[ List Latest Comments Only For Pages | Games | Rated Pages | Rated Games | Subjects of Discussion ]
Comments/Ratings for a Single Item
Back in the Big-board CV:s thread, I also had trouble when clicking on Next 25 item(s). I figured out how to make links like these: skipfirst=25, skipfirst=50, skipfirst=75. Also I started a Very Large CVs thread, for discussion topics related to Very Large CVs.
Here is another Piece Values thread from environs of 2004, but it is hard to read before its most recent 25 Comments, because 26-50 and 51 and over get lost in the indexing.
There is also a Positional Advantage Equation, to go with the Move Equation, both of which I am incorporating into an article to submit, following Mark Thompson's suggestion. There will be rigorous definitions and supporting examples applied to specific sets of rules. We used this thread at will mostly a year ago to test ideas for formulaic evaluation of CVs.
Rest assured, I am interested in and supportive of the effort to define a general mathematical formula for determining the average length of chess variant games ... if possible. However, I must echo Thompson in insisting that the persons responsible 'show their work' and publish it (without clutter) upon a seperate web page. A complete, step-by-step presentation and definition of each term in the calculation is needed as well as a logical, conceptual explanation of the indispensible nature of each term within it. It needs to be evaluated for fundamental validity and possibly, revised. I suspect the efforts to date are incomplete, inaccurate or conceptually flawed since I cannot rationally imagine what mathematical formula can predict or dictate the level of aggression freely chosen by both players and hence, the actual length of a game (measured in moves) with any accuracy or even within a strict range from minimum to maximum moves. Although I think an optimum, average level of aggression exists in theory and is somehow definable by formula, specific to a given chess variant, for rational, incisive play, I am certain that the rules of virtually every chess variant do not enforce its use upon its players in any way. Even if a valid, crude formula has been successfully produced by Smith and Duke, every chess variant will need a positive or negative adjustment, significantly sizeable in some cases, due to its opening setup. [Some stable opening setups are highly buffered; some stable opening setups are hair-triggered]. Furthermore, game-specific calculations focused upon trapping royal pieces with different, likely amounts of material are indispensible to make any estimate of the endgame length for various games. If I misunderstand in expecting a mere, useful estimate to be more rigorous than ever intended, I apologize.
The mathematical formula I worked out a year ago for M(=#Moves) helps explain the flatness of play in Medieval Chess in Game Courier. It simply can be expected to have a large number of turns on average for its 76 squares. Building on Smith's Exchange Gradient, #M = 3.5N/(P(1-G)), with P Power Density and G calculated as (PV1/PV2 + PV1/PV3...+ PV1/PVn + PV2/PV3...+ PV2/PVn...+ PV(n-1)/PVn))/(n(n-1)/2). That gives Gradient, but we want (1-G) for right directionality. For Medieval with Q9, P1, R5, and excluding K all the other pieces 3 points, G is 0.614, very high, representing not much benefit in exchanges. Plugged in above, it translates to predicted long-term average of 62 moves, long games for 76 squares. Contrast that to Orthodox Chess(64sq) Design Analysis giving just ave. 34 #M and Capablanca(80sq) ave. 38 #Moves in Comments there.
A smaller case that demonstrates the effect of many pieces with the same value is
<a href='../index/listcomments.php?subjectid=Rook-Level+Chess'>Rook-Level Chess</a>, which despite more power on the board, is a flatter, less interesting game than FIDE Chess.
Medieval Chess played in GC now is perfect example of Larry Smith's 'advantage in exchange'. In both GC games there has been one piece exchange so far after close to 30 moves[a 3rd game, zero]. Four pieces are of about the same value: Knight, Longbowman, Seer, Swordsman. 'If a game were populated with pieces of near equal value, the advantage of the exchange might not be significant.' --Smith in this thread. Few sacrifices suggest themselves for positional advantage; Medieval Ch. is from its onset like Orthodox FIDE Chess in rewarding caution.
To resurrect a discussion line and continue the topic of pattern pieces: In those games which have promote-able 'Pawns' restricted to pieces which have been previously captured, pattern pieces can offer a further restriction. If the game contains pieces bound to specific patterns, such promotions could be limited when promoting to these. In other words, if a player has lost a Bishop and brought a Pawn into the promotion zone, the promotion to this captured Bishop could be predicated on whether there presently exists another Bishop within that specific diagonal pattern. And with those pattern pieces which do not occupy every one of their specific patterns, a Pawn might be denied promotion to that particular piece unless it was in the necessary pattern. These rules would be at the discretion of the developer, and could impact the over-all strategy of the game.
It appears that we've had spill-over from another discussion. But to continue with the use of pattern pieces in Game Design. The only problem with such pieces is the possible end-game scenarios. This can be solved by the developer with the creation of particular rules to handle this. What if both players reach the point that they only have these pattern pieces and no possible way of threatening either goal piece? Most would call this a draw, XiangQi does. But another idea would be to include these pieces in a condition for a win. Example: If the game is reduced to such pattern pieces and goal pieces, the player with the majority of pieces could win. Thus creating the secondary goal of capturing the opponent's pattern pieces.
Hmm... hey! I want to see that! I would have chosen some other actresses, but Fergus's are mainly good (at least 2/3)
Your comments about the Alfil and Dabbabah remind me of the Dragon in British Chess. This piece is a compound Alfilrider and Dabbabahrider. So, like the Dabbabah, it is limited to only one quarter of the board. Each player gets two Dragons, which are enough to cover only half the board, and the four initial Dragons in the game each cover a different quarter of the board. The only way for two Dragons to cover the same area would be through Pawn promotion to a Dragon. But since the only way a Pawn may promote to a Dragon is if one has been captured, no player will ever have more than two Dragons. Despite the fact that a player will never be able to cover the whole board with his Dragons, I don't think the game suffers from giving each player only two Dragons instead of four. The Dragon is useful mainly in support of other pieces. Also, given that a player's Dragons cannot capture each other, there is a greater potential for uneven piece exchanges, which may help to make the game more interesting.
Like the Bishop, there are other pieces which occupy specific patterns on a square playing field. For example, the Alfil and the Dabbabah. The first leaps to the second diagonal and the other leaps to the second orthgonal. It would take four distinct Dabbabah to occupy each of its patterns, and eight Alfil of its. But this is not entirely necessary. A developer may choose specific patterns for each of these pieces to influence and thus encourage particular tactical behaviour during play. Sacrificing or avoiding the risk of pieces on those patterns during play can make interesting strategy. Allowing each player to control particular patterns will give them both similar advantage, just seperate. A good example of pattern play is in XiangQi. The Elephants in this game are restricted to a limited portion of the field and yet they are significant during the game. Being able to properly use these Elephants can often determine the outcome of the game. In several Shogi variants, there are also strong pattern pieces. For example, the Capricorn which preforms a diagonal hook move. Usually this piece occupies a specific pattern at set-up, when captured it is permanently removed and can only be recoverd by the promotion of another specific piece on the field.
Also, the Bishop in Shogi can promote to the Dragon Horse and gain the ability to step one orthogonal. Thus being able to shift diagonal patterns. And to continue the potential of inner game dynamics. Most FIDE-style games allow for Pawns to promote to Bishops. Thus creating the potential of Bishops on either diagonal pattern. So, the initial set-up of the Bishop is not the sole determination of any game. And it actually can create definite strategic dynamics. So a game most be evaluated in its full potential and not just its initial set-up. What if a game has a Bishop on a single pattern and there is never the potential of a Bishop on the other? Does this, in itself, negate the value of the game?
One significant difference between Shogi and Chess is that the Bishop in Shogi can change color, so to speak, by being captured and then dropped. It is also possible in Shogi for a player to possess both Bishops. So, the drop rules of Shogi are making up for the imbalance created by each side beginning with only one Bishop. If Shogi were played without drops, it would be a significantly less balanced game than it is with drops.
Let me deviate a little and discuss the concept of balance in Game Design. Most would assume that a perfectly balanced game is the optimal, and this is often demonstrated by comments about the placement of Bishops (long diagonal movers) in games. In a square playing field, there are two distinct diagonal patterns, and FIDE has offered a Bishop for each of these. But in Shogi initially the Bishops occupy only one of these patterns. Both games are considered good. Whether or not a game has Bishops occupying each diagonal patterns is not the sole foundation for its evaluation. In fact such imbalances can be considered a potential factor in the overall strategic dynamic of the game. Both diagonal patterns can be occupied, one diagonal pattern can be occupied or opposing diagonal patterns can be occupied, the game will still have the potential of being good. In fact, there could be no Bishops in a game, like XiangQi(excluding its Elephants). 'Now now, perfectly symmetrical violence never solved anything.' ----Professor Hubert Farnsworth, Futurama, The Farnsworth Parabox
Seems like this idea of formulaic evaluation of CV's should be written up on a page of its own. A thorough investigation of how the various popular CV's fare under different formulas, and hence of how the formulas ought to be interpreted, would take a lot more exposition than could be done in comments. The challenge is to come up with formulas that will not only 'predict the past', by telling us what we expect them to tell us about well-known variants, but that will also provide useful insights into new games. It's far from obvious that such formulas could be found, but it would be quite a discovery if they were.
Note that M = 3.5ZT/P(1-G) is useful form of Move Equation because T, piece-type density, will figure in the Positional-advantage Potential Equation, yet to be posted. Use of T, piece-type density, in both enables other comparisons later. Actually, of course, for Game Length, #M = 3.5N/P(1-G), N simply number of piece-types, is all that is necessary, eliminating Z Board Size from numerator. Z still contributes to determination of Power Density. So, original equation reduces to M = 3.5N/P(1-G)
In the recent long comment, Antoine Fourriere names 7 CVs I believe in first paragraph, and seven more through article, only two of his own 'portfolio'(both which I rated Excellent), the rest I suppose from his 'repertory'. Another mind might list a different 7 as standard, or as formative. Not everyone uses Shogi, for ex., as model for western CVs. Still another team may have 7 more, theme-based perhaps, another 7 violent games, and so on to another group with 70 micro-regional-based, 700 small CVs, 7000 larger variants, 70,000 more sacrosanct to some. What way out except to begin to have design analysis criteria? Or, historicocritically, as Vladimar Lenin says, 'What Is To Be Done?'
In Bigamous Cavalier Chess, I did not use a 9x9 board, because the Nightriders would be attacking the back rank, and the solutions for fixing this caused problems of their own. If I stopped this by moving the Cavaliers up one rank, both sets of Cavaliers could immediately move to the 5th rank. In the initial position, a Cavalier could move forward only to the 5th rank. Thus, the first Cavalier to move forward would be moving to a space where it could be immediately captured by an enemy Cavalier. This could result in a quick exchange of Cavaliers, which would undermine the reason I chose Cavaliers over Knights in the first place. I chose Cavaliers (aka Chinese Chess Knights) for their ability to block each other, sort of like Pawns can block each other. To make this more feasible in the opening, I needed at least four empty ranks between the Cavaliers. If Cavaliers started on their player's 3rd ranks to prevent Nightriders from reaching the back rank on a 9x9 board, they would have only three empty ranks between them. Compromises that put some Cavaliers on the 2nd rank and some of the 3rd did not work out well either. Using a 9x10 board eliminated all the problems caused by a 9x9 board without introducing any new problems. I did not include an Amazon for the same reason I never included one in Cavalier Chess. This piece to too powerful, resulting in a less interesting game. I don't like to include any piece that is so powerful, it can force checkmate on its own. It makes the other pieces superfluous. I find a Chess variant more interesting when it involves the strategic marshalling of a variety of forces, and I don't like games where the main strategy is to get one super piece into a position where it can proceed to force checkmate. That's why I hate Frank Maus's Cavalry Chess.
Fergus, In the new Bigamous Cavalier Chess, why did you decide to use a 9x10 playing field? Why not the 9x9? Also, why the Queen and not the Amazon? You may have covered these topics before. Just a few questions that might help the interested see what goes into some of the decision process of Game Design.
As an experiment, I made a preset for a version of Cavalier Chess with an extra Queen. I doubt it is an improvement. But we shall see. Paladins begin on the same color squares, but that's not the problem it would be for Bishops, since Paladins change color with Knight leaps. Here is a link to the preset: /play/pbm/play.php?game%3DBigamous+Cavalier+Chess%26settings%3DMotif
I see no need for adding an extra Queen to Cavalier Chess. The Queen is still the most powerful piece in the game. My only complaint about the game is that it is played in a tight space given the power of the pieces. I fixed this with Grand Cavalier Chess, which I think is the better game.
We may need an Advanced Exchange Gradient, per Antoine Fourriere's method, for some studies, to reflect all individual pieces' value relationships. So far the only formula out of EG is No. of Moves, and for that any imprecision of not counting each piece separately is offset an extent by over-all Power Density and the constant in M = 3.5(Z*T)/(P*(1-G)), keeping this remark brief. I am also working on a variable to reflect Lavieri's cry for measure of positional-advantage potential too.
Regarding George's comment, I'm considering overall strength by piece-type. EG would value the Queen similarly whether there is one, two or eight Queens on the Board. I think one Queen is better for Chess and two Queens would be better for Cavalier Chess, because they better match the overall strengths of 2 Rooks, 8 Pawns, 2 Bishops and 2 Knights in the former case, and of 2 Marshals, 2 Cardinals, 2 Nightriders and 8 Cavaliers in the latter case. On 10x10 or even 12x8 (without a hole), a Bishop is significantly stronger than a Knight -- the Omega Chess pages suggest Q=12, R=6, B=4, C=4, W=4, N=2(.5) -- and a third (Pocket?) Knight would make sense. (Of course, I didn't follow my own advice on ClB, but there were other pieces to drop, and the armies were strong enough, an argument which makes some sense for Cavalier Chess too, but that Queen/Marshall or Queen/Cardinal disparity still bothers me.) A third Nightrider for Cavalier Chess on a 9x8 Board would also be mathematically consistent, but maybe two Nightriders exert enough influence on the nervous systems of the players, like one Coordinator in Ultima/Maxima.
25 comments displayed
Permalink to the exact comments currently displayed.