The site has moved to a new server, and there are now some issues to fix. Please report anything needing fixing with a comment to the homepage.



The Chess Variant Pages




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

Comments/Ratings for a Single Item

Later Reverse Order Earlier
Foolish King Chess. Players have different armies and victory conditions. White has a fool for a king. (10x8, Cells: 80) [All Comments] [Add Comment or Rating]
H. G. Muller wrote on 2020-09-28 UTC

The diagram's AI uses a bit of a quirky method for detecting game end: it incrementally keeps track of a 'royalty count', where each piece type has its own 'royalty weight'. Pieces defined as royal get weight 1024, pieces defined as baring get weight 1. If the opponent's royalty count is negative after your (pseuo-legal) move, you win immediately. If not, and if your own royalty count is exactly zero, you lose. Except when the opponent is at 0 too. Then, if this was the consequence of the latest move (i.e. before the move the opponent was still positive) the game continues to check the legality of this counter-baring; otherwise it is ruled a draw.

Normally the royalty count starts at 1, so that loss of any royal makes it negative. For extinction royalty 1024 times the number of spare royals is added to the initial value. I there are any baring piece types, the initial 1 is replaced by the number of baring pieces (so that you will hit 0 when they are all gone).

You can specify which piece types should be counted for the baring, through a number of baring=N parameters. For cases like Shatranj, a short-cut baring=0 is recognized as the instruction to make all non-royal pieces baring. (Piece-type numbering starts at 1.) This system so far was able to handle all cases I encountered; in this case I defined the Rooks and Archbishops as baring, and to prevent black from losing when he loses both Rooks, I defined the King as baring as well. Then black can only lose by baring if he loses his King, but as the King is defined as royal, that would make him lose anyway.

 


Greg Strong wrote on 2020-09-27 UTC

Wow. Very nice! Thanks for updating the script to accommodate!

Flexible castling is fine. I emailed the author to inquire about the castling. The email didn't bounce but I haven't heard back either. If I don't hear back, I'm inclined to change the rule to flexible castling. I think that's fair for a couple of reasons, which I will go into if that time comes...

After a few days, if I haven't heard back from the author, and others haven't weighed in, I will update this page and include the interactive diagram.


H. G. Muller wrote on 2020-09-26 UTC
files=10 ranks=8 promoZone=1 promoChoice=*H*F graphicsDir=/graphics.dir/alfaeriePNG/ squareSize=50 graphicsType=png symmetry=none maxPromote=2 extinction=1 royal=9 royal=8 baring=4 baring=5 baring=9 pawn:P:ifmnDfmWfceF:pawn:a2,b2,c2,d2,e2,f2,g2,h2,i2,j2,,a7,b7,c7,d7,e7,f7,g7,h7,i7,j7 knight:N:N:knight:c1,f1,h1,,b8,c8,h8,i8 bishop:B:B:bishop:d1,g1 rook:R:R:rook:a1,j1,,a8,j8 cardinal:C:BN:cardinal:b1,i1 queen:Q:Q:queen:,,f8 hunter:H:BnDD:archer:,,d8,g8:0,10 fool:F:mkNmkW:fool:e1:10,0 king:K:KisO2isO3ilO4kN:king:,,e8

It seems the above diagram does approximately what is needed. The diagram script initially choked on only black having a piece that can castle; the code for determining the castling partners only considered white pieces, assuming symmetry, and thus left things undefined, which propagated through the calculation to finally make everything undefined. I fixed this now.

I also had some problems with the implementation of baring. This counted only non-royal pieces. But in this variant black should not suffer even from being totally bared, and I thought I was clever by defining the King as a piece that should be counted. But the old code overruled that. I changed the code so that it now only discounts the royal from the total number of pieces when it was not explicitly defined by a royal=N line (in which case the last piece in the table is taken to be royal). This allowed me to define the black King as a baring piece as well as a royal, so that being bared is the same as losing your King. For white, however, the Rooks and Archbishops are baring pieces, and the Fool an extinction royal.

Another problem was the promotion. I could have defined it as Shogi promotions, as there is apparenly no choice. But then white and black pawns would have to be different piece types, as they promote differently. (And it would allow youto defer.) So I opted for another solution: I allow promotion to Hunter or Fool, but only to pieces 'in hand'. And the white hand starts with only 10 Fools, and the black hand with only 10 Hunters.

The capture restriction on Fools is implemented by a BadZone JavaScript routine embedded in the page. The AI pays attention to this. For castling I just took flexible castling.


Greg Strong wrote on 2020-09-25 UTC

Ok, thank you.  That is helpful.  Maybe something like:

A Fool may never move to a square that is a kings-move or knights-move away from a King.  (In other words, it cannot move into a position where it would "check" a King.)  Likewise, a King cannot move to a square that is a kings-move or knights-move away from a Fool (as though it would be moving into check, even though Fools can't actually capture.)

I am a bit worried about the promotion rule: can white promote only to Fool, or is this just an extra possibility? The Fool is practically useless as attacker, similar to an Elephant in Xiangqi. That they are uncapturable and that you can have many cannot compensate for this; N times zero is still zero. The possibility for white to win an end-game seems bleak if he can only promote to Fool.

That is a good point... However, I think it is very, very hard to balance unequal armies, even without adding different victory conditions.  So, I think this game is almost certainly unbalanced.  But I think it could still be fun to play.  If you could get the interactive diagram to play a close approximation, that would be awesome.  I'd add that to the page.  But can it really do the free castling rule?  (If not, I think that's ok, I'm not even certain that was the intention.  The page literally just said "castling is free".)


H. G. Muller wrote on 2020-09-25 UTC

It is just a variant of the Xiangqi King-staring rule. In this case it is illegal to have the royals see each other throug a W or N move, rather than an R move. In XBetza I would configure it as a kWkR move on both King and Fool. The Fool is an (extinction) royal. (And since there will never be more than one King, the latter could be considered an extinction royal too.) The white majors are subject to a baring rule (and the Diagram's AI would allow you to specify which piece types participate in the baring). I guess the Diagram could give an almost perfect implementation of this variant with the aid of a user-supplied JavaScript routine BadZone() to outlaw Fool capture by most piece types, and switching off the checking rule. The only imperfection then is that some stalemates would be seen as a win for black.

I am a bit worried about the promotion rule: can white promote only to Fool, or is this just an extra possibility? The Fool is practically useless as attacker, similar to an Elephant in Xiangqi. That they are uncapturable and that you can have many cannot compensate for this; N times zero is still zero. The possibility for white to win an end-game seems bleak if he can only promote to Fool.


Greg Strong wrote on 2020-09-25 UTC

I have made a first pass at cleaning up this page. Graphics have been added, formatting issues have been fixed, and some of the writing has been edited for clarity.

The cryptic part in the description of the Fool remains. That will need to be replaced with something, but I'm not sure what.


Greg Strong wrote on 2020-09-25 UTC

Ah... This part made sense to me. The wording can be improved but the noteworthy thing here is that the hunter is a lame dabbabah-rider. I believe the bishop move is entirely normal. But it was difficult for the author to describe an orthogonal move that can only land on same-color squares but can still be blocked on the other squares.


Ben Reiniger wrote on 2020-09-25 UTC

On fools, the paragraph seems to mean that neither fools nor kings can move such that the fool "attacks" the king. The convoluted language comes I think in part because the fool cannot capture any pieces at all, and so the author is trying to avoid the word "attacks" or its variations.

Several of this author's pages say just "castling is free," which I can only attribute to the historical rule? I haven't read all of them to see whether they elaborate somewhere.

I'm also confused about the hunter movement:

...The hunter can't jump over any agent when it's travelling either diagonally or orthogonally. When travelling purely orthogonally, the hunter can never even jump any agent located on squares of opposite color. ...

The second sentence here seems to be redundant, unless diagonal movement is supposed to allow jumping over opposite-color squares?


Greg Strong wrote on 2020-09-25 UTC

I don't really understand this game.  This paragraph is particularly cryptic:

Fools are forbidden to step onto a square that means that the enemy king is on a square that the fool could have reached (had it been unoccupied). On the other hand, the enemy king is forbidden to step onto a square that the fool could have reached in just one move, so the fools can contribute to mating the enemy king. The king can take out the fool, but the fool can not take out the king.

Also, the statement that "Castling is free" isn't very descriptive.  Does this mean the historic free castling rule?

And there are other confusing things as well.  If the author is around to clarify, that would be great.  If not, and someone else understands what is meant here and can help clarify this text, that would be also be fine.  If this doesn't happen in 30 days, I will hide this page.


9 comments displayed

Later Reverse Order Earlier

Permalink to the exact comments currently displayed.