Comments/Ratings for a Single Item

A minor quibble here about board size: it can be considered a mutator. There are variants which propose placing an 8x8 standard chess set-up in the middle of a 10x10 or 12x12 board. This does change the game a fair bit. Now, with a 10x8, you can use it a few ways. You might have a file a rook can step to on either side of the standard 8x8 set-up, or you could 'play the long way' and set up with 6 rows of 8 squares empty between the 2 sides, or even move both sides up 1 square, so they ar the standard distance apart, but there is an extra row behind each side. Circular boards have long been used, also, for example. But what you are really doing here is examining how board geometry affects play and affects the utility of various pieces (eg: on a Byzantine circular board, bishops are nerfed and rooks are enhanced. In other words, you're playing Chess with Different Boards.

Isn't that exactly what chess variants are: games with non-standard armies and/or non-standard boards? I fail to see how there is anything new in your proposal.

I also don't see the merit of trying different armies on all kinds of different boards. It seems the famous adagium "it only takes 10 sec to invent a new chess variant, and unfortunately some people do" applies here. The idea of using different pieces for each player is a great idea, but what would you get by using, say, 10x8 boards that you don't get on 8x8 boards? It seems a difference just for the sake of being different, without adding anything new. There might be an incentive to do that when the original game has been 'beaten to death', like orthochess, but that doesn't seem to be the case here at all.

Not sure how interesting they would be by themselves, but for now one might be content just with new variant(s) in which both sides use the same new armies (but a new army which is different from e.g. the standard 10x8 Capablanca Chess army) in e.g. new 10x8 variant(s).

At any rate, I'm not quite sure if e.g. a variant of Marsailles Chess based on Chess With Different Armies wouldn't work okay if just using the standard 4 armies as in Chess With Different Armies.

Oh, I make no claims about that position being good.  It was just an idea I was toying with a long time ago that I haven't tought about since Kevin's post.  Maybe I'll start thinking again ...

@Greg

I don't see why there are two queens and only one colour bound caliph. I asume it is for balance but maybe something else could be done as from an artistic point of view it does not make much sens. What about caliph's on the starting squares queen squares as you have drawn them and instead of your central caliph a more powerfull queen. If you are ok with the many bishops moves, maybe BDHN would work fine. This is obviously a first glance judging and not a thorrow investigation :)!

It is an excellent concept, and, in my opinion, an excellent game.  That said, balance is very difficult to accomplish.  There are complaints about Betza's standard armies not being balanced, and while they are not perfectly balanced, they are pretty darn close - especially given that they were developed before the advent of sophisticated engines capable of evaluating them.  This was the culmination of Betza's life's work - it is unfortunate that he disappeared right before such tools became available.  The first version of ChessV, 0.1, was pretty much a Betza engine - Chess with Different Armies and Chess with Differently Augmented Knights.  As soon as it was ready I emailed him, but he had gone dark a few months before ...

Regarding the suggetions, I think 10x8 "Capablanca with Different Armies" is certainly worthy of pursuit, and probably easier to balance than 8x8 since there are more pieces.  In fact, I made such a game of standard capablanca vs. colorbound capablanca about 11 years ago.  Here's the starting array:

/play/pbm/play.php?game=Capablanca+Chess+with+Different+Armies&settings=SS_vs_CC

An excellent concept!

Inspired by it, I can suggest many Chess With Different Armies-like variants that, to my taste, might be especially interesting to try out sometime:

That would be with the Different Armies idea used to make any number of versions of the following variants 1) 9x8 Symmetric Chess; 2) 10x8 Janus or Capablanca Chess; 3) any number of 10x10 variants, such as Grand or Sac Chess; 4) 4x16 or 5x16 Circular Chess variants; 5) Glinski's, Symmetric Glinski's, McCooeye's or Hexajedrez (91 cell Hexagonal Chess variants); 6) 4-Way Chess; 7) Crazyhouse, Chessgi or Hostage Chess; 8) Pocket Mutation Chess; 9) Backlash; 10) Marsailles or Progression Chess.

@HG Muller

It seems to me that you have placed the pieces for bent bozos in reverse order of what would it be in the opening. My take on this would be that most pieces should as much as possible move towards the center with their first move. I believe that happens only for black, i.e. the example you gave.

For example if bozos are white then the left aanca should be placed on g1 as it moves towards the center that way. But from black's point of view the right aanca would move towards the center if you place it at the beginning on g8. I think it's not that tricky what I'm saying and it goes that way for all three pair of pieces. See you soon!

Yes, castling with the corner pieces is always possible in Chess with Different Armies. The Chiral Griffons are not color bound, so this would be O2 castling. It seemed better to set up the Sastiks on c- and f-file, even though they are really Knight replacements. This suits both the move of the Satstik and Chiral Aanca better. The latter can get out of the Kings way for castling througj the trajectory g1-f1-c4/b5 after the Sastik moved from f1 to f3 or g3. On f1 the King itself would block it, and you would be forced to break the Pawn shield to get the Right Aanca out.

Good point. Fixed that now. Unfortunately the diagram will always only consider one piece royal, so it now will not mark stepping into check with the Clobber King as forbidden move. But I guess that is just a minor flaw; this forbidden-move feature was only introduced for Caissa Brittannia, to indicate the numerous forbidden moves of the Royal Queen there. For symmetry I should perhaps also switch it off for the other King (by adding "royal=0" to the diagram definition).

The king is supposed to move three squares on the queenside if the queenside rook is colorbound. This is so the piece doesn't change colors.

Thanks Fergus,

It looks like there's a big set of Alfaerie graphics. I was looking for a war machine icon that I saw somewhere, and i was able to find it. In fact, there's a few versions. I appreciate it.

Thanks! :)

It is "Alfaerie for Chess w/ Diff. Armies."

After reading these comments I became curious what graphic set is used for Chess with different armies (the version shown on this thread). I couldn't find out. Does anyone know?
 

Hi Nick,

Good questions.  I don't believe there are, as yet, definitive answers to them, but I'll take a shot ...

The relative balance of the armies is a topic of some disagreement.  I think it is fair to say that they are not perfectly balanced, but they are probably within one pawn.  I believe I remember Fergus saying he conisdered the Nutty Knights to be the strongest army.  David Paulowich, a very strong player, considered the Remarkable Rookies to be the most powerful.  I, personally, think the Colorbound Clobberers might be strongest, so long as they can avoid getting their extreme colorboundness exploited (they start with four colorbound pieces, if they get down to only two, but both on the same color, they are in trouble.)

I also think the strength of individual pieces definitely depends to some extent on what other pieces are on the board.  Therefore, the armies have a sort of Rock-Paper-Scissors effect where an army can be better against one type of army than another.  I was doing some in-depth computer analysis to try to determine this, but I stopped part-way through and it was a long time ago.  If I remember correctly, I was beginning to think that the Rookies may be the best for countering the Clobberers, and the Knight may be best for countering the Rookies.  But don't put too much faith in that.  In one of the most recent comments, H. G. Muller provides a chart of win/loss percentages of the various army matchups based on large numbers of games with Fairy-Max.  That's the best we have to go on now, but even that should not be considered definitive.  The problem with computer analysis is that it is applying the evaluation terms we programmers provide.  Automated tuning is possible, but requires a ton of games and is still limited.  And there probably should be other evaluation terms we haven't included at all because we don't know we need them, e.g., a huge penalty for the clobberers having only two colorbound pieces on the same color.  We have a similar 'two bishops' bonus in orthodox chess because we know it works and we know how big it should be.  But we don't know what we need here and computers are difficult to use to find it - kind of a chicken-and-egg problem.  But, in any event, the four official armies are close enough for a fun, fair, friendly game between human players.  If we had a library of thousands of expert games like we have in orthodox chess, however, we might well determine that an army is at a significant disadvantage.  Also note that the experimental armies are very uncertain.  The Meticulous Mashers, for instance, is way over-powered and should not be used except in a handicap game.

If I ever became ambitious and maybe decided to come up with my own army (loosely translated to "if I ever have to time"), what are the processes to do this.

Well, you can always do what others have done and just post it :)  There is no official process.  If you have an idea, start a thread and get feedback.  I can be of assistance with some basic computer testing if the pieces aren't too radical.  This is limited, as already mentioned, but can identify some that are way out-of-whack.  (It didn't take too much testing to determine that the mashers were far out of balance.)

Personally, one thing I frequently aim for when designing a variant is multiple different types of pieces with roughly the same material value.  This encourages uneven exchanges, leading to more different-army match-ups despite starting with the same armies (and thus avoiding the whole sticky question of balance.)  For an example, look at Opulent Chess.  It has four different piece types, the Bishop, Wizard, Lion, and Knight (which is augmented), which all have roughly the same material value.

155 comments and most of them extensive.  I do apologize if I'm asking material that has already been covered, but there's not a lot of time in my day as of late.  I've recently started looking at (haven't played) variants with different armies - this is notably one of the most popular variants on this site.  

I'm in no regards an expert (or anywhere close) to determining piece/army values and I have several questions surrounding it.  

- How even are the armies used in this variant?
- How are piece/army values determined and are they static based on board position, what pieces are currently in play, the army of the opponent, etc?
- How does one test and determine equality of different armies?
- If I ever became ambitious and maybe decided to come up with my own army (loosely translated to "if I ever have to time"), what are the processes to do this.  

I've seen this variant around for many years, I've come across recently Spartan Chess, I'm familiar with the Maharaja Chess where it's one piece vs an army (which I think is always a win for white if played right).  Does anyone have experience with this and have tips?  What if I wanted to do something on a 10x10 board?

I'm sorry if this has already been covered.  Thanks!

Nick

Well, if you were not talking about stalemate, it seems you had no point at all. Without stalemate, King capture and checkmate are fully equivalent, so that using one or the other can certainly not be responsible for a "gulf of draws". Fact is that most Chess-like games have a rather large 'draw sector', and the initial position is usually in that sector. This is a logical consequence of that you run out of material, and whatever remains at some point will not give you enough advantage to force a checkmate / king capture. The only remedy against that which I am aware of is piece drops. Then the material is recycled, so that you never simplify to a trivial draw. Games could still end in draw through repetition, however. In mini-Shogi they even plugged that hole: repetition there is a loss for sente.

I don't think that comments to Chess with Different Armies is the proper place to discuss this, however. So to get back on topic:

I played a few test games between Fairy-Max and ChessV. The latter at 15 sec/move, the former at 40 moves / 5 min. Fairy-Max seems to be a bit stronger at these settings: it won by 16.5-7.5. There always was an interesting fight, however. It is important to do a test like this to see if any engine has obvious 'blind spots', which would not be revealed in self-play.

I stated before that misconceptions about piece values hardly affect the rusults of self-play tests, where both sides share the misconception. This is because no matter which piece it thinks is stronger, one of the two players will always avoid the trade. But this doesn't apply for some strategic misconception. For instance, an when playing the Nutters, pushing all your pieces forward is a mistake, as they cannot outrun opponent passers. But even when the opponent doesn't know it either, and also thinks advancing is good, there really isn't any way he can stop you from advancing, in the end-game. So the engine handling the Nutters blindly runs into the knife, the opponent initially being sour about it. Untill at some point that opponent sees "wow! by trading/sacrificing such and such, I am suddenly left with an unstoppable passer!". This now doubt hurts the performance of the Nutters, compared to armies where you cannot make this mistake. It is a bit like telling an engine P is worth more than Q. There is no way the opponent will ever be able to prevent you from trading Q for P, even if he thinks P is worth more too. Having no Q to waste under those conditions should be a big advantage!

In watching the Fairy-Max vs. ChessV CwDA games, I learned that ChessV is pretty poor in the end-game. It gets completely passive, moving its King to a corner in stead of to the center, and moving it back and forth there. While the opponent in the mean time centralizes his King, advances his passers, etc. This behavior should hurt all armies somewhat equally, though. In general ChessV seems to have an inclination for pointless King moves; during the middle-game game it usually shuttles his King between g8 and h8 several times. What is also strange is that it seems to have a large preference to 'develop' his Knights (when handling FIDE) or to a lesser extent Fibnifs (when handling Nutters) to d1 and e1! This very odd behavior must hurt the performance of the mentioned armies, even in self-play, as again there is no way an opponent could ever prevent you from moving your Knights to e1&d1.

I didn't say or infer anything about making stalemate a loss.  That would be unfair to one player and you didn't specify which player- white or black, the attacking player or the defending player.  Yes, the stalemate rules in FIDE Chess also annoy me because it is possible to design chess variants that are absolutely drawless without decreasing fairness.

Simpler does not necessarily mean less drawish. The one extra move could hardly be a serious concern, as in practice people would resign when mate or capture of their King is unavoidable.  It seems that your real gripe with FIDE rules has nothing to do with checkmate, but with the fact that stalemate is a draw.

Have you any idea which fraction of FIDE GM games end in stalemate? Declaring stalemate a loss will have almost no impact on the draw rate. Have you measured whether the fraction of draws in engine self-play is measurably lower when you let them play under rules where stalemate is a loss?

Your idea of making the King an immobile piece is interesting. (Whether such a game would still qualify as 'Chess like' is debatable, of course.) Your claim that this would decrease draw rate seems unproven, however. My guess is that it won't. Players will just reserve more other material to evade checks, when the 'King' itself (I guess 'Flag' would be a more appropriate name for such a piece) cannot step out of check. Material they could no longer use to attack the opponent, as making it leave their defensive positions is suicidal. What matters is the number of ways you have to evade check. Whether some of these ways are by moving the royal piece or not doesn't really matter.

E.g. consider a 'Chess variant' with initial position: Flags on a1/h8 Rooks on b1/g8, Pawns on a2/h7. Seems a dead draw to me, you can do nothing but moving the Rooks over the back-rank. with an occasional excursion to protect the Pawn from the side when the opponent attacks it.

Making it possible for the king to be captured as a game-winning condition is significantly simpler than numerous check & checkmate rules.  In effect, it ends the game once move sooner.  Also, making the royal piece incapable of movement would render a game similar to Chess much less drawish.

> the maze of overcomplicated rules associated with checkmate from FIDE Chess that unnecessarily create a wide "draw gulf"

You lost me completely. What 'maze' are you talking about? AFAIK 'checkmate' in FIDE is that your King is up for grabs, and there is no move that alters that. I cannot see anything complicated in that, let alone 'over-complicated'. In fact I hardly could imagine anything simpler.