Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Masters of the Chess: on Chess Capablanca (10 x 8) 40 pieces
1 2 3 4 Previous Next  
Parent - - By GothicChessInventor (**) Date 2007-09-25 20:13
I came up with a set of "balancing combinations" for the 80-square board.

2 Knights > 1 Rook
1 Archbishop > 2 Bishops
Knight + Bishop > Rook + Pawn
Rook  >= Bishop + 2 Pawns
Rook + Bishop > Archbishop + Pawn
Archbishop + Knight > Queen
Queen + Pawn > 2 Rooks
Archbishop + Bishop > 2 Rooks
3 Knights  > Archbishop + Pawn
Chancellor + Pawn > Queen
Chancellor > Archbishop + Pawn
Chancellor + Bishop = Archbishop + Rook
Rook - Bishop = Chancellor - Archbishop

Which of these should be adjusted?
Parent - - By lkaufman (*****) Date 2007-09-25 20:42
Speaking only theoretically, most of these terms look right. However the last term is clearly (to me) wrong, it should be greater than instead of equal. The reason is that a bishop benefits more than a rook by the additional of knight power, as this addition cures the bishop´s weakness of moving on one color only. Similarly the second from the last term should read less than for the same reason. As for chancellor being greater than archbishop plus pawn, this may be true but it is not clearly so. As for rook and bishop vs. archbishop and pawn, again this may be true but is not clear. A term you could add would be queen minus rook = archbishop minus knight, though this might not be exactly true. Finally I would say that in normal chess the bishop pair is big enough to rate a term of its own, and I don´t see why this would not be true in Gothic chess. Naturally adding such a term would decrease the value of a bishop by itself.
Parent - - By GothicChessInventor (**) Date 2007-09-26 03:36
Chancellor + Bishop vs. Archbishop + Rook is an interesting comparison. For one, they are comprised of the same decomposed pieces; Knight + Rook + Bishop are in each group, though fused differently. The second item of  interest is that more often we see a Chancellor + Bishop doing more damage as a "functional metapiece" than we do the Archbishop and Rook.

The Chancellor + Bishop can create what is known as a "Chancellor's Vortex", a situation where the Chancellor can slide or hop, revealing the check of the Bishop, and then checking again with the Chancellor in such a way to force the enemy king to be pinned again after the next  move. This is like the famous "windmill check" Torre was able to inflict on Lasker in Moscow, 1925. In the case of the Chancellor's Vortex, just about every piece could come off the board. It is this formation that gave the name to my program, "Gothic Vortex."

We don't see the Archbishop and Rook combining in this fashion. One reason is, the Rook would need to be the revealed piece with the Archbishop doing the active revelations. Usually a King does not allow himself to be in the same file as an unexposed-but-exposable Rook.

For this reason, I always thought Chancellor + Bishop should be given a "bonus" since the potential exists to inflict greater damage.
Parent - By lkaufman (*****) Date 2007-09-26 18:13
In principle, the archbishop plus rook should be the better pair, because this pair does not have the one color restriction that the other pair suffers from. I haven´t played Gothic enough to comment on specific maneuvers like you are quoting, but my experience with normal computer chess suggests that general factors like the one I quote are more important iusually than specific themes like the one you quote.
Parent - - By neoliminal (*) Date 2007-09-26 01:49
I agree with Larry and I'd add:

Queen + Pawn > 2 Rooks


Queen + Pawn = 2 Rooks
Parent - - By GothicChessInventor (**) Date 2007-09-26 03:23
How can you agree with Larry when he says Queen - Rook = Archbishop - Knight when you have Queen = Archbishop? You obviously don't agree with that.
Parent - By neoliminal (*) Date 2007-09-26 16:07

You have shown me the error of my calculations and I am recalibrating my system to see where the error lays.  For time being I accept the values being show by all parties until I can tweak my evaluation system.  The error in my system probably relates directly to non-evaluation of movement, which is why knight related pieces are scaled too high.  This was pointed out to me in a private message on this board.

So looking over your evaluations I'm inclined to agree with the exception I noted... all of this on consideration that I may change my mind when I reevaluate my system.
Parent - - By lkaufman (*****) Date 2007-09-26 18:10
In normal chess queen plus pawn is on average much better than two rooks, though not in reduced endgames. I don´t see why this would be much different in Gothic.
Parent - By neoliminal (*) Date 2007-09-27 00:12 Edited 2007-09-27 00:18
I was thinking in terms of end game.  My thinking is that in a wide board, Rooks are more valuable (although Ed contests this).  If I'm right, then the queen is also partially increased, but not for her bishop move.  This would reduce her slightly against the other Rook.  But now thinking on it, the increase in a Rook's value vs. a Queen is not a full

Queen + Pawn> Rook + Rook.
Parent - - By GothicChessInventor (**) Date 2007-09-28 03:53
I published a new spreadsheet for comparing the values:

It compares and contrasts the values from 8 different Authors who have discussed their 10x8 computations somewhere on the net.

I computed the averages for the "differences" of 14 different hypothetical trades, then computed the standard deviations. This allowed me to "rank" from lowest to highest for which piece trades there is the most diagreement.

The most harmonious was for Knight + Bishop vs. Rook + Pawn with about a 0.14 standard deviation. This outranked even Rook vs. 2 Bishops, which was surprising.

The most disagreement seems to be over Archbishop vs. 2 Bishops, but this result might be skewed by the Lewis value of Archbishop = 9.0. This category had a 1.22 pawn standard deviation, which is rather large.

Download and enjoy. If you want to see other comparisons, let me know, I will add them, then re-rank the "Agreement Factor" (although this formula will do this automatically.)
Parent - - By lkaufman (*****) Date 2007-09-28 04:13
Regarding my own values listed on the spreadsheet, I believe that I was giving my estimates of what the values of the ¨special¨pieces would be if we ASSUME the standard 1-3-3-5-9 values for the normal chess men. Actually I consider these values pretty inaccurate for normal chess and perhaps even more so for Gothic, so these values do not actually reflect my opinion of the real values of all the chessmen in Gothic chess. Just for starters, if there is no separate bishop pair term, then the bishop is clearly worth more than the knight in normal or Gothic chess.
Parent - - By GothicChessInventor (**) Date 2007-09-28 23:38 Edited 2007-09-28 23:41
I think most people confuse "piece values" for "evaluation function". Most programs do have things like a bonus for the Bishop pair, or a mobility function to determine at least the sliding freedom of Bishops (since they are most susceptible to pawn chains encroaching from a distance), or a halo region surrounding the King to determine safety, etc.

What I mean by "piece values" is strictly that portion of what is fed to the Static Exchange Evaluator to "order" the way in which trades should logically be considered.

This can be the pawn = 1, knight = bishop = 3, etc. "traditional numbers", or it can be very granular with decimal point precision (although almost everyone uses a "centipawn" implementation that is rounded by multiplying by 100.)

That having been said, if anyone wants to change their numbers that appear in the spreadsheet, or if they want to sumbit new numbers, let me know, and I will roll out a new version of the spreadsheet.

As a footnote, I noticed that my Queen - Rook value exactly equals Archbishop - Knight, a comparison suggested by Larry.
Parent - - By lkaufman (*****) Date 2007-09-29 03:56
okay, I will make a stab at it. I will keep the knight at 3, since it is worth somewhat more than this in normal chess but perhaps is worth a bit less on the wide board of Gothic. I will call the bishop 3.5, since the average value of the bishop (paired or unpaired) is somewhat higher than the knight in normal chess (though this is mostly only true of paired bishops), and in Gothic the average bishop has roughly an extra square of mobility on an empty board. I call the rook 5.5, since the rook-bishop spread is a bit under 2 in normal chess and should be slightly higher in Gothic as the rook gains more mobility than the bishop on the larger board. In normal chess, the queen is worth about rook plus bishop plus 1.5 pawns, and I see no reason that this would be much different in Gothic, so queen = 10.5. Now the queen´s 1.5 bonus consists of two parts, a bonus for combined power and a bonus for curing the bishop´s confinement to one color. I will say that 1 point is for combined power and 0.5 for curing confinement (one may argue with this division of the 1.5). Then the archbishop would get the same 1.5 bonus for a value of 8, and the chancellor would get only a 1 point bonus for a total of 9.5. So these would be my values, derived only from chess experience with modifications for the board size.
Parent - - By GothicChessInventor (**) Date 2007-09-30 04:43
There is a new copy of the spreasheet here:

This has the new values in place.

Now among the 8 authors, Knight + Bishop vs. Rook + Pawn is the comparison most in agreement, and Archbishop vs. 2 Bishops is the trade that is least in agreement.
Parent - By Octopus (**) Date 2007-09-30 09:41
I suggest to experimentally normalize the piece values that way, that a Knight is always valued exactly 3. That seems to work out the differences much better.
Parent - - By GothicChessInventor (**) Date 2007-09-30 20:27
I was playing around with the spreadsheet where I looked at:

queen + archbishop vs. chancellor + 2 bishops + 2 pawns

The Kaufman set has this at exactly equal, but I would think the side with the chancellor might be doing better.

Any comments on this trade? It is not easy to come up with  piece values that work all the time, but I like trying to figure out values that better fit the various trade scenarios.
Parent - By lkaufman (*****) Date 2007-09-30 22:03
I have no idea if you are right about this trade, but if you are it might mean that the archbishop is slightly overvalued on my scheme (maybe by a quarter point) because the 1 point bonus for combined power of the component pieces of the queen and chancellor should be a bit less for the archbishop since the pieces being combined are somewhat weaker. If you find clearer examples where the archbishop side of some imbalance seems to be overvalued then I would accept reducing it from 8 to 7.75.
Parent - - By h.g.muller (****) Date 2008-04-21 10:46
Wow, this is a very interesting discussion, to which google brought me. I never visited this forum before, because I thought it was only about Rybka.

I made an attempt to determine empirical (opening) piece values for Capablanca-type Chess, by self-play of my 10x8 engine (Joker80) from shuffled (FRC-like, but with normal castling) opening positions where some pieces were deleted to create a material imbalance. I played nearly 20,000 40/1' games on a 2.4GHz Core 2 Duo to collect data with good enough statistics. (For the full list, see The Pawn-odds score is about 62% under these conditions (so a quarter Pawn would result in a 3% advantage).

The score percentages from the various imbalances are best fitted by the following piece values:

P = 1
N = 3.5
B = 4.5 (+0.5 for pair)
R = 5.5
A = 10.25
C = 10.5
Q = 11

Or, when normalized to Q=950 to bring it more in line with other reported vaues (I like to normalize on Q, because the Q value is the least variable piece with respect to position on the board, while the value of Knights and Bishops is strongly affected by positional patterns).

P = 85
N = 300
B = 350 (+40 for pair)
R = 475
A = 875
C = 900
Q = 950

(The values were rounded to quarter Pawns, as the required rounding never involved more than 10 cP, and I really cannot claim that my statistical error is much smaller than that anyway.) Note that the Pawn value was that of g2/g7, which in Capablanca Chess has a similar role as the f-Pawn in normal Chess. So not a particularly valuable Pawn, but also minimal compensation when it is missing. As the measurements were done from the opening, the Rook value obtained this way might be an under-estimate, and that it goes up in later game phases. One way to incorporate this, would be to say that this is the value of the Rook without any open-file bonus, and that later in the game he Rooks can almost always be developed to an open file, (in the late end-game any file is open!), upping their value by 25cP or more.

Note that this data set fully corroborates lkaufman's intuitive feeling that most people strongly underestimate the value of A. The Archbishop is a devastating and illusive piece. Observing how the engines wield it, I really got the impression that the name 'Dancer' would describe this piece much better.

Another surprise was the rather large (50cP) gap between B and N, on top of the pair advantage, and that this is mostly caused by a stronger Bishop, rather than a weaker Knight.
Parent - - By Vasik Rajlich (Silver) Date 2008-04-22 23:33

quick question about your methodology.

> Q-BNN    (172+ 186- 75=) 48.4%
> Q-BBN    (143+ 235- 54=) 39.4%
> C-BNN    (130+ 231- 71=) 38.3%

How did you "randomize" the 400 games from each position?

Note also that your method gives you values of the pieces in the opening position. These values may not be appropriate for other types of positions (for example, endgames). There is no really perfect way to do this.

Parent - - By h.g.muller (****) Date 2008-04-23 09:52
For one, the engine used (Joker80) randomizes its play during the entire game, by adding a quasi-random value to each evaluation (derived from hash key and starting time of the game, so the same position will receive the same random contribution consistently).

To make absolutely sure that the games do not resemble each other too much during the first few moves, I play all games from a different CRC-like opening setup. I do not touch the King and Rooks, (so that normal castling rules can apply), and shuffle the other pieces with the restriction that there is one B and one N on either side of the King, and the Bishops are on different colors. On a 10x8 board, this leaves 216 setups. Each setup is played with white and with black to move first, so that makes 432 different game starts. So an entry like 'Q-BNN' is really 432 different positions. Opening positions that were different in general do not become the same when I delete pieces: you can still see what type of piece was supposed to start at a certain vacant square because the opponent still has it. For the Q-BNN test I could even have created two variations on each position (for 864 different games) by deciding which Bishop to delete. (On 8x8 the number of different opening positions you can generate this way is far smaller, and it would probably be better to play with FRC castling and shuffle K and R as well to create diversity.)

Indeed, the method measures opening values, with as main effect that it tends to underestimate Rooks. In principle it could be applied to other game stages as well, though, if one can devise a good way to generate randomized non-tactical positions for that phase. I only tried this once, to test if the anomalously high Archbishop value was game-stage dependent. To this end I started from the position

which seemed non-tactical enough, and relied on the engine randomization to create diversity. I watched a couple of dozen games to make sure that they did not tend to develop along the same forced path initially. This might not be an entirely sound method (I would like to average over more initial pawn structures, perhaps also including passers (equal numbers of passers for both sides), but at the time I was only interested to see if their was a gross devaluation of the Archbishop as the board got more empty (as some players had suggested to me). The self-play test confirmed that the above position is more or less balanced, though, showing that constant piece values during the game is not an extremely flawed approach. The increase of the Rook value towards later game stages is covered to a large extend by an open-file bonus.
Parent - - By Vasik Rajlich (Silver) Date 2008-04-24 08:20
Ok, thanks for the description.

We've done similar studies to this (for chess, of course). Instead of shuffling the pieces on the back rank and also randomizing the eval, we use the following randomization scheme: when a root position is encountered during a game which had previously been enountered (probably during a previous game), we determine the frequency with which all root moves have previously been played. The frequencies will be X, X-1, and 0. Moves which have been played X times are penalized by 8 centipawns - their searches are done as follows:

int val = search (-beta - 8, ...) - 8;

instead of

int val = search (-beta, ...);

In general, we find that this leads to quite a few biases. In quite a few cases, the results are a little bit unusual (compared to chess theory) and can be improved by head-to-head games between competing values.

Note also that your shuffling scheme may be ok for testing material weights but will probably not be very appropriate for testing other features.

Anyway, I think the methodology question is crucial. You solve that and you solve everything.

Parent - By h.g.muller (****) Date 2008-04-24 09:47
I use a similar trick to randomize in my other engine, Fairy-Max, which I use to determine the value of more exotic piece designs (as I can simply define the pieces there in the .ini file, without recompiling). There I add a quite large (+/- 25cP)  random value in the root score of the first 3 moves, to create diversity, (where the clock-derived random replaces the 8 in your posted statement). That random is large enough to effectively make Fairy-Max try all opening moves that do not lead to outright material loss. (Except that King moves receive such a large eval penalty that they are almost never chosen.)
Parent - - By tomski1981 (*****) Date 2008-04-23 00:49
what values would you give to normal chess men in normal chess? (if not 1-3-3-5-9)
Parent - - By lkaufman (*****) Date 2008-04-23 02:17
My published values are, with pawn at 1, knight and bishop each 3 1/4, bishop pair 1/2, rook 5, and queen 9 1/2 (or 9 3/4 if you go by statistics and not by my personal opinion).
Parent - - By BB (****) Date 2008-04-25 22:06
Soltis turned some of the considerations with the LK values into a book Rethinking the Chess Pieces. See here for a review by Heisman, and this PDF by Garcia for the ChessCafe review.
Parent - By lkaufman (*****) Date 2008-04-25 23:38
Of all the grandmasters, Soltis is the one I have played the most tournament games with, although most were before he become a GM. Our first eight games were all draws, although not without a fight. I believe he won the ninth encounter.
Parent - By h.g.muller (****) Date 2008-04-23 10:26 Edited 2008-04-23 10:29
I have not systematically applied my method to normal Chess. I only did some occasional measurements for comparison with 10x8 results that seemed different from what they should be based on 8x8 lore. E.g. BB-NN for an anti-pair of Bishops (= on like color, to avoid a pair bonus). This confirmed base value of B and N is identical in normal Chess (I did not carry it to the resolution needed to see the difference between having 5 Pawns or 8 Pawns per side on the board), while in Capablanca Chess the Bishop anti-pair beats the Knights by ~62%, and you need BB-NNP to make a balanced position.

Bishops on different color versus Bishops on same color (by swapping a Knight and Bishop on the same side of the King for one side) turns out to be a good way to measure the pair bonus. In 8x8 Chess this produced about half a Pawn. So the few tests I did all confirmed the 'Kaufman values'.

Getting a good Pawn value is always a problem, though, as the Pawn value is extremely variable and non-additive. Pawn-structure evaluation is a difficult issue, and in deleting a Pawn, one really should value it as the difference of the evaluation for the entire Pawns structure before and after deletion. So it is very difficult to understand what kind of Pawn the observed Pawn value corresponds to. Even consensus evaluation of extreme Pawn structures usually is totally off. Just for fun I tried deleting a Queen versus 8 Pawns from the opening, so white played entirely without Pawns. So black has 8 connected passers! Well, the Pawns are toast! Of 20 games I tried, there was only one draw, (19+, 0-, 1=) due to the side with the Queen going for an early perpetual, as the evaluation was overly scared for the 8 connected passers, and I had not set a contempt factor. (This was with Joker, though. Perhaps Rybka would do a much better job in evaluating the initial position
rnbqkbnr/8/8/8/8/8/PPPPPPPP/RNB1KBNR w KQkq - 0 1
Parent - By neoliminal (*) Date 2007-09-29 19:56
For the record I no longer stand behind the value of 9 for the Archbishop.  As noted elsewhere, my calculations for both the Archbishop and the Knight are likely overvalued.  If I had to guess at a value I would put the number at 7 (rounded down).  The knight likely staying at it's original value (3 rounded up).
Parent - By GothicChessInventor (**) Date 2007-10-08 07:05
Some people have been "clamoring" for a Zillions release of Gothic Chess. I'm not sure why, Zillions is a sub C-class engine.

For those that have Zillions-Of-Games, I am releasing an official Gothic Chess zrf implementation:

Place the .zrf file into the "Rules" folder of your Zillions directory, and place the entire "gothicchess" folder into your "Images" directory.

Launch Zillions, then open the GothicChess.zrf rules file, and you can play Gothic Chess using Zillions.
Parent - By Octopus (**) Date 2007-09-23 16:13
There have been some other people, proposing me a value model for the Capablanca piece set. For some of those proposals I had made a special SMIRF compile for to find out, how those models would behave. In fact this is only an approach to correctly implement another value model. Nevertheless from corresponding testgames of such implemented different models there often could be found positive and negative aspects of those models. If you would have another consistently derived model, why not have such test games then between different SMIRF compilates as a first evaluation approach?
Parent - - By neoliminal (*) Date 2007-09-21 01:06
Has anyone seen a version based on this board configuration?

      a   b   c   d   e   f   g   h   i   j
    +---+---+---+---+---+---+---+---+---+---+    Black
  8 |*Q*|*R*|*N*|*B*|*A*|*K*|*B*|*N*|*R*|*C*|  8
  7 |*P*|*P*|*P*|*P*|*P*|*P*|*P*|*P*|*P*|*P*|  7
  6 |:::|   |:::|   |:::|   |:::|   |:::|   |  6
  5 |   |:::|   |:::|   |:::|   |:::|   |:::|  5
  4 |:::|   |:::|   |:::|   |:::|   |:::|   |  4
  3 |   |:::|   |:::|   |:::|   |:::|   |:::|  3
  2 |:P:| P |:P:| P |:P:| P |:P:| P |:P:| P |  2
  1 | Q |:R:| N |:B:| A |:K:| B |:N:| R |:C:|  1
    +---+---+---+---+---+---+---+---+---+---+    White
      a   b   c   d   e   f   g   h   i   j

The thing I like about this configuration is that the more powerful pieces are put to the outside.

Other factors covered by Casablanca variants:

1. Every Pawn is protected.
2. Normal Chess set up is maintain more than any other version I've seen.  (only one piece is different...)
3. The King is now flanked by a *diagonal* mover, so that pushing the kings pawn opens up two pieces... something other variants have missed.
4. Castling with the rook works as in normal chess... no need to learn new castling rules.

What do you guys think?
Parent - - By Octopus (**) Date 2007-09-21 02:07 Edited 2007-09-21 02:10
You could create, store and replay such a position with the SMIRF FullChess GUI.
Queen and Archbishop are standing on equal coloured squares, which prevents,
that this starting array could be regarded as one of those many CRC positions.
Exchanging Q and C or A and K would help to change this.

Nevertheless it is looking a little bit strange at the first sight. More investigations
might be needed to judge on that array (X-FEN string of that position):

qrnbakbnrc/pppppppppp/10/10/10/10/PPPPPPPPPP/QRNBAKBNRC w KQkq - 0 1
Parent - - By GothicChessInventor (**) Date 2007-09-21 02:26
1. Nd3 is practically a forced white win since e5 is surrendered by black and can never be recovered without incurring significant weaknesses, especially with 2. f4 being "unstoppable." The Bishops are on the wrong colors as their 8x8 chess counterparts, they can't fianchetto, so what's the point of not being able to contest control on the primary long diagonals?  The two strongest pieces are furthest from the action, and the knights can impede the knightlike components of the first move of the Archbishop and Chancellor on each side of the board. 1. d4 is weak since it hems in the Bishop on g1 and 1. e4 almost loses by force due to overexposure of the king since black has 3 pieces on the light diagonal (Queen on a8, Archbishop on e8, Bishop on g8).

So, we have a game with limited opening scope for white but a forced white win, with no similarity to chess, and no appeal what so ever.

Congratulations on discovering a pathetic configuration.
Parent - By neoliminal (*) Date 2007-09-21 17:44
1. Nd3 is practically a forced white win since e5 is surrendered by black and can never be recovered without incurring significant weaknesses, especially with 2. f4 being "unstoppable."

I don't agree.  This looks very much like a normal chess opening and is countered in the same way.  Perhaps if you take your example further...?  My response to Nd3 would be a mirrored opening of the black knight.

2. The Bishops are on the wrong colors as their 8x8 chess counterparts, they can't fianchetto, so what's the point of not being able to contest control on the primary long diagonals?

Actually they ARE the the right colours.   You are look at Octopus' layout.  On the original ascii board you can clearly see you are wrong.  They fianchetto just fine for "normal" chess, in the center.

The two strongest pieces are furthest from the action, and the knights can impede the knight-like components of the first move of the Archbishop and Chancellor on each side of the board.

This is in line with the history of Chess.  With the exception of the "Mad Queen" which was accepted into play in the 1500's, the pieces were arrayed from weakest to strongest.  Minor pieces are supposed to suppress higher pieces at the start of the game.

1. d4 is weak since it hems in the Bishop on g1 and 1. e4 almost loses by force due to overexposure of the king since black has 3 pieces on the light diagonal (Queen on a8, Archbishop on e8, Bishop on g8).

1.d4 has the same ramifications as those in normal chess.  After taking Octopus'  advice and flipping the Queen/Chancellor there is no longer and triumvirate on the light diagonals.

So, we have a game with limited opening scope for white but a forced white win, with no similarity to chess, and no appeal what so ever.

Congratulations on discovering a pathetic configuration.

Thanks for your input Ed.

Anyone else care to comment?
Parent - - By neoliminal (*) Date 2007-09-21 17:27 Edited 2007-09-21 17:33
i agree with octopus.  Flip the Chancellor and the Queen is in order.

      a   b   c   d   e   f   g   h   i   j
    +---+---+---+---+---+---+---+---+---+---+    Black
  8 |*C*|*R*|*N*|*B*|*A*|*K*|*B*|*N*|*R*|*Q*|  8
  7 |*P*|*P*|*P*|*P*|*P*|*P*|*P*|*P*|*P*|*P*|  7
  6 |:::|   |:::|   |:::|   |:::|   |:::|   |  6
  5 |   |:::|   |:::|   |:::|   |:::|   |:::|  5
  4 |:::|   |:::|   |:::|   |:::|   |:::|   |  4
  3 |   |:::|   |:::|   |:::|   |:::|   |:::|  3
  2 |:P:| P |:P:| P |:P:| P |:P:| P |:P:| P |  2
  1 | C |:R:| N |:B:| A |:K:| B |:N:| R |:Q:|  1
    +---+---+---+---+---+---+---+---+---+---+    White
      a   b   c   d   e   f   g   h   i   j
Parent - - By neoliminal (*) Date 2007-10-01 19:11
It appears, after some research, that this particular set up was already invented.  See for more information.  Funny how few good combinations there really are.
Parent - By neoliminal (*) Date 2007-11-30 02:51
The link, if anyone is interested, is Victorian Chess.
Parent - - By JhorAVi (***) Date 2007-09-21 05:50
If a new variant should replace normal chess, it's very important that the initial position should be similar to chess or else the transition wont work.  Likely candidates are Gothic and Embassy configurations.
Parent - - By neoliminal (*) Date 2007-09-22 02:18
This new variant on Capablanca has almost exactly the same set up as regular chess.  The center contains a normal chess set up with the exception of the Queen being moved and replaced with an Archbishop.  You can't get much more similar.  Certainly this is more similar than Gothic or Embassy Chess to regular chess.
Parent - - By GothicChessInventor (**) Date 2007-09-22 07:38
The Center Counter Defence is out the window in that variant. And 1. f4 2. Af2 already threatens to win a pawn with Axa7.

The whole thing is unplayable for those reasons alone.
Parent - - By neoliminal (*) Date 2007-09-22 22:32
Ed, you're really losing your touch here.  That pawn is protected by a knight. If you really want to trade an Archbishop and Bishop for Knight... 

I think you'll throw up any excuse now. You're not even trying.
Parent - - By GothicChessInventor (**) Date 2007-09-23 08:33
threatens.... as in makes the feigning of... as in, unless you park your knight there all game, you will lose it.
Parent - By neoliminal (*) Date 2007-09-24 19:56
How many of these can I find in Gothic Chess? I can threaten pawns from the opening.   What's your point here? 

If you really think the threatening of the a7 pawn (which is trivially resolved by moving it) makes the game flawed beyond repair I think you're really grasping at straws.  Do you really feel this layout is a threat to Gothic Chess?  That's the only reason I can think you might even be interested in this thread.... why are you even talking about this?

You seem so reactionary.
Parent - - By GothicChessInventor (**) Date 2007-09-22 18:33
An interview of me appears now on a French online website in both English and French:

Parent - By saxon (**) Date 2008-04-21 13:02
Interesting strategic board game that "Gothic chess".
However,I dislike introducing new "strange" pieces ,asimetry of the board,or even comparisons with normal chess.
I would think next chess variants to be explored by the players is chess960 ,and Capa chess (10×10).The latter will be explored by computers in the future more than played by humans.10×10 chess is extremely physicaly demanding variant of a game for humans.With normal rate of  play a typical game lasts over 10 hours.While with faster time controls quality of the game degrades.That's not very good.    
Up Topic Rybka Support & Discussion / Rybka Discussion / Masters of the Chess: on Chess Capablanca (10 x 8) 40 pieces
1 2 3 4 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill