, variants in which the royal piece is not a king, such as Knightmate, and variants in which each player has more than one (semi-)royal piece, such as Weak Dual Chess.
On the other hand, within the field of heterodox chess, and especially in the case of scientific variants supported and practised even by top players, the idea of abolishing or restricting castling has recently been explored: see the study by Kramnik and others, where among the variants considered there are no-castling chess and a form of delayed-castling chess -- in the latter, castling is only allowed from move 11 onwards (link here, and full paper here). Among the variants playable on Chess Dot Com, there are No Castling, and Opposite Castling chess -- where same-side castling is forbidden.
I find it genuinely interesting to vary the rules of castling, in these or in other ways. In particular, I have in mind two types of variation. Both derive from my reading of Murray 1913 and other works relevant to the history of chess, including some primary sources, such as the books of the Modenese masters (Lolli, Del Rio, Ponziani: Italy, 18th century -- Wikipedia link here).
Variations on castling (part one): Italian-style free castling
Between the French and English chess of Philidor and that of the Modenese masters of the same period, there were some important differences in the rules. The most striking is this: in Italy, free castling was still practised, in a very specific form defined and described in the books of Lolli, Del Rio, and Ponziani: Italian-style free castling (arrocco libero all'italiana). Most other countries, by contrast, knew only the then-called arroccamento alla calabrista, that is, the two fixed forms of modern standard castling (one kingside and the other queenside).
The Modenese masters were aware of this difference and proud of it, and defended the richness of the Italian system. The Italian system continued to be practised in Italy well into the 19th century, and it had important supporters even outside Italy: in particular, the authoritative English chess player and teacher J.H. Sarratt (1772-1819 -- see Murray, pp. 874 ff.; Wikipedia link here) "was ready to adopt free castling and all the Italian rules" (Murray, p. 881).
The Modenese masters codify the Italian rules of castling in a very precise way. Among other things, they state that in castling the king and one of the rooks move in a single move, but it is not permitted to move a pawn at the same time: they felt the need to specify this, because this practice was evidently still in use (it had been standard in earlier periods, but it seems they were not aware of that).
Even in this Italian system, castling had constraints similar to those we know: you could not castle if one of the castling pieces had already been moved, etc. There were also some additional constraints: the Modenese masters state that it is forbidden to castle if castling immediately gives check, or if it produces an attack against any opposing piece or pawn. For these details, one must consult Murray and, above all, the primary sources in Italian; and it will be important to know them precisely if, one day, digital versions of 18th-century Italian chess (and other local and historical variants of European chess) are to be developed, in accordance with the idea expressed in the previous section (see §14 above). But this is a different section!
Here we limit ourselves to considering Italian-style free castling as regards the exact movements of king and rooks. How many and which positions are possible after free castling? As far as I have seen, the correct answer is not readily available anywhere. Searching on Google leads one astray. It is possible that some modern book gives the exact answer, but I would not bet on it.
[A short intermediate postscript, with a spoiler about the numerical result. I have now discovered that, by searching on Google for something like arrocco libero Italia sedici, one can find excellent confirmations. The resulting positions are indeed sixteen, veramente sedici, exactly as my count shows. See: www.mantovascacchi.it, centurini.altervista.org, and, in English, new.uschess.org... However, apparently no webpage reports the calculation procedure or lists all the positions. There are many inaccuracies. One of the pages gives 5+11 instead of the correct 6+10. Total 16, in any case. -- Ah! This 2012 forum discussion had already covered the topic thoroughly: www.chess.com/forum/view/general/bring-back-free-castling]
The scheme I provide below is, so to speak, original research: I have constructed the schematic illustration myself, on the basis of what can be read in the Italian books of the 18th century (fortunately available on sites such as Google Books). It must be emphasized that, whereas in kingside castling both castling pieces necessarily remain on the kingside, in queenside castling the rook may move to the king's original square, thus invading the opposite side.
Here are all the final configurations allowed by Italian-style free castling (18th and 19th centuries).
Kingside: 6 possible configurations
o o o o R K - -
o o o o - R K - (alla calabrista)
o o o o R - K -
o o o o R - - K
o o o o - R - K
o o o o - - R K
Queenside: 10 possible configurations
- - - K R o o o
- - K R - o o o (alla calabrista)
- - K - R o o o
- K R - - o o o
- K - R - o o o
- K - - R o o o
K R - - - o o o
K - R - - o o o
K - - R - o o o
K - - - R o o o
In brief. The idea is this: just as various platforms allow us to play online no-castling chess or opposite-castling chess, I would like it to be possible to choose to play standard chess with Italian-style free castling as well: exactly the standard chess of today (and 960 as well), with the sole difference being the movements of king and rook in castling. See the schemes above. 6 + 10 = 16 possible configurations.
(The same idea, with a greater number of resulting positions, can also be applied to variants with larger boards, such as Gothic Chess, 10x8, or Begnis's Reformed Courier Chess, 12x8!)
Variations on castling (part two): allowing castling out of check and/or through check
Castling, in itself, is strange, illogical, not very reasonable. As Kramnik says, and as anyone can observe, it is the only illogical part of the rules of chess.
There is also a certain illogicality in some of the constraints concerning castling. These constraints have changed greatly over time and space. For example, today it seems obvious to us that one may castle after having previously been in check, provided one is no longer in check and the king has not moved. But in some rulesets of the past this was not allowed. If you played chess in 16th-century Europe, you had to adapt to the rules of the country and city where you happened to be.
But enough talk! What I am proposing here is simply this. Digital platforms for heterodox chess could also allow castling out of check and/or through check:
a. Standard rules apply;
b. Castling is allowed while in check, but the king may not pass through attacked squares;
c. Castling is not allowed while in check, but the king may pass through attacked squares;
d. Castling is allowed while in check, and the king may even pass through attacked squares.
(In standard chess, of course, the king crosses a single square when castling, on either side. But the number of squares crossed may be zero or more than one, if for example we play 960, or variants with larger boards… or if we activate Italian-style free castling -- see above!)
Endnotes
Game balance
I think that some subvariant combinations, at least in section §6 above, may perhaps give rise to balance problems. Some of my default choices are possibly questionable. In any case, standard chess and a lot of other important games are unbalanced too. Inverted-colours rematch can be a good and simple solution, as is well known for Fischer Random starting positions, etc.
Xiangqi or janggi cannons?
A general note on the variants with cannons: we can choose either the Chinese-rule cannon or the Korean-rule one. They are both nice and interesting. (The choice must be stated in advance, of course, as a part of the ruleset. Having several types of cannons in one individual variant may also be interesting, as the fans of cannon shogi will teach us.)
KOTH and campmate subvariants
I like King of the Hill, and Couch Tomato's campmate rule as well. Either hillmate or campmate can be added to plenty of chess variants, including the ones listed in this web page.
Si parva licet componere magnis: other famous variants, put together with the ones of my list
A lot of existing and more or less important variants could combine with most variants listed in this web page: Chess 960, Duck Chess (!), Three-Check, Crazyhouse and Bughouse (!); Seirawan Chess and Illescas Dragon Chess (perhaps with some adjustments, see also section §1); Torpedo, Sideways Pawns, Capture Anything, etc.
Appendix on my personal Chess Dot Com Custom Variants [new, 25 March 2026]
Whenever I start a game in one of my custom variants on Chess Dot Com, I usually open the chat and paste a brief description of the variant.
Chess Dot Com sergeants in the place of P's, see §8 above
SERGEANTS instead of PAWNS. (They capture & move, both forward & diagonally forward. Standard-like 2-square advance in both kinds of peaceful movement; and en passant. STANDARD PROMOTIONS.) The rest is CHESS.
Variations on Perfect Chess (without knights or without bishops -- same text), see §10 above
NORMAL chess (including CASTLING), except for: each player has a R+N, a B+N, and even a Q+N! (Same change for promotions)
[ It's a variation on Perfect Chess: greenchess.net/rules.php?v=perfect ]
Another variation on Perfect Chess (two bishops but only one knight), see §10 above
NORMAL chess (including CASTLING, only K-side), except for: each player has a R+N, a B+N, and even a Q+N! (Same change for promotions)
[ It's a variation on Perfect Chess: greenchess.net/rules.php?v=perfect ]
Non-royal N+K's in the place of N's, see §10 above
NORMAL chess, but nonroyal N+K instead of N (same change for promotions). Good luck! (The value of a N+K is about 7. A queen is 9, a rook is 5.)
Royal N+K in the place of K, see §4 above
NORMAL chess (including CASTLING), except for: ROYAL PIECE -- it's a KING-KNIGHT compound.
Major pieces in the place of K, see §13 above
This is: Royal *Queen*, WITH (nonroyal) AMAZONS AS PAWN PROMOTIONS. -- NORMAL chess, except for TWO points: (1) The ROYAL PIECE is a QUEEN; (2) You can PROMOTE to a Q+N too. **CASTLING: DRAG PIECE ONTO ROOK**
[new, March 2026]
This is: *ROYAL ARCHBISHOP*, WITH AMAZONS AS PAWN PROMOTIONS. -- NORMAL chess (including CASTLING), except for: (1) ROYAL PIECE is B+N; (2) another B+N in the place of Q; (3) you can PROMOTE to a Q+N too.
[new, March 2026]
This is: *ROYAL ROOK*, WITH AMAZONS AS PAWN PROMOTIONS. -- NORMAL chess, except for: (1) ROYAL PIECE is ROOK; (2) R+N (chancellor) in the place of Q; (3) you can PROMOTE to a Q+N too. **CASTLING: DRAG ROYAL PIECE ONTO ROOK**