Check out Symmetric Chess, our featured variant for March, 2024.


[ 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

LatestLater Reverse Order Earlier
Asymmetric Chess. Chess with alternative units but classical types and mechanics. (8x8, Cells: 64) [All Comments] [Add Comment or Rating]
💡📝Dmitry Eskin wrote on Tue, Nov 22, 2016 05:22 PM UTC:

H. G. Muller:
Thank you. But there are some details.
The Hunter has the first part of moving (to range of 2) as a leaping, although continues his moving as a linear rider. In other words, he ia as classical Bishop but always skips (and ignores) the first square in each direction (at melee range of 1), like the Knight.
And the Wyvern has too.
The Pegasus is R3 but 100% leaper to all of his moves (can leap through up to 2 pieces). It gives him a great flexibility and makes him as stronger as the Rook.

If there some unclear details in the current rules (at the main post), please propose that such determinations would be clear to all. Maybe my current determinations are not clear now.

Sorry if it is inconvenient for the current notations: these types of moves were invented before I got to know this theory.

"If such a move would go to the square skipped over by the opponent Pawn on the preceding move, it could both be interpreted as a normal non-capture move or as an e.p. capture."
Hmm, its interesting. I think that this move has a capture's priority (always e.p. capture if possible). But there are some cases in which e.p.capture is put own king under a check - then moving to this square is not legal right now, even it is legal without e.p. capture.

And yes, there is a problem with e.p. capture in a Fairy-Max - the pawns may capture such a way any piece (not only pawns), even friendly, its a bug and later it is out of sync for engines :)

P.S. How I had found, the strongest configure for elvish AI is classical R and Q for a rook and queen, not centralized, as a standard rook and a queen.


H. G. Muller wrote on Tue, Nov 22, 2016 04:00 PM UTC:

As a demo I have put up a page for Orcs vs Elves in my turn-based server ( http://hgm.nubati.net/variants/orc-elf/). It is rather easy to set up such a page for any variant; it only requires the interactive diagram included in it to be adapted to the variant in question.

It is not clear to me whether this variant has e.p. capture, though. Although the diagram is rather general in how it defines e.p. capture, there arises an ambiguity when Pawns have moves that can both capture and non-capture. If such a move would go to the square skipped over by the opponent Pawn on the preceding move, it could both be interpreted as a normal non-capture move or as an e.p. capture.


💡📝Dmitry Eskin wrote on Mon, Nov 21, 2016 10:01 PM UTC:

I would update auto statistics at this post (with the new extended Pegasus instead of blink):

Orc vs Elf = 22,5 : 17,5
Orc vs Human = 21,5 : 18,5
Elf vs Human = 41 : 39

P.S. I like the new Pegasus, this unit became a brilliant with a range of 3, and so I had updated the rules (removing a blink and upgrading the Pegasus).

H. G. Muller wrote on Mon, Nov 21, 2016 06:27 PM UTC:

Ah OK. Sorry I misunderstood the Knights. So 43 is the correct move rights after all. Or perhaps better, because there are two moves passing over the same first square: make one of those lame (70), and allow the other to terminate there (43). That way you prevent that Fairy-Max will try to search the same move twice. (The hash table would largely cure any ill effects of that, but sometimes hash entries get overwritten.)

If there is no incentive to develop, opening play might be VERY poor. It is never very good, because the first 5 moves are heavily randomized. So opening moves like h2h4 will not be uncommon. But because both sides suffer from that it willnot skew the results. And it is needed to cause enough game diversity; when you play 100 games you don't want the same two games be repeated 50 times each.

Note that a 13-7 score hardly proves anything; you should at least do 100 games, and even then the standard error in the score will still be 4%. I had played 40 games before I realized I was using the wrong Knights, and the Orcs were leading by about 75% against the Elves, though (26+ 7- 7=).


💡📝Dmitry Eskin wrote on Mon, Nov 21, 2016 03:40 PM UTC:

To H. G. Muller:

Alternative Knights are not lame Knights! They have up to all 12 squares to moves, like a linear piece (with a limited range up to 2). Regarding the rest, I will try it, thanks.

I tested the old configure and had found that:
1) Blink-Phoenix is terrible! Elf vs Human defeats a pawn with 1.B:a7 (if the 1....R:a7 then 2.Qe3 with a diagonal a7-h8 attacking and capture one of the rooks). It need to be fixed and I'll remove this ability at the next update of the Asymmetric Chess.
2) AI is weak at the openings and is strongly influenced by the way played alternative pieces, maybe its a problem with my mistakes of configuring (upper case) or with starting arrangements or with some pieces/pawns' design.
3) The total results of the balance for now (with time-control of 1 minute per 40 turns):
Orc-Human 13:7
Elf-Human 8,5:11,5 (but 5:5 when the elf is white and capture a pawn with 1.B:a7 and 3,5:6,5 when the elf is black)
Elf-Orc 9:11
4) The Wyvern seems undervalued and strong as soon as the Griffin (classic Rook) because of the starting activity

Maybe instead Phoenix' blinking I'll give to the Pegasus range of 1-3 (instead of 1-2), I'm testing for it. The old Pegasus has very big difference with a classic Rook in his power, because a classic Rook (Griffin) has sieged range (longer that long-range of bishop's diagonals). An average range of diagonal's direction is about 2,5, and an average range of orthogonal's direction is 4. That's why I think that 4+ distance is sieged range but a range of 3 isn't sieged. New Pegasus will be as strong as a Griffin (Rook) because of a great jumping ability.

P.S. Alterntative Knights (unicorn/werewolf) may be counted as lame knights + ferz/wazir.


Greg Strong wrote on Mon, Nov 21, 2016 03:26 PM UTC:

Hi Dmitry,

I'm not sure how you determined your piece values but they are almost certainly off from the true values.  I suspect the elves are as strong as the humans without the phoenix blink.

If you would like to try it, I created a preset for Game Courier and posted an open invitation.  I will take the Elves (black) without the blink power.  You can use the following link to accept the invitation:

/play/pbm/play.php?game=Asymmetric+Chess&log=mageofmaple-cvgameroom-2016-325-681&submit=Accept

Cheers,
Greg


H. G. Muller wrote on Mon, Nov 21, 2016 02:41 PM UTC:

" Do you know any online chess service supporting illegal moves (free moves by agreement between the players themselves)? It would be the best for playing many non-classic variants without any programming (no AI analysis, no AI player, no AI tips - but the natural, origin wood form of a chess doesn't have these options too). "

I took the liberty replying to this in the Asymmetric Chess comments, as it did not seem to have anything to do with ChessV. You could use a page like this one. Although it highlights how pieces can move, it does not enforce this, and you can move them any way you want. If you have your own web space you could copy the page and change the rules it uses for highlighting to be the rules you want.


H. G. Muller wrote on Mon, Nov 21, 2016 11:11 AM UTC:

I tried your game definition, and it needed the following fixes:

  • The first was my bad: the lame Knights did not require 43 move rights, but 70 move rights, because otherwise they would be able to make the W or F step already with full rights (3), and the secondary step would only toggle the termination bit (4) on. But the W or F step should not have any rights (0).
  • The piece ID of N and B equivalents in the piece lines was made lower case; this makes them seek the center. Otherwise Fairy-Max becomes very tardy developing them. I also did that for the Elvish Queen, as in Capablanca Chess the Archbishop also seemed to be worth more when seeking the center. Not sure if this is also true for the cylinder version, however. Perhaps the short Rook should also be center seeking to let Fairy-Max make better use of it.
  • I changed the name into orc-elf; it should now appear under that name in WinBoard's New Variant dialog
  • I appended " # fairy" to the game line to specify 'fairy' as parent variant. This variant implements normal Chess rules w.r.t. promotion, check, stalemate, castling etc., and is thus a good parent for most western Chess variants. Because it is a 'catchall' variant where WinBoard has no prior notion of the initial setup, WinBoard will pay attetion to the engine's setup command even when legality testing is on,which is another advantage.
  • I appended Betza move definitions to the game definition, which don'tmean anything to Fairy-Max, but are sent to WinBoard at the start of the game, to make WinBoard aware of how the pieces move. This enables WinBoard to highlight target squares in human-engine games, and properly disambiguate the move notation. (It is of course bad design that two independent, and therefore possibly conflicting move descriptions are needed. But the ability of WinBoard to accept piece definitions from the engine is a recent enhancement, and was then added to Fairy-Max as an afterthought. Ideally Fairy-Max would derive one set of move definitions from the other.)
  • I played a bit with the 'pieceToChar' string in the game line. Each position in this string corresponds to a WinBoard piece glyph, and specifies the piece it has to be used for. The first half of the string is for white pieces, the second half for black. The first five for each color are standard symbols for PNBRQ, the last one for K. In between come upto 17 other piece glyphs built into WinBoard. I selected some of those that seemed applicable. (Masked knight and unicorn for the lame Knights, commonly used pictograms for Archbishop and Chancellor. WinBoard does not have enough rookish or pawnish glyphs, however, so the 'Skip-Rook' is still depicted as an orthodox Rook. Of course it is a matter of taste whether you want all pieces to look different or just use the orthodox images. This whole issue can also be solved purely on the GUI level, by configuring use of (and providing) external images in WinBoard.)

______________________________________________________________________

// Asymmetric Chess (Orc-Elf)
Game: orc-elf # P..R..B.....N...Q....K.......q.....br...p.nk # fairy
8x8
6 4 5 7 3 5 4 6
8 11 10 9 3 10 11 8
p:125 -16,24 -16,7 -1,5 1,5
p:130 15,24 17,24 15,7 17,7
k:-1  1,34 -1,34 1,7 16,7 15,7 17,7 -1,7 -16,7 -15,7 -17,7
n:430 1,070,-15 1,070,17 16,070,17 16,070,15 -1,070,15 -1,070,-17 -16,070,-17 -16,070,-15
b:280 17,7 15,7 -17,7 -15,7 34,7 30,7 -34,7 -30,7
R:450 2,3,1 32,3,16 -2,3,-1 -32,3,-16
Q:860 1,3 16,3 -1,3 -16,3 14,7 31,7 33,7 18,7 -14,7 -31,7 -33,7 -18,7
R:310 1,7 16,7 -1,7 -16,7 2,7 32,7 -2,7 -32,7
q:1060 15,103 17,103 -15,103 -17,103 14,7 31,7 33,7 18,7 -14,7 -31,7 -33,7 -18,7
n:440 17,070,1 17,070,16 15,070,16 15,070,-1 -17,070,-1 -17,070,-16 -15,070,-16 -15,070,1
b:280 34,3,17 30,3,15 -34,3,-17 -30,3,-15
#
# P fWscWifnD
# p fFifnA
# N afsW
# n afsF
# B FA
# b jB
# R jR
# r WD
# Q RN
# q oBN


💡📝Dmitry Eskin wrote on Mon, Nov 21, 2016 01:48 AM UTC:

to H. G. Muller
Thanks you for your program and advices!

I put this variant and it works but I need to configure pieces' images: elvish (black in this case, and white for future presets) bishop and knight must exchange their starting places.

I use classic preset (images), but maybe it's better to draw alt. queens as chancellor and archbishop, but I don't know how to configure it.

// Asymmetric Chess (Orc-Elf)
Game: normal # PNBRQ.........Kpnbrq.........k
8x8
6 4 5 7 3 5 4 6
8 11 10 9 3 10 11 8
p:125 -16,24 -16,7 -1,5 1,5
p:130 15,24 17,24 15,7 17,7
K:-1  1,34 -1,34 1,7 16,7 15,7 17,7 -1,7 -16,7 -15,7 -17,7
N:430 1,043,-15 1,043,17 16,043,17 16,043,15 -1,043,15 -1,043,-17 -16,043,-17 -16,043,-15
B:280 17,7 15,7 -17,7 -15,7 34,7 30,7 -34,7 -30,7
R:450 2,3,1 32,3,16 -2,3,-1 -32,3,-16
Q:860 1,3 16,3 -1,3 -16,3 14,7 31,7 33,7 18,7 -14,7 -31,7 -33,7 -18,7
R:310 1,7 16,7 -1,7 -16,7 2,7 32,7 -2,7 -32,7
Q:1060 15,103 17,103 -15,103 -17,103 14,7 31,7 33,7 18,7 -14,7 -31,7 -33,7 -18,7
N:440 17,043,1 17,043,16 15,043,16 15,043,-1 -17,043,-1 -17,043,-16 -15,043,-16 -15,043,1
B:280 34,3,17 30,3,15 -34,3,-17 -30,3,-15

Update: randomly get exchanging of the black knight and bishop, but how to get chancellor and archbishop images, it's a riddle...


H. G. Muller wrote on Sun, Nov 20, 2016 11:12 PM UTC:

"For example, if I put a move like 15,FFFF070 (-16) then a piece can move from g8 through f7 to e7. If put 15,1070 (+1) then a piece can move from g8 through f7 to... d6. But any other tries with -1 and +16 do nothing. I need help for it."

I agree, for pieces that change direction the definitions are awful. But the latest Fairy-Max version (5.0b) actually improved on that. Rather than having to specify the step change in hexadecimal in the upper bits of the move rights, (which is how it is interally stored) it allows you to simply write the secondary step as an (optional) 3rd element in the move definition. So 17,3,1 would now be a Griffon move, a slide that first steps diagonally (17), and then continues orthogonally (1). Fairy-Max itself then calculates what bits have to be changed to make a 1 from a 17.

The 'Skip Rook' in a sense is also a 'bent slider' like the Griffon, except that the move step does not change direction, but only length. The first step of the slide is that of a Dababbah, then the slide continues with single steps. The description of moves that do this would look like 2,3,1 32,3,16 -2,3,-1 etc. The 'lame Knights' move description would be like 1,43,17 (blockable orthogonally) or 17,43,1 (blockable diagonally).
 

I might equip future versions of Fairy-Max with a compiler for Betza notation, so that the piece definitions can be given in this form. That wouldmake it a lot more user-friendly.


💡📝Dmitry Eskin wrote on Sun, Nov 20, 2016 09:58 PM UTC:

to Greg Strong
Hi Gregory!
Thanks for the feedback.
The blink is added only to the Phoenix (Archbishop) not the Hunters (Bishops).
Adding the blink wasn't a simple decision, but at the current moment I think it is necessary (and is a minimal fee for the balance). A Phoenix without the blink (classical Archbishop) has an estimated value 8.2 (instead of 10.6), a half pawn less than a Dragon (Chancellor). And a Fairy promoting to non-blinking Phoenix has an estimated value 1.25 (instead of 1.33). Summary a blink removing makes elves 3.0 classical pawns weaker. It will be wonderful if such elves with so much estimating weakness will be balanced. But now I believe to these estimatings more than to a wonderful equality.

As an alternative variant without the blink, there is possible to buff the Pegasus, from 2 up to 3 range (leaping) but this is controversial and gives more advantages to elves (according to my estimates extended Pegasus has value of 4.9 instead of 3.1)


Greg Strong wrote on Sun, Nov 20, 2016 05:46 PM UTC:

Hi Dmitry,

This is an interesting looking invention you have here.  I like the concept of the game and the principals you followed.  It has led to a promising game.  Of course, something so radical will require testing.  The only thing that is unfortunate is the bishop "blink". I think the Elven pawns may be powerful enough that the extra blink might be unnecessary.

Regarding playing, your best bet is Game Courier on this site.  You can open one of the Chess presets, click Edit to modify it, and just erase all the code in the seven boxes at the bottom to stop it from enforcing any rules.  Then, in the first box, change the name from Chess to Asymmetric Chess, enter your user id and password, and click Save.  It will give you a link that you can then use to send out invitations to play.

Cheers,
Greg


💡📝Dmitry Eskin wrote on Sun, Nov 20, 2016 01:16 PM UTC:

Yes, the Pawns are the most dangerous because they are very complex and progressive units, especially with pawn's chains in any different forms.
But I can't configure Fairy-Max to play this variant because of problems with configure linear-leaps (unicorn/werewolf, wyvern/hunter). It has very strange instructions and unclear mechanics of configuring complex moves, even for a programmer.
For example, if I put a move like 15,FFFF070 (-16) then a piece can move from g8 through f7 to e7. If put 15,1070 (+1) then a piece can move from g8 through f7 to... d6. But any other tries with -1 and +16 do nothing. I need help for it.

Do you know any online chess service supporting illegal moves (free moves by agreement between the players themselves)? It would be the best for playing many non-classic variants without any programming (no AI analysis, no AI player, no AI tips - but the natural, origin wood form of a chess doesn't have these options too).


H. G. Muller wrote on Sun, Nov 20, 2016 11:26 AM UTC:

I is pretty hard to judge if your other armies are anywhere near balanced with the FIDE ('Human') army, without computer assistance. Because of the unconventional Pawns. Any uncertainty in their value will be multiplied by 8.

It seems all the pieces you use are well within the capabilities of Fairy-Max. So it should be easy to configure Fairy-Max to play this variant.


💡📝Dmitry Eskin wrote on Fri, Nov 18, 2016 07:09 PM UTC:

English is not my native language, that's why if you'll find any mistakes, please point on its to me.
Also, I'm opened for new suggestions, including the balance of some units and estimating their real value. But please consider that the main goal of this chess variant is the asymmetric balance with very simple rules and saving all basic mechanics and types.
And it will be very good if this chess variant will be implemented on any online chess engine, including AI testing balance, I hadn't done it yet.

Thanks for this site. And for "Chess with different armies" also, I'm not agree with some of things (basically with unaltered pawns, strange extra moves of typical pieces, extended/limited pieces), but learning of this experience was important for me, helping to realize what is better to choose the implementation of pieces.


15 comments displayed

LatestLater Reverse Order Earlier

Permalink to the exact comments currently displayed.