Talk:Chu shogi

Tables
While this article is outstanding, its old-style HTML tables are completely unreadable, to the point of being a barrier to further editing. Let this be a lesson on the judicious use of whitespace. --Ardonik.talk* July 3, 2005 10:48 (UTC)


 * Hi. Thanks for your interest.  I am not much of a writer, but I felt tables were the best way to demonstrate the piece movements.  I didn't want to make up a bunch of graphics.  Is there a more appropriate way of doing it?  --Sivak. July 4, 2005 2:01 (UTC)


 * The tables seem fine to me.


 * Notice you're missing a few pronunciations. Are you just missing vowel length? The only possibility for 盲虎 "moko" in standard Japanese is mouko (from mauko); for 猛豹 "mohyo" is mouhyou (from mauheu), etc. However, others are wrong: 将 can only be shou in any of its compounds, I believe. (Well, there's always the possibility of ideosyncratic pronunciations, but that's doubtful.)


 * Well, a lot of these Romaji I am getting from Jim Breen's site. I have a hard time reading Hiragana, so I only did a few for the initial release.  I intended to add the others later.  I don't speak Japanese and only know Katakana a few Kanji (mainly for fun).  It was hard to get some of them for this article.  If you want to contribute, feel free. --Sivak 21:06, 13 July 2005 (UTC)


 * Of course, this is assuming that Roger Hare's got the basics right, because there are other possibilities if you let the vowels vary (*myouku for 盲虎, etc.). Easiest just to lift the pronunciations from the Japanese Wikipedia.


 * Anyway, I'll add these in. kwami 20:02, 13 July 2005 (UTC)


 * Weird - Japanese Wikipedia got one wrong, or so it appears. 歩兵 has two pronunciations: hohei (infantry, foot soldier), and fuhyou ([shogi] pawn). The abbreviation in shogi is fu, but Japanese Wikipedia has hohei. I'll "correct" the Japanese page and see what happens - meanwhile I'll leave it as fuhyou here. kwami 20:59, 13 July 2005 (UTC)


 * Well, it's been seven weeks, and my correction to the Japanese site stands, so I'm taking that as confirmation. kwami 07:30, 2005 September 4 (UTC)


 * Sorry about being anonymous above [correcting now]; I got signed out while doing this. I touched up the article a bit, hopefully for clarity; see what you think. Tenjiku would be a fun article to do too (my favorite variant that I haven't actually played). Last I heard, some of the moves were still debatable in the English-speaking world; it would be fun to check out Japanese Wikipedia. Well, maybe some day. kwami 00:39, 2005 July 14 (UTC)


 * Hey, I had a quick question about the game and the Taishi (Crown Prince). If you have one of these and it gets put in check, do you have to move it out of check?  Same for the king:  If you've got the prince and your king is checked, do you need move the king away?


 * What if your opponent checkmates your king, but not the prince? Can he capture the king on the next move and then have to go after the prince?  I've been wondering about this matter and if either piece can be physically removed from the board. --Sivak 05:35, 4 August 2005 (UTC)


 * As far as I know, there is no rule against capturing kings in shogi. A player will normally concede the game when mated, but if he doesn't, his opponent takes the king like any other piece, and the game is over. In chu, however, the game is on as long as each player has a king, and the prince counts as a king. (Perhaps the prince is assumed to be promoted to king if the king is captured, but I don't believe anyone says that explicitly.)


 * There's that famous (if perhaps apocryphal) story of the (Western) chess master who told a beginning player that he could queen a pawn into any piece he chose, so of course the student chose to have a second king. The master claims he then won the game by checkmating both kings simultaneously. No such difficulties are entailed by a second king in chu: you simply capture the kings one at a time. And no, I don't believe you have to move a king that's in check, though of course it would be suicide not to do so if you have only one king. But with a prince, you might even sacrifice your king as part of a gambit! kwami 06:08, 2005 August 4 (UTC)


 * Interesting. Maybe we should add notes about this in the article.  I actually realized there is no mention of the Crown Prince acting as a second king.  Maybe you could put something together?  Lastly, if BOTH pieces are checkmated, does the game end?  I know that a resignation is more likely, but it's these little points that interest me.


 * I hope you like what I've done. I've tried to do my best at both fact-finding and editing.  I don't mind the revert, I was just following what I found in Manual of Style (Japan-related articles).  --JTTyler 06:44, 4 September 2005 (UTC)

I assume you mean my reverting the vowels with macrons? I'm not sure what style has to do with it: the macron indicates that the vowel is long in Japanese, and these weren't long vowels. I'll make the corrections in the other variant pages too.

I doubt that double check mate ends the game. Only capturing the last king does. Since you can only capture one piece at a time (unless you're mating with the lion or similar piece), the other escapes, and the game goes on. But even if you checkmated both kings with the lion, so that you could capture both on the next move, it's still the capture that ends the game. kwami 07:27, 2005 September 4 (UTC)


 * Ah, tenjiku: I don't know if it's as fun as it looks, but it's cool to be able to take out eight pieces at once! kwami 07:40, 2005 September 4 (UTC)

I went and converted the tables from XHTML to the pipeline wiki markup. This aught to make Ardonik and Sivak happier. @u@ Not only does this make things a little easier for the editors, but it also cuts 6KB from the file size. Don’t worry Kwami, all your changes are still there in the new format so you didn’t waste your time. I thought it was a good idea anyway. Maybe we could take a vote. If folks don’t like it, we could always go back to the XHTML. I always did like Sivak’s idea for the tables, and wouldn’t mind seeing them in other shogi variant articles but that may prove time intensive. Let me finish fixing the others first, then time will tell. --JTTyler 18:24, 23 September 2005 (UTC)


 * Sorry, I didn't see that comment until just now. (I got logged out when editing, which of course removed this page from my watchlist - what a pain!) I appreciate the change - I prefer pipeline markup too, but the conversion is rather tedious. kwami 09:23, 2 October 2005 (UTC)


 * I cheated --JTTyler 01:23, 6 November 2005 (UTC)

movement color
Hi Akamemoto,

I appreciate you color coding the movement diagrams here and in the shogi article. It looks good, and makes the diagrams much more legible. However, I see you've now gone from a system of giving each type of movement its own color to using different colors for each piece. May I ask why? As it is now, there is no continuity across the game, and several of the diagrams have colors that are rather difficult to distinguish, meaning that they're little better legibility-wise than when they had no color at all. kwami 23:41, 24 March 2006 (UTC)

Kirin to Qilin or maybe Kylin
Subject says it all. I was wondering if the users here thought it would be a good idea to change it to the "English version" of the name. I'm really not sure why I used the Romaji when I began the article. What do you think? I think Qilin might be the best option, as it seems to be the "proper spelling". I personally don't care on the matter, but was wondering if others did or if we should just leave it as Kirin. --Sivak 22:10, 8 April 2006 (UTC)
 * There is no common English version: Qilin is Chinese and Kirin is Japanese. It's normally seen as Kirin in English. Kylin is a fully anglicized word, but hasn't been used much since the nineteenth century. kwami 01:57, 9 April 2006 (UTC)
 * So I guess I was right to guess Kirin then? How interesting...  --Sivak 02:21, 9 April 2006 (UTC)
 * I would actually go with Kyrin, since most websites that allow playing Chu Shogi (Richard's PBM Server, Chess Variants, etc.), as well as George Hodges' Middle Shogi Manual all use "Ky" for the piece in moves lists (Ky-7j) and Kyrin for the name of the piece.TKR101010 (talk) 06:41, 5 January 2011 (UTC)

Looks like we're back to this. I see both "kirin" and "kylin" in the article. Which should we use? OneWeirdDude (talk) 17:35, 19 February 2019 (UTC)
 * It doesn't matter much to me; I'd go with "kirin" as that's what we seem to have more of. Double sharp (talk) 23:34, 19 February 2019 (UTC)
 * I've changed them all to "kirin", as that is what our larger shogi variant articles had. Double sharp (talk) 14:16, 1 April 2019 (UTC)

Heisei shogi
I don't see any mention of Heisei shogi at wiki-ja. Is this at all notable? Is the setup correct? What is an "adjacent square"? Adjacent to a friendly piece? kwami (talk) 07:05, 17 October 2008 (UTC)

Piece images on initial setup
These are very nice. Can you show the promoted pieces too?

Hidden Protectors
I've never heard of this, and so would at least like a citation, and the rule needs to actually be stated, rather than just an example given:
 * Note that there exist "hidden protector" positions. For example, if a Black Lion lies directly on a line between a White Lion and a White Free King, then the White Lion is considered to be protected by the Free King, as it will be bearing down directly on the Black Lion were it allowed to capture the White Lion (it is not so allowed, owing to this rule).

Which are these positions? They are not defined. Is the rule saying that you're not allowed to swap lions? I've never heard of such a thing. — kwami (talk) 22:18, 15 May 2010 (UTC)

A Hidden Protector position involves the Lion capture rule restricting the capture of a Lion that's two squares away and protected ... "If an opposing Lion is two squares away from your Lion, then your Lion can only capture that opposing Lion if that opposing Lion is unprotected.", #7a in the rules of the European Chu Shogi tournament (Reference #1 on the Chu Shogi wiki page), which continues ... "Note: If your attacking Lion is positioned directly between the opposing Lion, and another opposing piece, such that if your Lion were to be removed from the board, then the opposing Lion would be protected, then the opposing Lion is considered to be protected, and so may not be captured under the conditions of this rule (the "Hidden Protector")." Although worded slightly differently, this information is also stated in The Middle Shogi Manual by George Hodges.



So in the diagram above, the Lions on 10j and 11h are not able to capture each other. Both are two squares away from each other, and they are both protected (10j by a Copper General, and 11h by a Free King). On the other side of the board though, the Lion on 1g would be able to capture the Lion on 3g as it is unprotected, BUT the Lion on 3g would not be able to capture the Lion on 1g because it is protected by a Hidden Protector (i.e. the Freek King on 12g) according to the Note quoted above. The Lion on 1g doesn't seem to be protected as one normally thinks of it, but since the Lion on 3g is the only piece in between it and the Free King on 12g, it is still being protected by the Free King. This isn't dependent on the protecting piece being a Free King, it could be Rook, Bishop, etc.

Personally, I think the way the rule on the wiki page should be restated so that it's as it is on the referenced cite.TKR101010 (talk) 06:41, 5 January 2011 (UTC)
 * If you ask me, the Hidden Protector rule is redundant. All the rule-set needs to say is that a lion can capture another lion that is two spaces away if the capturing lion will be safe from capture for at least one turn.  (Otherwise, it must capture something else as well, other than a pawn or go-between—or else the move is illegal.) — Right?  OneWeirdDude (talk) 21:50, 17 March 2011 (UTC)
 * Someone agreed with me in an edit summary. Should I change it?  Thing is, I certainly didn't make it up.  OneWeirdDude (talk) 22:19, 31 March 2014 (UTC)
 * That was me! I think it should certainly be changed, for clarity, though I wouldn't mind keeping as secondary the "hidden protector" phrasing (only because that one was widely used by Western sources). Double sharp (talk) 02:10, 1 April 2014 (UTC)
 * How about: "2. It can always capture a non-adjacent lion (on a "☆" square) if, supposing it had done so without capturing anything else, remains safe for at least one turn. 3. If capturing a non-adjacent lion with nothing else leaves it vulnerable next turn, it may only capture the other lion if it first captures another piece so that it is then adjacent (double capture), and then only if the other piece is something other than a pawn or go-between."  (The wording might need fixing.)  OneWeirdDude (talk) 01:14, 11 April 2014 (UTC)
 * Seems good to me. Double sharp (talk) 11:55, 11 April 2014 (UTC)
 * OneWeirdDude and I have now edited the table entry for the Lion. The text now seems to make the note on protection below redundant, so I removed it, along with the diagram (which unfortunately wouldn't fit in the table: otherwise I would have kept it). I still think the diagram is a helpful illustration for the situation, though (and similarly there could be one showing the "hidden protector" pinned to a royal piece): @OneWeirdDude, what would be a good place to put it, in your opinion? Double sharp (talk) 11:40, 22 April 2014 (UTC)
 * I am with you on this, but beware that not everyone agrees on the cases where a hidden protector was blocked by a P or GB that is captured together with the Lion, or where a 'false protector' P or GB is captured together with the Lion. Modern as well as historic rule descriptions do not mention these cases explicitly, and it depends on whether one interprets the explicitly mentioned case of the protector being blocked by the capturing Lion as a (strictly speaking redundant) example that was added to illustrate that one should judge the situation after the capture rather than before it, or as an exhaustive summary of what is allowed. H.G.Muller (talk) 14:17, 22 June 2014 (UTC)
 * I restored the definition of a hidden protector, and incorporated some of your text (is that fine with you?) into the article to explain this particular point. Is it better now? Double sharp (talk) 14:58, 22 June 2014 (UTC)
 * Well, I am not a native English speaker, so I am not sure if "captured with the Lion" is sufficiently unambiguous that it cannot be mistaken for "captured by the Lion", which is why in the text above I added the word "together". If you think it is sufficiently clear, I trust your judgement. H.G.Muller (talk) 19:10, 22 June 2014 (UTC)
 * I added the word "together". (I actually missed it when reading your comment, and thus inadvertently removed it: it's in the article now.) Double sharp (talk) 03:34, 23 June 2014 (UTC)

@Double sharp:  My opinion? I'm not sure my opinion counts for much; I'm not that good at composing documents. OneWeirdDude (talk) 03:12, 24 April 2014 (UTC)

Here's another example I found on the japanese Chu Shogi wiki page. Here the red Bishop is a hidden protector of the red Lion. The black Lion would not be able to capture the red Lion in this case. TKR101010 (talk) 06:43, 18 January 2011 (UTC)
 * Thanks. I'll add this to the article. — kwami (talk) 07:15, 18 January 2011 (UTC)
 * Is there a way to do this so that the opponent's pieces are upside-down and not red? Red makes them look like promoted pieces. OneWeirdDude (talk) 03:39, 16 March 2011 (UTC)
 * With graphics, not with text. — kwami (talk) 04:29, 16 March 2011 (UTC)
 * Hey, I'm fine with that. I tried to upload a bitmap and failed.  Would you do it, please?  OneWeirdDude (talk) 21:44, 17 March 2011 (UTC)

If I were trying to create a historical standard for the rules, I would probably interpolate this to mean that you should judge the position after the capture. A lion can't capture a lion if it can then be recaptured. That would be a simple way of phrasing it. It's a pity though that we cannot explicitly simplify things to that due to a few small issues.

Although if I were trying to do something with this game, I would note: this is surely the best game of Grand Chess ever designed. It could work as an expansion of chess, too: the worlds of chess and shogi could be said to meet, as both contribute pieces, and we have a chesslike material balance. If I were trying to use it as a chess expansion primarily, though, I would probably use FIDE pawns to make the pawn-line more lion-proof (so that it need not be perfectly flat), and thus also FIDE go-betweens (that can also move and capture backward). I might also get rid of the lion-trading rules, as there are still many strong sliders, just to prove that (like in chess) the loss of the strongest piece still leaves a mightily interesting game. (The same may be applied to dai or tenjiku! With tenjiku I'd use go-betweens as well, converted to FIDE-style two-way pawns. Dai-dai, maka-dai-dai, and tai should probably work, but I need to check this.) Double sharp (talk) 10:38, 4 February 2016 (UTC)
 * My Chess variant 'Elven Chess' is a step in that direction.H.G.Muller (talk) 19:37, 18 September 2016 (UTC)
 * In fact, it seems like the treatment of the fire demon in tenjiku is a good point that shows that we don't need lion-preserving rules.
 * I have went through dai and tenjiku shogi, to edit their rules to be more in conformance to that of chu shogi, which was probably the case. Nonetheless, we cannot be certain, so I added a few caveats before the main meat of the move descriptions begin.
 * I would add that I would use Western knights in dai and tenjiku shogi, so that they aren't jokes. The shogi knight is a joke. The only reason it works in standard shogi IMHO is because of the drop rule, which removes its binding and lets it go anywhere. I would also eschew dai and let tenjiku take its place in the shogi-variant hierarchy, as having all bishops on the same colour (and all short bishops "flying dragons" on the other colour) is annoying. I would also get rid of the rules disallowing repetition, as I think that the possibility for a losing side to perform a miracle save by perpetual check is charming and beautiful. Maybe I am biased because I am primarily a chess player, but these seem to be reasonable minor changes to make these games just a bit more attractive to Western chess players. The main point of divergence is how the Japanese inventors took the concept of promotion to such lofty heights. Double sharp (talk) 13:45, 4 February 2016 (UTC)

Promotion
When dicussing the rules with Japanese Chu-Shogi players on the 81Dojo server I learned that they use different promotion rules there. I investigated the matter, and it seems that the rule mentioned in the Wikipedia article is based exclusively on a misunderstanding. For one, the rule mentioned in the article is different from what the Middle Shogi Manual states: it says the possibility to promote on a non-capture returns after one turn, while the MSM states that it only returns after a move with that piece. Both seem to be wrong, based on faulty translation of the Japanese rule descriptions, as well as by drawing conclusions from erroneous solutions to historic mating problems. The correct solutions provide zero evidence that promotability on non-capture could ever be regained without leaving the zone first. A summary of my findings is currently posted at http://hgm.nubati.net/MSM/ChuPromoRule.html. H.G.Muller (talk) 14:48, 22 June 2014 (UTC)

Also note that the Japanese Chu Shogi Association explicitly states in their rules ( http://www.chushogi-renmei.com/kouza/kouza_main.htm section 4, sub 4) that lances that reach last rank become immobile pieces: "If the lance is not advanced in formation to the end-stage and 4. Infantry, is treated as a piece (piece of not going) to death on the spot." (Sorry for the Google translation, but native Japanese speakers confirmed this.) H.G.Muller (talk) 18:46, 22 June 2014 (UTC)
 * OK, I correccted the article. Now it says that a piece which declined promotion at its first opportunity cannot promote unless (1) it makes a capture or (2) it leaves the zone and reenters, unless it is a pawn moving to the twelfth rank. It also now says that lances that reach the last rank without promoting simply become immobile. (Sorry for reverting you earlier: I didn't see your comment here, and chose to trust the English-language sources based on not knowing of the evidence you have clearly pointed out on your website that they are wrong.) Double sharp (talk) 13:04, 29 June 2014 (UTC)
 * BTW, do you have any info for the other historical shogi variants with promotion zones? Do the same promotion rules apply? Double sharp (talk) 13:06, 29 June 2014 (UTC)

Repetition
Similarly, the repetition rule mentioned in the article (which is the same as in the MSM) can also be questioned. It would allow one to force a win by perpetual checking, when the checked side would have repeat first. This is something that Chess players in general would perceive as very unsatisfactory, and both modern Shogi and Xiangqi have rules against it. Some of the historic matings problems do offer an opportunity for immediately forcing a perpetual which then turns into a mate when the only safe evasion would have to be repeated, but the given solutions never make use of this. So it is unlikely that even in the Edo period the repetition rule was as strict as mentioned here. A minimal amendment would be that escaping from check would always be allowed, repetition or not. H.G.Muller (talk) 14:53, 22 June 2014 (UTC)
 * That would be interesting, though. As Ralph Betza said for his Nemoroth, which uses the same rule (you cannot repeat a previous position): "Nimzovich would approve of this proposition. Victory by Zugzwang!" Of course, given the evidence from the historical mating problems, your amendment would make sense. I'd codify it as: "It is illegal to make a move that results in a repetition of a previous position, unless you are in check. A position is repeated when all pieces are in the same position and the same player is to move, so that turn passing is legal if you have a lion, horned falcon, or soaring eagle." Note that Nemoroth does not have royal pieces, which makes this (as you put it) unsatisfactory situation impossible. Double sharp (talk) 08:23, 13 February 2016 (UTC)

I am not happy with the modifications recently made to the repetitions section. These suggests that the rule we mention primarily is not based on the TSA rules, while in fact it is. There was no 'rule change' at all in 1976, it has been George Hodges' interpretation of the historic sources all the way. Which could be wrong, BTW: I am pretty much convinced now that the historic documents were formulating this in terms of 'move sequences', not positions. And this usually puts the opposite player at fault, not the one who creates the repeated position, but the one that has to play from it. The Chu-Shogi-renmei rules for turn passing still reflect this: an even number of turn passes (i.e. returning to the same position) are allowed, as the player that passed first must eventually make a move. Whether the first move or position repetition is forbidden, or only the Nth is just a minor difference, a relaxed application of the basic rule to allow for human error. H.G.Muller (talk) 14:48, 9 August 2017 (UTC)
 * It seems to be a generally applicable principle that if a rule in a historical shogi variant has multiple interpretations, George Hodges' is probably the wrong one. May I ask what passages in the historical documents are the ones that point clearly to your described interpretation, though? Double sharp (talk) 00:50, 10 August 2017 (UTC)
 * Well, George Hodges is not entirely to blame for this; probably his Japanese coworkers were unconsciously influenced by the rules of modern Shogi, assuming by default that they would also apply to the Edo games. I have no access to primary sources, so my main source of information is the website of the Chu Shogi renmei, http://www.chushogi-renmei.com/kouza/kouza_main.htm . The third section of links on that page leads to excerpts of two historic rule descriptions, a recent one (1928) from Fumiaki Okazaki, and an Edo source 1703. Hidetchi once translated the latter for us on the now defunct 81squareuniverse.com forum. (Perhaps a mirror can still be found in some web archive?) The Okazaki rules seem to mention repetition rules in point (4): 千日手は仕掛けた方より変えなければならない. and （注）待ったの禁と、将棋の昔の千日手の規定と同じ.  . The ancient Japanese of the Edo document is too difficult for Google translate to make anything intelligible out of it. H.G.Muller (talk) 08:04, 11 August 2017 (UTC)
 * Well, from what I understand, the Edo-era sources are written in very Sinicised Japanese, so I may yet be able to make something out of them! ^_^ I'll take a look at this in a few hours. Double sharp (talk) 10:24, 11 August 2017 (UTC)
 * It seems clear from Okazaki's description, at least from my understanding, that for him it was the player who starts the repetitions who must deviate. The second sentence simply means that the rule for chu shogi is the same as an old one for shogi (so evidently not the modern one). This accords with the historical rule described by John Fairbairn in Shogi for beginners (1986, 2nd ed.) and attributes to the 2nd Meijin Sōko Ōhashi, and mentioned in our article on sennichite. Unless there is some old Sino-Japanese term for repetition that I do not know of, I do not see anything about repetition in the Edo source.
 * I am not sure if Okazaki's description is representative of the historical rules in the heyday of chu shogi. However, I would grant him a great deal of weight, as he would come near the end of that unbroken tradition until the 1950s. Curiously he seems to imply that the same thing had previously happened to the other variants: when describing the large shogi variant exhibition of about three years before he wrote the article, he described the situation as the exact rules and even movements of the pieces having been lost or forgotten, and said that chu shogi seemed to be going the same way. His efforts notwithstanding, I fear it pretty much has by now: I was in Kyoto about three and a half years ago, and it seems that for the most part only older people seem to even remember that chu shogi ever existed, and even then about nobody seems to play it. So I would take Okazaki's description as the last one coming from the time when the game was truly "alive". If it can be proven that these were different from earlier versions of the game, as in the Okazaki rule (counterstrike permitted against a hanging lion), we can always mention both. Double sharp (talk) 16:16, 11 August 2017 (UTC)
 * Indeed, he who starts to repeat needs to deviate. But repeats what, moves or positions? What I feel uneasy about is this: an anti-repetition rule is expected to prevent repetitions by forcing the player that benefits from them to deviate. Not to encourage them by awarding a quick win to a player that can force one. But it is very hard to formalize what forcing is, as the complex rules for perpetual chasing in Xiangqi demonstrate. Usually it is the forcing player that first reaches a position in the repetition loop, so it seems to make sense to forbid him repeating that position. But this method fails, because after the forcing player deviates by playing an unrelated forcing move, (e.g. initiating a trade). That does not make the repetition loop go away, but because of the trade it now consists entirely of fresh positions, so that the forcing player starts moving already from a position on the repeat loop. This would then put the burden of deviating on his opponent after all. So it seems rules solely based on who repeats a position are doomed to fail. The moves that follow the repeat loop are the same before and after the trade, however. E.g. a perpetual check would go like (c=check, e=evade, ~ = reversed move): c1, {position A} e, c2, e~, {c2~ would repeat A} capture, recapture {position B} c2~ {allowed again} e, c2, {e~ repeats B, so checkmated}. The problem would go away if the move c2~ would remain forbidden even though the intervening trade would have changed the position (in a non-essential way). When looking at moves, however, it becomes a question when two moves should be considered the same. E.g. would it matter where the piece came from, or just where it goes to. In the example c1 and c2~ result in the same position A, although in c1 it came from a position outside the repeat loop. If these would be considered the same, c2~ would be forbidden (after the maximum of allowed repeats), and then remain forbidden. Of course the fact that the 1703 description doesn't even mention anything on repetitions may make this whole issue a red herring. H.G.Muller (talk) 13:53, 12 August 2017 (UTC)
 * Okazaki does not state which. One might be able to resolve this by saying that the burden to deviate is on the player who initiates and chooses repetitions, not the one who is forced to go along with them to survive. This may be hard to define precisely to cover all the cases, but always allowing check evasion seems to cover the vast majority of them, and would seem consistent with the historical mating problems. Double sharp (talk) 16:19, 12 August 2017 (UTC)
 * Indeed, this is what the Xiangqi chasing rules formalize. The rule of the Chu-Shogi renmei always allows all absolutely non-aggressive moves. This avoids that dead draws like K vs K can be decided by the repetition rule, but is a bit unsatisfactory in cases where, say, a Rook chases a Queen, as it will be almost impossible to withdraw the Queen without making new, but completely irrelevant attacks with it (e.g. on protected Pawns). Allowing only moves with pieces that received a new attack would solve this, but would not allow a natural draw in K vs K. Of course one could allow both absolutely non-aggressive moves and attack evasions. But to get back on the issue of the historic rules: that the 1703 description doesn't seem to mention anything about this makes me wonder whether there indeed was any rule of this kind in Edo times. Even the idea that perpetual checking is forbidden might be contamination by the rules for modern Shogi. (Note that perpetuals are one of the few ways a game with piece drops can remain undecided, while in games without such drops it is much more likely a draw is achieved by reaching an unwinnable end-game. Perpetual checking in Xiangqi is a problem because the King must stay in the Palace and has a weak move, so that it becomes an easy target for a perpetual in almost any game at some point; in Chu Shogi it would hardly be a problem until a late end-game where the defending side has a Queen-like piece (like 2 Tigers vs Queen), and forbidding perpetuals there solves very little, as the Queen could make hundreds of checks before repeating a position, after which a single other move would create a new position where the checking would start all over.) If perpetual checking was forbidden, I would certainly expect even a short description to mention it; it would not be more rare than 'hidden protectors'. Of course it could be that other Edo manuscripts mention it. But, considering how difficult it is to formulate anything satisfactory, it seems unlikely that there could be just a casual mention. H.G.Muller (talk) 10:32, 13 August 2017 (UTC)
 * Ummm,I looked it up in the MSM, and concerning repetition George Hodges says: If a perpetually recurring sequence of moves occurs, this is known as “sennichite”. This is not allowed and the player beginning the sequence must vary his move.. So it seems that attributing the rule that repeating positions is forbidden to George Hodges is wrong, as in general the move out of the repeated position is the first repeated moves. Unless moves are considered the same when they move the same piece to the same square irrespective where it came from. As this is different from modern Shogi(as the MSM also remarks) it must be based on something, and of course the MSM doesn't state on what. It could be solely based on Okazaki's rules. H.G.Muller (talk) 19:09, 13 August 2017 (UTC)
 * Perhaps we should check the historical repetition rule in use for standard shogi, as Okazaki refers to? Double sharp (talk) 00:05, 14 August 2017 (UTC)

Just an idea: given the sparse wording of the historical sources for most of the variants – they often seem to assume a level of common sense that we lack, not having grown up with chu shogi as a living game – I strongly suspect that this is another case where treating the literal wording of the sources as the literal truth gives unsatisfactory results. For example, it seems that in the past the rule for repetition in standard shogi was also "a player who initiates a repetition loses", but this seems to assume a level of common sense in its application, or else you get the unsatisfactory scenarios that you have outlined above. So as a result we now have the proviso about positions rather than moves for clarification.

Consider: when chu shogi was in its heyday, I very much doubt that most people could play it, since it requires recognition of characters and most people were probably illiterate then. That seems to be why, unlike Go, it seems to be exclusively an aristocratic game in that time. As a result, I do not think they would have found the strict, almost legalistic codification of rules to be necessary. I wouldn't be very surprised if the rules for xiangqi on perpetual chasing and checking also date from considerably later than most of the other rules; actually, the opposite would surprise me. Even in chess, the standardisation of stalemate and pawn promotion only happened in the 19th century, and even a few wrinkles involving castling and dead positions were only sorted out in the late 20th century! Chu shogi clearly had its development arrested by sho shogi before it could attain this kind of standardisation. So it is not surprising that it had none historically. So why not let the standard version be that of the foremost organising body for it today, the Japanese Chu Shogi Association? Double sharp (talk) 11:38, 15 August 2017 (UTC)

Piece value
According to Betza, the value of a doublemove commoner (non-royal king) is thrice that of a normal commoner. But this is probably for an 8×8 board (the larger 12×12 board would make it somewhat weaker), and doesn't take into account the lion's passing and jumping abilities (which would make it stronger). It is an interesting finding, though. Double sharp (talk) 04:51, 29 June 2014 (UTC)
 * Ralph Betza was a great thinker, but his reflections piece-value were based on very limited human play-testing combined with educated guessing, and often do not bear out in practice. I did a great many determinations of leaper piece values by the method of computer self-play from materially imbalanced positions, and the opening value of the Commoner turned out to be slightly below that of a Knight. Its gets slightly more valuable than a Knight in the end-game (but you have to test it with a program that is aware of its mating potential in order to see that.) The FWADN (i.e. the 'Toothless' Lion without its passing and hit-and-run abilities) tests as about 1.5 Pawn stronger than a Queen, i.e. about 1100cP. It attacks 3 times as many squares as the Commoner, but the dependence of short-range-leaper values on 'footprint' is super-linear (approximately 30*N + 5/8*N*N for the 'multiplet averages' of all pieces with equal number of target squares N). The ability to double-capture and igui enormously increases the Lion value, though (but turn passing seems virtually worthless), and 1500 seems a more realistic value for the Lion (i.e. about 5 Commoners). Chu players might voluntarily trade a Lion in the opening phase for FK + DH, but not for anything less. I did not get yet to making accurate measurements of Chu-Shogi piece values in the context of Chu Shogi itself; until recently there were no engines that played with strategy realistic enough that I would consider the piece values they would find meaningful. So currently HaChu still uses the values I found on the website of the German Chu-Shogi Association. H.G.Muller (talk) 09:02, 29 June 2014 (UTC)
 * Yes, I think for a piece as valuable as the Lion turn passing is essentially worthless. It could help against zugzwang in the endgame, but the lion is so powerful that in practice it's probably not all that useful (besides it can triangulate anyway).
 * I think the reason why the commoner is weaker in the opening is because it is (1) slow to develop and (2) it needs to be right next to the centre to exert influence there, e.g. an Nf3 attacks two central squares (d4 and e5) but a commoner needs to be on f4 to get two central squares (e4 and e5), and f4 is inherently harder to have defended or at least not under attack than f3, so it can be forced away. (This holds true for bishops as well, but a Bg3 still impacts the centre as well as a Bf4, whereas a Kg3 has no impact on the centre at all.) But its attack, though short-range, is concentrated: so if gets to occupy the centre itself in the opening or middlegame, it may well equal a knight in value.
 * Betza's speculations make reasonable first approximations, I think, but I agree with you that to get any sense of what a piece is actually worth you need to playtest. I wonder how much of that he did for Chess with different armies? He posted games he did of FIDE vs. CC (both ways around): but from looking at the pieces I think the Rookies are the most balanced to normal chess, followed by the Clobberers (I think the archbishop isn't a weak enough queen to counter the overstrong bishop) and then the Knights (their "bishop" replacement, the charging knight, is too strong, I think, because of its concentrated moving power). Double sharp (talk) 10:57, 29 June 2014 (UTC)
 * Betza did a lot of play-testing on CWDA, for sure, (he considered it his major invention), but it was all in human games (where 100 already qualifies as 'a lot') against opponents presumably weaker than himself. Since CWDA is one of the variants I have configured the standard release of my engine Fairy-Max to play, I did a lot of computer play-testing (meaning tens of thousands of games) with it on these variants. One thing this revealed is that all the unorthodox engines are significantly stronger (50-150cP) than FIDE. Humans seem to systematically overestimate the power of the familiar pieces. (The same holds for Spartan Chess, which also has unequal armies. This is perfectly balanced in computer play, but humans perceive the Spartans as disadvantaged compered to the Persians=FIDE.) The Clobberers have an advantage of over half a Pawn on FIDE (mostly due to Betza's underestimation of the value of BN, as you say), but the Nutters on average beat FIDE even when you give them Pawn odds. I haven't tested the Rookies as extensively as Clobberers and Nutters, because it was added to Fairy-Max' repertoire only recently (due to difficulty of implementing limited-range sliders), but what I have seen suggests it is close to the Nutters in strength. The charging Knight is indeed a quite strong piece. I once did some intreresting testing on divergent pieces. It turned out that K (non-royal), N and mKcN all had approximately similar opening value, but that mNcK was about 50cP stronger. It seems that the strength (in the K-N comparison) of the Knight is speed in moving, and of the Commoner concentration of its attacks. The charging Knight also seems to combine these favorable properties. Logically this would imply that mKcN had the worst of both, and should be worth significanly less. But low value of a piece with good forking ability is often masked in opening values, because you almost always can force a trade for something else, and so its value is drawn towards the value of the opponent's minors. I noticed the same when testing the Camel. It seems a near-worthless piece, that in the end-game almost always is lost without compensation, because it is simply too clumsy with its long stride (on 8x8). But its long-range forking ability makes it easy to trade for more valuable stuff in densely populated positions, so it still has a reasonable opening value. H.G.Muller (talk) 08:50, 30 June 2014 (UTC)
 * Two out of three right, then! :-) I wonder what I missed in the Rookies. The WD is quite interesting for a minor piece, being able to force mate. Is that too strong? Because it focuses everything on one direction, and the HDW (3-square x-ray rook) seems very dangerously strong, much stronger than would be expected for a 3-component piece. The RN and Q seem about balanced. The HFD may be a bit overstrength in the opening due to its large leap, but overall it seems OK to me: the R4 seems like a overstrength minor piece or understrength major piece (maybe using R3 instead would balance the Rookies?)
 * Your results for the divergent pieces are very interesting, and confirm my suspicion that the commoner's main strength is its concentrated attack. You raise the point that the knight has speed in moving, but at the same time isn't so fast as to be clumsy like the camel: naturally this will change with the board. Maybe on an n-by-n square board, the ideal leaper jump-length would be somewhere around n/4? It seems about right for the knight on 8×8 and the camel on 11×10 (Tamerlane chess: close enough to being a square, I guess). Double sharp (talk) 10:09, 30 June 2014 (UTC)
 * If you break it down to individual pieces, the WD is worth nearly as much as a Knight. (I set it to 310 in Fairy-Max, where N=325, despite its mating potential; this can be explained by the fact it is significantly slower than a Knight, and only two of its moves go forward.) HFD is about 480, which is average for a 12-target leaper, and marginally better than a Rook that are still 'boxed in'. Betza mentions 3-jumps are dangerous, but I guess this mainly applies to forks from that distance, and only a single forward 3-jump is not more dangerous than other forward moves. The R4 is about 400, though, significantly more than a Bishop. (A B-pair is 325+325+50 = 700, while two R4 would be 800.) RN is about 50 weaker than Q, so adding everything would only leave a half-Pawn advantage for Rookies over FIDE. It could be that the Rookies derive some extra cooperative advantage because all of their pieces have mating potential. Unfortunately Fairy-Max can only handle boards of 8 ranks, due to its Pawn-evaluation code, so I never did any testing on larger NxN boards. But I don't expect there to be a pronounced optimum; what you gain in travel time from one side of the board to the other by making larger strides, you lose in manoeuvrability by skipping over squares. So this is just a trade-off. But when the stride becomes so large that the part of the board where the piece is not hindered by the edge is so small that you cannot stay in it, it really starts to hurt. For a Camel on 8x8 this is only the 2x2 center, which is way too small. So I guess a stride of N (as only move) needs at least a board of size 3N+1 to be useful. H.G.Muller (talk) 12:25, 30 June 2014 (UTC)

Article expansion
Perhaps quite a few section titles from Chess (particularly "History", "Strategy and tactics", "Composition", "Competitive play") would be relevant here, with much possible content. Double sharp (talk) 04:53, 29 June 2014 (UTC)
 * (Although it has to be noted that much of it would be historical.) Double sharp (talk) 15:11, 24 February 2016 (UTC)

OK, here's the organisation I'm going for. The opening sections (rules, handicaps, notation, and variants) can stand, but after that I hope to expand that section on strategy (some of which appear in the CVP article and some in Colin Adams' Chu Shogi Library). Then a history section badly needs inclusion. That would basically be the history of shogi, as shown in my representation below:


 * Heian → sho
 * Heian dai → dai (if not an evolution, then still the same concept) → chu, dai
 * Chu → tenjiku (a spin-off that no one ever played; just a cursory mention will do)
 * Dai → dai-dai, maka-dai-dai, tai (to quote Fairbairn: "these games were never played, nor were they meant to be played. They are regarded as idiocies of the type nowadays inspired by the Guiness Book of Records.")
 * (Wa, ko, and taikyoku fall off the main story and are not that important anyway.)

After that would be the six historical games (or perhaps these should only be mentioned?), followed by a section on composition, mentioning all those historical mating problems. I do not know if competitive play was ever much of a thing, though there was a European tournament in 1997 IIRC. Double sharp (talk) 08:59, 7 March 2016 (UTC)
 * Why aren't wa and ko important? Why weren't the "extra-large" variants meant to be played? Just curious. OneWeirdDude (talk) 22:44, 14 March 2016 (UTC)
 * The Edo-era sources didn't give very detailed descriptions for ko, IIRC. As for wa, Fairbairn says that multiple sources give descriptions, but that one has many weak pieces and one has many riders(!), so evidently neither must have been felt important enough to need standardisation. Masukawa notes that the only sets of the extra-large variants were meant for display purposes only: presumably because they look impressive but are just not practical to actually play. Perhaps another thing to note is that only in chu, dai, and tenjiku do we have some sense of consistency between the historical sources. Double sharp (talk) 06:23, 15 March 2016 (UTC)
 * BTW, perhaps I should note that the Edo-era sources often disagree for the large variants in such large ways that it is difficult to imagine them being natural, local rule variations, along the lines of giving the queen the knight's move in Western chess. I am not sure how one could get some of the rule variations listed in maka dai dai shogi otherwise. The emperor move is quite striking (jumps to any unprotected square on the board), but according to Japanese Wikipedia there are also records giving it simply the combined move of the lion and the queen. Meanwhile the teaching king causes a problem with the lion dog, because the Edo-era sources for dai dai shogi do indeed give it the Western move of Q3, without mentioning anything about lion power! Only for maka dai dai shogi might they give the lion power, assuming that Japanese Wikipedia has got it right. And then apparently there was another variation giving it the combined powers of the lion and the tenjiku-shogi vice general. Furthermore not all the "free" pieces are actually free versions of the original, and their moves seem to vary between sources again. I suspect that what we have here are mere "display" sets intended to look impressive, and multiple people felt that there needed to be rules so that they could "play" them once or twice, but that these never were actual, codified games that had a real life; the moves were simply plucked out anew from thin air each time. Double sharp (talk) 12:40, 15 August 2017 (UTC)

Bare king rule
What if your Gold is hanging and the bare King can play KxG immediately? Double sharp (talk) 15:09, 10 December 2014 (UTC)
 * The entire baring rule is in fact controversial, and the reference given to it is not a primary source. In a rule description from 1703, there is a phrase "When pieces are gone, and there are only the 2 kings, one can mate only if he has a promoted gold." (Translation made by Hidetchi from: 盤の面、駒かれになりて、玉二枚、成金一枚一方にあれば、金一枚にてもつめるなり、). It seems that the 'rule' is based on that remark, distorted through time. The whole idea that there should be anything like a baring rule is based on the reasoning that the remark would otherwise be redundant, so that it must have meant something that is not implied by the other rules, even if this requires one to assume it means something completely different from what it says. This argument is rather shaky, as the 1703 document continues by stating "When a small piece knocks down a large piece, or for example a bishop knocks down a large piece by attacking the king and lion, the player with more pieces will going to win.", which no one tried to elevate to a rule yet. It is also not clear how the "promoted Gold" in the 1703 document morphed into an "unpromoted Gold". And that mutual baring should be a draw seems to come from analogy to Shatranj only, without any roots in historic Shogi literature. Yet the Japanese Chu Shogi Association has adopted the Shatranj rule, declaring any bare King a loss, unless you can bare the opponent King on the next move.H.G.Muller (talk) 11:14, 17 April 2015 (UTC)
 * What would be the point anyway of making it an unpromoted Gold? Surely you would be able to promote it to a Rook from any sane starting position, and easily bring about the mate? Double sharp (talk) 08:37, 26 April 2015 (UTC)
 * Speaking of that rule: imagine that in chess. 1.d4 d5 2.c4 dxc4, and then Black wins because White cannot immediately restore material equality. How ridiculous! Clearly, some of these are strategic guides. Double sharp (talk) 15:13, 20 December 2015 (UTC)

Winboard piece IDs

 * A – Reverse Chariot
 * B – Bishop
 * C – Copper General
 * D – Dragon King
 * E – Drunk Elephant
 * F – Ferocious Leopard
 * G – Gold
 * H – Dragon Horse
 * I – Go-Between
 * K – King
 * L – Lance
 * M – Side Mover
 * N – Lion
 * O – Kylin
 * P – Pawn
 * Q – Queen
 * R – Rook
 * S – Silver General
 * T – Blind Tiger
 * V – Vertical Mover
 * X – Phoenix

A promoted piece is prefixed by a + sign, and hence +X symbolises a queen that was originally a phoenix. The promotions that are not initially present are:
 * +D – Soaring Eagle
 * +H – Horned Falcon
 * +V – Flying Ox
 * +M – Free Boar
 * +T – Flying Stag
 * +L – White Horse
 * +A – Whale

Hence the initial setup is as follows:

This is the largest historical shogi variant that can be played with only single letters. (Tai shogi would start looking like NetHack in such a representation, and that makes me think that the emperor should absolutely be "@".) Double sharp (talk) 16:47, 13 February 2016 (UTC)
 * ✅ Added. Double sharp (talk) 15:50, 22 February 2016 (UTC)
 * In theory Dai Shogi would still be feasible, if you are willing to make the compromise to consider the primordial King, Queen and Lion promoted pieces, and write them as +E, +X and +O. Then there are exactly 26 unpromoted piece types, but you have to make some very unnatural assignments. With Chu it was not that bad, except N for Lion. (The I, O and X are supposed to be menemonic to the move pattern of the piece.) The problem with double-letter codes (especially mixed with single-letter codes) is that they break the FEN system for position encoding. I prepared the latest WinBoard (version 4.9) for handling bigger variants by allowing the use of 'dressed letters', i.e. letter + punctuation symbol. So now L, L' and L! can be used to indicate three different pieces, where I intend to reserve ! for the strongest pieces. E.g. L=Lance, L' = Left Chariot, L! = Lion. Even in variants that do not strictly need it (such as Dai) it would allow much more natural ID assignment, giving the N back to the Knight, using L! for Lion and, say, I', S', B', C', F', O' and W for Iron, Stone, Angry Boar, Cat Sword, Flying Dragon, Violent Ox and Evil Wolf.H.G.Muller (talk) 10:32, 19 September 2016 (UTC)
 * BTW, it is nice that the WinBoard notition is mentioned in the article, but isn't it in a completely wrong section now? I would say it belongs in the section about game notation. I also think that calling it 'westernized' is wrong; the Standard Algebraic Notation used by WinBoard originated in continental Europe. It is in fact the TSA notation which uses a westernized form of Japanese board coordinates (replacing the digits written as kanji by letters). H.G.Muller (talk) 18:02, 22 September 2016 (UTC)
 * I agree; I shall move it. Double sharp (talk) 03:49, 12 August 2017 (UTC)
 * Actually it seems to have been done already, so never mind. Double sharp (talk) 07:06, 12 August 2017 (UTC)

Lion-capturing rules again
We generally speak of the situation of recapture on the next turn, as does CVP: a lion cannot capture a non-adjacent lion if that would expose it to recapture on the next turn, unless it captures something else along with the lion that is not a pawn or go-between. This actually resolves the supposedly unclear cases listed in the box: you cannot do the first one (hidden protector blocked by P or GB captured with enemy Ln), because your Ln would be vulnerable to recapture on the next turn, and you would not have taken anything more valuable than a P or GB with it along with the enemy Ln on that turn. But you can do the second one (false protector of P or GB captured with enemy Ln), because your Ln is not vulnerable to recapture on the next turn.

An obtuse interpretation of the rules would only ignore the attacking lion in determining if the defending lion is protected, and count everything else on the path – even if it gets massacred by the attacking lion on its merry way. Look at the whole point of this rule! The point is to avoid exchanging lions! (And yes, after a week I've come round to accepting its usefulness.) What is the sense in allowing you to ever play Lnx(∅/P/GB)xLn if the opponent can reply (any)xLn? If the whole point is to avoid lion trading, then the ruleset should never allow this under any circumstances, even if the P or GB along with the attacking Ln blocks the line from the hidden protector to the defending Ln. I will edit the article to reflect the more probable interpretation (because the other one generally doesn't follow from the rule's motivation), just as we now give probable interpolations for lacunae in the rules for dai and tenjiku shogi. Double sharp (talk) 15:32, 22 February 2016 (UTC)


 * P.S. I suppose the statement that these rules apply to promoted kirins as well is because most chu-shogi enthusiasts will get there by way of shogi instead of Western chess. To a shogi player, there is a real difference between a promoted silver and a promoted knight, because they will revert to their different unpromoted states when captured. But to a chessplayer, when you promote a pawn to a queen, it is a queen for all intents and purposes. It does not matter one whit that it was originally a pawn. And thus it is in chu shogi. When you promote a kirin to a lion, you know from the red text that it was originally a kirin. And such may be interesting to the casual historian of that particular skirmish, wanting to discover the histories of each of his or her troops. But its life story does not figure into it, no matter how moving its tales of promotion against the odds might be. The promotion has been obtained, there was great rejoicing, and it is now a lion and behaves exactly as one. Double sharp (talk) 15:40, 22 February 2016 (UTC)
 * I agree with all of that (and I am not even sure promoted pieces were drawn in red in historic times. But they were written with "italic" kanji. In fact I played Chu Shogi last month using a piece set someone had bought in Japan, and there the promoted pieces were black and italic as well. There is one worry, though. Of the historic tsume problems, two (D34 and D55 in the MSM) become entirely trivial when you cannot reply to Ky x Ln+ with a recapture of the +Ky. One even becomes a mate in one, while an intended beautiful line exists after recapture.H.G.Muller (talk) 18:47, 18 September 2016 (UTC)
 * I just conceived a solution to this dilemma: the rules distinguish two cases of consecutive Lion capture, on the same square (protected Lion) or on different squares (counter-strike). In the case of Ky x Ln+, x +Ky the Ln and +Ky are captured on the same square, so the counter-strike rule does not apply. But Ky x Ln+ does not violate the protection rule, as x Ln is always allowed (if no counter-strike itself), whether the Lion is protected or not. That by the time of its recapture the Ky is now a +Ky apparently is of no consequence; the net effect is a Ky-for-Ln trade, not a Ln-for-Ln or +Ky-for-Ln trade.H.G.Muller (talk) 11:26, 19 September 2016 (UTC)
 * ✅ Double sharp (talk) 15:51, 22 February 2016 (UTC)
 * But shouldn't "If capturing a non-adjacent lion and NO OTHER piece leaves it vulnerable on the next turn, ..." say instead "If capturing a non-adjacent lion and ANOTHER piece leaves it vulnerable on the next turn, "? I am abit in doubt about whether using this "leaves it vulnerable" phrase so often makes it very clear. Wouldn't it be more natural to say "can be recaptured"? H.G.Muller (talk) 18:47, 18 September 2016 (UTC)
 * Sorry, but I now see I misunderstood the sentence because it is ambiguous. I thought "no other piece leaves it invulnarable" was one sub-sentence with 'piece' as subject, while in fact it is the object of "If capturing ... no other piece". To avoid this we could rephrase as "If a Lion is left vulnerable after capturing a non-adjacent Lion and no other piece". But in fact I would move for leaving this "and no other piece" out completely. This restriction only serves to exclude the case where you do capture something together with the Lion that was blocking a protector, or was the only protector that Lion, and those cases are already explicitly mentioned as controversial in the third sub-clause. So it does not matter how the rule we state here would decide them. The same holds for the first sub-clause; the phrasing is now needlessly complicated, to the point where people trying tolearn Chu complained to me that they could not understand it. Another error is that the stuff about "additional eating" and "shooting the Lion" is now placed at a point where it seems to refer to hidden protectors, while in fact it refers to double-capture. Furthermore, the addition "which is always allowed" to the other x Lion capture in the sub-clause whose very purpose is to explain that this is not always allowed is a funny way to contradict yourself. BTW, I am starting to wonder where this idea comes from that after other x Lion you could retaliate against a Lion elsewhere with another Lion if the Ln x Ln rules allow that. Historic manuscripts will most certainly not have addressed this case explicitly, as they even fail to address the much more common cases of doubly-hidden and false protectors. And the more I think about it, the less sense it makes. If I attack one of my opponent's Lions with (say) a Rook, why should he now be allowed to counter-attack my Lion with his second Lion, so that we can trade Lions? The whole idea of the rule is to outlaw indirect trading, and it is only a moot point which piece would make the counter-strike.H.G.Muller (talk) 19:52, 20 September 2016 (UTC)
 * Blame Fumiaki Okazaki. ^_^ Since the Japanese Chu Shogi Association has apparently adopted his suggestion, we have to mention it even though it doesn't quite make sense. Then again, I think the whole Lion-capturing rule thing is a horrifically complex kludgy solution to a real problem (lions seek each other out to trade). Hmm, maybe this is why Tenjiku was invented! ^_^ Seriously, from the way you speak of the historical problems, I suspect the historical rule was just "you can't initiate a sequence of two moves that would make each player lose a lion", and cases like these would just have been argued among the players. Double sharp (talk) 14:52, 21 September 2016 (UTC)
 * It depends on what you consider 'initiating'. For a direct trade (i.e. second capture on same square) the initial capture is forbidden. But for a counter-strike the second capture is forbidden. OTOH, this situation almost always arises from the sequence Attack 1st Lion - Counter Attack 2nd Lion - Capture 1st Lion - Capture 2nd Lion, as in practice Lions will not be left under attack. It is the counter attack on the 2nd Lion here that offers the trade, and although it is not forbidden in itself, forbidding capture of the 2nd Lion makes it a losing move. Anyway, it seems that indeed this Okazaki is responsible for the idea that the rules for Ln x Ln capture have precedence over those for counter-strike. I am now convinced that, like his other rule, this was a change from what the rules historically were.
 * My other problems with the current formulation remain, however. It is needlessly complex, not fully consistently talking about forbidden or allowed, and therefore very hard to understand for a novice. I would like to change it into something like:
 * 1. A Lion cannot capture a non-adjacent other Lion when it could then be recaptured on the next move, unless it captures something substantial (i.e. other than P or GB) together with the Lion in a double capture. (The latter is called "additional eating", and recapturing that Lion is called "shooting the Lion".)
 * 2. A non-Lion cannot capture a Lion when a Lion was captured by a non-Lion on another square in the immediately preceding move.(In recent times this has been amended by the Okazaki rule, that such a counter-strike is allowed against Lion that is unprotected.)
 * Historic rule descriptions explicitly discuss a case where recapture slides over the square evacuated by the capturing Lion ('hidden protector'), but do not mention cases where a P or GB taken together with a Lion in a double capture affect the possibility to recapture, making it controversial whether the latter could be exceptions to the stated rule (1). Furthermore, it is generally assumed that the Lion-capture rules do not apply recursively in case of multiple Lions, so that hypothetical recaptures ruling out a capture would not have to obey those, just like they also would not have to keep their King out of check. H.G.Muller (talk) 20:42, 21 September 2016 (UTC)

There is another wrinkle, though. Suppose Black wishes to capture a White lion with her lion, but the White lion is protected by a promoted kirin. Thus far, the capture would be illegal. But then what if the proposed recapture would be illegal, because the White lion is additionally attacked by a Black rook that is pinned to Black's king by a White bishop?! (The proposed sequence is 1.LnxLn, because 1...LnxLn could be met by 2.RxLn.)

My interpretation is: since leaving your only king in check is not illegal (just stupid), it matters not that 2.RxLn is suicidal. The article already says that if we went to the position after 1.LnxLn, White cannot play 1...LnxLn. Hence Black's lion is indeed safe from recapture in the following turn and hence the capture must be legal. (It would be legal even without the pin, but then 1.RxLn would be far simpler. The pin creates a real reason to play 1.LnxLn.) So here is an example of applying the rules recursively. Double sharp (talk) 05:58, 2 March 2016 (UTC)
 * Indeed, this is an ambiguous case. It seems that the modern rule at least is to apply the rules 'front to back': 1.Ln x Ln would be forbidden here, despite the fact that the 1... +Ky x Ln would be forbidden too. The reasoning here could be that by that time you have already lost. Similar to capturing a Queen that is attacked by King + Rook put protected by King with your King in FIDE. This is not allowed, even though the recapture would also not be allowed, because it exposes itsef to the Rook. You are not allowed to expose your King to pseudo-legal capture in FIDE. Likewise, you are not allowed to expose your Lion to pseudo-legal capture after Ln x (distant) Ln in Chu. I sometimes express this by saying the Ln x (distant) Ln makes your Lion 'absolutely royal' for one turn.You just lose the instant the opponent recaptures your Lion, and that you could make him lose the next ply by taking that Lion again is just too late. Game over... H.G.Muller (talk) 18:47, 18 September 2016 (UTC)
 * Note that this means that as far as I am concerned example 8 is wrong, and should have the opposite conclusion: Black canot capture the white Lion. I already edited the diagram of example 5, as in the original diagram there was no way to capture Lion and the Pawn that protected it together, as both were distant.H.G.Muller (talk) 22:05, 18 September 2016 (UTC)
 * Hmm, yes, your interpretation does make sense too. I have added a paragraph showing the other conclusion, since it is really not sure what would happen. My logic was that illegal moves should not be considered as true protection, and hence the lion is not really "in check". It is not quite like moving your king into check from a pinned piece, but rather the impossible FIDE scenario of moving your king into check from another king. So the FIDE analogy is a little weakened. It's also important to note that Chu has a distinction between actually illegal moves (lion-trading) and moves that are not illegal but just very stupid (moving your king into check when you don't have a prince). So, while I would argue that the latter sort counts as protection (we need an example 9 now!), the former sort does not.
 * In fact, the way I see it, the immobilisation of pinned pieces in FIDE is a property of the king, not that piece. The pinned piece could move if the king was not behind cowering behind it, so its immobility is situational. But the king cannot move to an attacked square, no matter where it is: that is an innate property of the piece. So also for the lion. It is not allowed to expose itself to recapture when it captures a lion, and I do not see why this innate property of the lion would suddenly stop being true when it is itself doing the recapturing. But, in general, it does seem like the sort of fine print that would never come up in practice, but could lead to spiraling discussions. Double sharp (talk) 14:43, 21 September 2016 (UTC)
 * I have been trying to digest this, but I don't get your point. FIDE rules say it is not allowed to expose your King to PSEUDO-LEGAL capture, i.e. explicitly ruling out recursive application of the rule. (Which defines the difference between legal and pseudo-legal.) That is a choice; defining it recursively would be different, but not inconsistent. The choice is no doubt motivated by the idea that the game ends when the first King is captured, so that any exposure you would suffer from that is illusory, as the opponent will never get to move anymore. The problem is that in Chu Shogi 'forbidden' means something different as in Chess, where it means 'rejected, and retry until you get it right'. But in Chu Shogi it means you lose. This has erased the difference between a Lion capture being forbidden, and the recapture being instantly winning. If the recapture is instantly winning, the possibility to re-recapture is no longer relevant.H.G.Muller (talk) 20:43, 2 October 2016 (UTC)
 * BTW, the most general way I could think of for defining 'pseudo-legal' is this: "moves allowed when considering WHERE all pieces are, and WHAT they are, but not HOW ENEMY PIECES MOVE." I.e. allowed under the fiction that the opponent will forcibly have to pass his next turn. The "what they are" is important for things like 'iron' pieces, or more specific restrictions on what piece type can capture which other. But also for Atomic explosions and Fire Demon burns, to determine the full effect of the move. Rules like e.p. capture, or the Lion counter-strike rule still fall in the realm of pseudo-legality. The rules for not moving into check, or 'protected Lion', however, rely on opponent moves, and define the difference between legal and pseudo-legal. Such rules can be formulated (non-recursively) in terms of pseudo-legal moves, or (recursively) in terms of legal moves.H.G.Muller (talk) 11:22, 4 October 2016 (UTC)

Is there an easy way to do these diagrams with graphics? The current "red equals Gote's piece" is confusing to the eye, like an unambiguous passage that nonetheless requires a second reading to clarify. OneWeirdDude (talk) 16:03, 4 May 2016 (UTC)
 * Yeah, it's not ideal when red is pretty firmly associated with promoted pieces. I guess you could flip the enemy pieces with CSS, but it gets annoying to read (and is patently illegible for people who can't read kanji). The Japanese article uses the red=Gote's piece convention, so I used it here. Double sharp (talk) 03:46, 8 May 2016 (UTC)
 * Just an idea: couldn't we use 'reverse video' for the other side? White (or promoted red) kanji on a black background? (I now tried this out in the diagram on the tak page)H.G.Muller (talk) 18:54, 18 September 2016 (UTC)

There ought to be a way to do this so that one player's pieces are upside down w.r.t. the other's. OneWeirdDude (talk) 20:49, 20 February 2019 (UTC)

Importance rating
In the context of the Chess WikiProject, this surely is "low". In the context of a hypothetical Shogi WikiProject, this variant would at least deserve "Mid" as the most-played of the lot. (I suppose one can argue about tori and wa, but they have nothing like the history of chu.) Perhaps tenjiku would deserve it too as having had a Western cult following for a while, until it was given up due to a mistaken reconstruction of the rules rendering the game probably a forced win for Sente. Double sharp (talk) 14:45, 24 February 2016 (UTC)

Readings of piece names
I've been struggling to get through the Chushogi Renmei's site, but here are their given readings for the piece names (which I should not have gotten wrong: I can at least read kana competently).

Additionally, they abbreviate the blind tiger as 盲 instead of 虎, and the drunk elephant as 酔 instead of 象. Double sharp (talk) 15:35, 29 February 2016 (UTC)
 * Pawn: hohei, fuhyō, fu. Promotes to gold general, but here read と金 tokin. (Is this special name a spilling over from standard shogi?)
 * Go-between: chūnin. Promotes to drunk elephant.
 * Ferocious leopard: mōhyō. Promotes to bishop, but here read chorokaku.
 * Copper general: dōshō. Promotes to side mover.
 * Silver general: ginshō. Promotes to vertical mover.
 * Gold general: kinshō. Promotes to rook, but here read ginbisha.
 * Blind tiger: mōko, mekura. Promotes to flying stag.
 * Drunk elephant: suizō. Promotes to prince.
 * Prince: taishi. King: ōshō, gyokushō.
 * Kirin: kirin. Promotes to lion.
 * Phoenix: hōō. Promotes to queen.
 * Lance: kyōsha, yari. Promotes to white horse.
 * Reverse chariot: hensha. Promotes to whale.
 * Vertical mover: shugyō. Promotes to flying ox.
 * Side mover: ōgyō. Promotes to free boar.
 * Flying stag: hiroku.
 * White horse: hakuku.
 * Whale: keigei, geigei.
 * Flying ox: higyū.
 * Free boar: honcho.
 * Rook: hisha. Promotes to dragon king.
 * Dragon king: ryūō. Promotes to soaring eagle.
 * Soaring eagle: hijū.
 * Bishop: kakugyō. Promotes to dragon horse.
 * Dragon horse: ryūma. Promotes to horned falcon.
 * Horned falcon: kakuō.
 * Queen: hon'ō, honnō.
 * Lion: shishi.

On the lion move
It strikes me that pedagogically (for a chess variant player), the best way to describe the lion is "knight, dabbaba, alfil, or doublemove king". The reason is simple. If you treat it simply as a doublemove king that cannot be blocked on its first step, then it becomes legal to pass a turn when it is surrounded by friendly pieces or the edge of the board, which it isn't. The turn-passing is a separate thing. (You cannot pass your first turn!) Double sharp (talk) 06:14, 2 March 2016 (UTC)

Stalemate in chu shogi
(A reply to an edit summary by OneWeirdDude.) It is possible that a player simply has no legal moves. Because the king's safety is not an issue, this must be because moves are totally blocked. One way of doing this is to advance pawns and go-betweens into the promotion zone and declining to promote them, creating a box that you can cram your king into as well as some stepping or ranging pieces. (For example: Black king 1a, rook 2a, bishop 3a, go-between 4a, pawns 1b, 2b, 3b, 4b; White king 6a, kirin 6b. Black to move.) Unlikely? Yes. But a rule set cannot be complete unless it deals with all possible scenarios. So, what is the ruling here? The historical sources do not say anything about such a case: presumably such ad hoc situations, like the recursively applied lion-capturing rules, would be argued about when they came about in reality. (Black could argue that it must be a draw, because the game cannot continue, but her king cannot be captured; White could argue that he should win, as the straightforward rule in chu shogi evolved into the current rule in shogi where a checkmated or stalemated player loses.) The only definite source for this is The Chess Variant Pages, which simply says "This makes stalemate non-existent in real games, and if there ever has been a rule for it, it is no longer known. For definiteness we can assume that stalemate is a win." This acknowledges the aforementioned pathological possibility, while simultaneously noting that it is not really a relevant concern, but puts a ruling on it anyway: if you are stalemated, you lose. Double sharp (talk) 19:41, 14 March 2016 (UTC)

Fumiaki Okazaki
Is the Okazaki rule named after him? It seems highly plausible, but I struggle to comprehend the sources. Double sharp (talk) 03:50, 8 May 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 2 external links on Chu shogi. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://archive.is/20150423170917/http://forum.81squareuniverse.com/viewtopic.php?f=18&t=1754&start=23 to http://forum.81squareuniverse.com/viewtopic.php?f=18&t=1754&start=23
 * Added archive https://web.archive.org/web/20070803010434/http://www.luckydog.pwp.blueyonder.co.uk/games/shogi/chu.htm to http://www.luckydog.pwp.blueyonder.co.uk/games/shogi/chu.htm

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 02:37, 7 August 2017 (UTC)

Colin Adams
Who is Colin Adams and why should we care? Just asking for the purpose of the wiki; I think he was originally my suggestion. OneWeirdDude (talk) 03:06, 1 January 2018 (UTC)

A translation of this source would be nice
I would dearly love to see a proper translation of this page from the Japanese Chu Shogi Association explaining many hard cases in the rules! Double sharp (talk) 14:38, 1 April 2019 (UTC)