Comments by crazytom

That is, for one of my games the source has

< SELECT NAME='crazytom_bgimage'

< SELECT NAME='crazytom_bgimage' >

Thanks for adding these new features and taking the time to work out the kinks. As far as my efforts are concerned, I figure the opportunity to change to something other than shogi-simple.png is easily worth looking at a little HTML.

What specifically is it that you think is defined wrongly?

The rule itself is very simple:

The king moves to the c-file and the a-side rook moves to the d-file, or the king moves to the g-file and the h-side rook moves to the f-file.

That's it. One sentence (not including the restrictions on when it is permissible to castle, which are identical in all the rules discussed on this page).

The Chess480 rule, even though it was introduced as 'an appeal for simplicity', is no simpler, and arguably more complicated than the FRC rule.

Of course these are not the only possible rules. If I had been asked, before learning about FRC, how the castling rule should be generalized for random starting posiitions, I probably would have said that the king moves half the distance (rounded up) toward the rook, and the rook moves to the other side of the king. This rule is left-right symmetric and matches the Chess480 rule in 11/16 of the possible positions. But without the need for awkward special cases, it is in my opinion simpler.

I am predisposed to like symmetry, and it wouldn't have occurred to me to choose an asymmetric rule like the one in FRC. Yet there is something appealing about the asymmetry, particularly in this context where it produces twice as many actually distinct positions. For this reason I'm still inclined to prefer the FRC rule.

Yes, a fascinating game, and a victory for the Chess960 champion over the
FIDE champion!

1. When the tournament rules mention 'White' and 'Black', presumably 'White' means the first player (White in Chess, Red in Xiangqi, Black in Shogi) and 'Black' means the second player (Black in Chess and Xiangqi, White in Shogi)?
2. The first round will be a true round robin (meaning everyone plays everyone) only if exactly seven players participate.
3. If my understanding of the Buchholz-Solkoff and Sonneborn-Berger tiebreaks (as explained here) is correct, then this page's description of Buchholz-Solkoff is incorrect, and that of Sonneborn-Berger is incomplete and potentially misleading.

Oh, and sign me up, please.

Antoine's proposal should work very well if the number of players is
prime.  If the number is divisible by 2 or 3, there's a small weirdness. 
Look at the Shogi pairings in the 9-player example: players 1, 4, and 7
play one another; 2, 5, and 8 play one another, and 3, 6, and 9 play one
another.  No one in any of these groups plays Shogi against anyone in
another group.  A similar partitioning into 2 groups would happen with
Xiangqi with an even number of players.  I don't know how much, if at
all, this should bother us.

Here's an interesting possibility, inspired by Fergus's idea of having a
champion for each game.  Perhaps the final round could consist of the top
Chess player, the top Xiangqi player, the top Shogi player, and the top
overall player (and, if some of these should be the same person, the 2nd
overall player, etc.).

I would prefer to allow draws by agreement even in Shogi, although they
should be discouraged except in clearly drawish positions.  It doesn't
seem fair to me to penalize both players for playing equally well just
because the game ran long.  If length is a concern, there must be some set
of carefully chosen time controls that will address the issue.

The problem seems to be that the link from the PBM Game Logs page points to http://play.chessvariants.org/pbmlogs/displaycomment.php?commentid=11172, while the link from the corresponding Comment Listing page points to http://www.chessvariants.org/index/displaycomment.php?commentid=11172.

I won't say much about game selection and tournament structure; I'll play
in just about any event as long as the games are appealing and I have the
time.  All the ideas posted here so far are good, but I hope the
multivariant tournaments with democratically selected games won't go
away; part of the fun of these has been the exposure to games that I might
not have played otherwise.  My only real 'complaint' about the polling
process is that good old Chess doesn't seem to stand much of a chance of
getting through, and I'd like to play it alongside other games.  (Hmmm...
maybe a 'Big Three' chess/xiangqi/shogi tournament?)

I have some thoughts about scheduling, which apply mainly to largish round
robins.  In GCT1 the games were divided into three rounds; this worked, but
led to periods of relative inactivity if a round had one or two very long
games.  For GCT2 it was decided that games would be assigned as players
became available.  This was a good idea in principle, but in order to run
smoothly it seems to require an inordinately large (and prolonged)
time commitment from the director, who must continually check for finished
games and determine which players are available, how many new games can be
assigned, which of the remaining games should be assigned first, etc.

It seems to me that the game-assignment process could be automated.  I'm
thinking of a script which could run periodically (once a day, maybe) and
assign games until it couldn't assign any more without exceeding a
specified maximum number of ongoing games for any player.  The algorithm
to work out which games to assign wouldn't have to be very complex.  The
part that I don't know anything about would be the interface with Game
Courier.  Perhaps Fergus can tell us whether this would be feasible.

Michael,

In Case 2, 4 BCQ pieces must be placed on 4 squares (c1, d1, e1, d2),
giving only 1 combination of filled squares.  The same applies to Case 3. 
Removing this factor of 4 reduces Case 2 to 72 combinations and Case 3 to
144, making the total number 864, which matches the figure I found by
counting in a somewhat different way.

Nova Chess features a great many unusual and interesting pieces, but the
pieces in any given set should be fairly easy to learn, and the game looks
quite playable.  Nevertheless, it will be a while before I've played all
possible scenarios. ;)

I like the new Clodhopper and Fuddy-Duddy pieces in the Smess-style set. I preferred the name Dumbo, though, as it seemed so perfect for a piece based on the elephant. Fuddy-Duddy makes some sense too, but I've known ministers who are anything but dull, conservative, and unimaginative.

I think we need to start teaching topology in elementary schools.

• If the nickel comes up heads, we believe the written rules, which have consistently indicated that Rococo's Long Leaper moves and captures in the same way as Ultima's Long Leaper.
• If the nickel comes up tails, we believe the testimony of the game's inventor, Peter Aronson, who has consistently indicated that Rococo's Long Leaper moves and captures in the same way as Ultima's Long Leaper.
• If the nickel balances on its edge, then we'll have to think of something else. Pistols at dawn?

The link on this page to the GCT1 game logs appears to have an incorrect tournament filter, and produces an empty list of logs.

Here is a working link.

Of course, chess on a Klein bottle has got to be at least as cool as chess on a sphere, right?

I like the inclusion of the Wazir, Ferz, Camel, and Elephant. These pieces form a natural sequence: two Wazir's moves at right angles make a Ferz's move, two Ferz's moves at right angles make a Camel's move, and two Camel's moves at right angles make an Elephant's move. That makes me wonder, though, whether the Trojan Horse should contain a Bishop instead of a Knight. Or maybe I just think that because I missed the Bishops while playing Shatranj recently.

Speaking of which, the inclusion of drops will of course prevent the game from ending in a long slow war of attrition.

I would probably play quite badly at first, but I'd love to try this game.

Hey, what happened to all the comments from the past two days?

There are several points where the rules need clarification:

• What is the definition of 'forward' for Pawns and Squires? The definition that seems most natural doesn't satisfy the statement that a Pawn has three forward directions to choose from.
• Regarding promotion, what exactly is meant by 'any higher-ranking piece'? My guess is Queen, Rook, or Templar. Certainly promotion to a second King would change the game drastically. But can a Pawn promote to Squire? Can a Pawn or Squire promote to Obelisk? If so, how does that Obelisk behave, given that it is outside of its Home Territory?
• Is promotion mandatory when a Pawn or Squire enters the opponent's Home Territory? If not, is it mandatory when it reaches the opposing King's starting hex?
• How does a Templar move? There seem to be two contradictory statements: 'The Templar is a combination of the traditional Knight and Bishop.... The Templar moves three spaces diagonally, or two spaces forward and one to the side.' Which of these is correct?
• Can a Queen, Rook, or (maybe) Templar make a 'null move', traveling all the way around the board and returning to its starting space?
• Is the King really in Exile only when forced out of its Home Territory?

[2005-11-14: I've edited this table after the recent and long-dreaded demise of the Black army in Gifford-McElmurry. Just a few minutes later I'm already feeling withdrawal and hoping the last few games will be assigned soon. Has anyone heard from Fergus lately?]

I've been keeping track of the results, and I think the following is correct, although it's possible that I've made mistakes. The last column is the Buchholz-Solkoff tiebreaker, which is guaranteed not to break ties when the tournament is complete.

64 games have been completed, and 14 (listed below) remain.

Alice Chess

• McElmurry-Duniho

Chess with Different Armies

• Duke-Joyce

Extinction Chess

• Madsen-Duke

Hostage Chess

• Duke-McElmurry
• Duniho-Fourrière

Marseillais Chess

• Fourrière-Nelson

Maxima

• Joyce-McElmurry

Rococo

• Duke-Lavieri

Switching Chess

• Carlos-Joyce
• Duke-Strong
• Fourrière-Madsen
• Henriksson-Nelson
• Kaufmann-Gifford
• Lavieri-Duniho

There are a couple of points where the wording could be improved. 'If it steps on an opponent piece it is captured and is permanently off the board' seems to imply that the moving piece is captured, rather than the opposing piece. Also, the phrase 'your other Mook' in the description of seeing red could be taken to mean that one must have two Mooks in order to see red.

In these four games, I count three moves of the type in question: 6. dxc3 in Svidler-Anand, 25. bxc3 in Leko-Kasimjanov, and 50... hxg6 in Morozevich-Adams.

If your friend considers every word on the Internet to be inherently untrustworthy, you might try the local library. Just about any book on chess should contain either the rules or records of games, and you won't have to dig through very many games before finding examples.

My understanding is that KBN vs. K can win in under 50 moves from any
starting position except those where the lone King can immediately capture
the Knight or Bishop.  If my memory's not too crazy, I think the maximum
number of moves required is somewhere around 30-35.  It can be done in 50,
but there's often not much room for error.

Rather than thinking of 2 sets of 480 positions, perhaps it's better to think of 480 sets of 2 positions. The two positions in each set are related by left-right reflection, and have equivalent strategy trees.

I agree with John that it's aesthetically preferable to play all 960 positions, but if you want to choose just one from each pair, how about the one where the Queen is to the left of the King in White's starting position?

I've noticed this also.  The page starts to load, and then seems to get
stuck for a long time (sometimes several minutes), usually after loading 4
kB of data.  The problem is inconsistent; some days I have trouble getting
to my games at all, while on other days everything loads quickly.

The FIDE rule can be stated something like this:

In order to castle...
(1) ...the King moves two squares toward the Rook, and the Rook moves to the other side of the King.
In order for castling to be permissible...
(2) ...neither the King nor the Rook may have moved.
(3) ...none of the King's initial, intermediate, and final squares may be attacked by the opponent.
(4) ...the initial, intermediate, and final squares of the King and of the Rook must be vacant except for the King and the Rook.

(1') ...the King moves to the c-file and the a-side Rook moves to the d-file, or the King moves to the g-file and the h-side Rook moves to the f-file.

From the standard starting position, these two statements are equivalent. But when we try to generalize to other starting positions, they differ, and we must choose one or the other (or a third statement not listed here). If we follow (1'), then we have the Fischer rule, with no change to the statement. If we follow (1) and include additional language to deal with the special case where the King begins on the b- or g-file and can't move two squares toward the near edge, then we have the rule proposed here (which I'll call the Lewis rule). Because of the necessity of handling this special case, I consider the Lewis rule to be not cleaner, but less clean than the Fischer rule.

At a glance, the Fischer rule may seem a bit ugly, since it explicitly breaks the symmetry of left-to-right reflection. But this actually makes the game richer, as it allows more possibilities. Under the Lewis rule (or any other symmetric castling rule), there is no meaningful distinction between a starting position and its mirror image, and the game should really be called Chess480.

It could be that in some games the player to move hasn't checked frequently to see whether the site is back up.

As for the emails, I can't speak for others, but I've received the notifications of my opponents' moves only inconsistently throughout the summer.

CwDA is one of those games that includes FIDE Chess as a subgame, and I don't see a problem with allowing it to do so. It's worth noting that if the players choose their armies, then anyone with a sufficiently strong desire not to play FIDE-FIDE can avoid it by not choosing the FIDE army.

Also, we've already played Fischer Random Chess in this tournament. One FRC game in 960 is identical to FIDE Chess, and we didn't make any provision to prevent that from happening. Granted, the probability is somewhat higher in CwDA, but it's not all that bad. With unrestricted random assignment, there's about a 77% probability that FIDE-FIDE won't occur in the tournament.

Peter,

Regarding game design, I agree with Michael Nelson's last comment, but after your last comment I understand your position much better. I now see this as a question, not of logic, but of taste. And it's not my game, so I'll shut up about that.

I'm glad to see that the rules have been revised to reflect your intentions (although I had hoped that the tournament issue would be resolved first), and I believe that the revised rules are equivalent to what you have indicated in recent comments. But, if you're interested, I have a few suggestions as to how I think they could be written better, so as to avoid future misunderstandings.

The sentence that was most controversial in the original rules was this one:

Or in other words, a piece may only end up on an edge square by making a capturing move that would not be possible without landing on the edge square.

The section labeled 'Rules' seems unnecessarily complex and extremely redundant. The same statement is made at least four times, and if I were unfamiliar with Rococo, I would have to read this section three or four times to be sure I had it right.

You've now included a precise mathematical definition of the term 'capturing move'. This is good, because I now understand what you mean by the term, but I'm not sure that this is the best term for this definition. To see what I mean, consider this example: A Long Leaper on a4 could capture an opposing piece on a3 by moving to a2 or a1 (assuming both of these squares are vacant). In the absence of a definition of the term 'capturing move', I would regard these as two distinct capturing moves, since they are two distinct moves. But it seems natural to say that they result in the same capture. The rules would be clearer (to me, at least) if the term 'capturing move' were replaced everywhere with 'capture'.

The inclusion of the clarifying example is perhaps helpful, but it is not the best possible example, since it does not fully distinguish between the three interpretations of the original rules. A better example can be formed by placing the piece to be captured on x3 rather than x2, so that under the revised rules the Long Leaper can capture by moving to x2, but not to x1 or x0. Also, the example breaks up the flow of the redundant rules; perhaps it could be moved to its own paragraph. And if the rules refer to the square x9, the ASCII diagram should be modified to refer to the last rank as 9, rather than 00.

The formal statement of the edge-square rules is unambiguous, but statement 4 is completely unnecessary. It follows from the first three statements, and should therefore be treated as a theorem rather than a rule. If I'm not mistaken, the entirety of the edge-square rules could be stated in one sentence, something like:

A piece p capturing a set P of opposing pieces may land on or pass over only the minimal number of edge squares necessary for p to capture all the pieces in P.

The piece descriptions of the Long Leaper and the Withdrawer should include statements that these pieces, when moving along an edge, may move only the shortest possible distance for a particular capture. In fact, since each piece description contains the edge-square rules applying to that piece, I wonder whether it would be better for the 'Rules' section simply to introduce the general concept of edge squares, and leave the details to the individual piece descriptions.

Finally, I've noticed some typos: 'mininal', 'Moves that captures', and 'fewist'.

For the sake of posterity, here is a link to the game in question.

I see two logically independent questions here:
1) What should the rules be?
2) What is the correct interpretation of the rules as written?

The first question is relevant to future games of Rococo and should be answered by the inventors, taking input from the rest of us if they wish to. For the particular case of the present game, I think that the second question is relevant, and that ideally it should be answered by consensus.

The rules as written contain a general statement describing the concept of edge squares and their role in the game, and a specific statement for each type of piece, describing how that piece behaves with respect to edge squares.

The general statement:

These marked squares on the edge of the board are edge squares, and a move may only end on an edge square if necessary for a capture. Or in other words, a piece may only end up on an edge square by making a capturing move that would not be possible without landing on the edge square. This includes moves that start on edge squares.

The specific statement for the Long Leaper:

It may end its move on an edge square only when that is the only way to make a particular capture.

As I read these rules, the only thing that could be construed to forbid my move to x0 is the definite article in the second sentence of the general statement. It's worth noting that David's interpretation, based on this definite article, is paradoxical (as Mike Nelson has pointed out), and that the definite article only appears in a sentence which is presented as a rephrasing of the previous sentence. The first sentence of the general statement and the specific statement both imply (in my opinion; there may be some room for debate) that one edge square is as good as another. Since I can't capture the black Leaper by moving to an interior square, x1 and x0, both being edge squares, should be equally permissible under the rules as written.

Peter's emendation of 'landing on' to 'landing on or passing over' does produce a well-defined rule, but in my opinion it is inconsistent with the intention of the edge squares. If the black Leaper were on x3, then this rule would say that I can capture only by moving to x2, not x1. But a piece on x1 is no safer from a Leaper than a piece on x2. On the other hand, a piece on x0 is safer, so it would seem reasonable to forbid capture on x0, not because it is not the first square beyond the victim, but because it is on two edges rather than one.

In fact, I think this was Mike Madsen's understanding of the rules. (I hope he'll correct me if it wasn't.) While I maintain my belief that this is not the rule as written, it seems sensible, and in my opinion would be the best rule for the game. It could perhaps be stated most easily by defining three classes of squares (interior, edge, and corner), and forbidding a piece to move to an edge square except to make a capture which cannot be made by moving to an interior square, or to move to a corner square except to make a capture which cannot be made by moving to an interior or edge square.

I agree that such a position is very unlikely -- presumably it's unusual for a game of Extinction Chess even to last so long -- but it's not inconceivable. If the ruling had been for a draw, then one could imagine a game in which Black had no winning chances, but could force a draw by moving the Bishop to c8, forcing the Pawn to promote or die.

On the other hand, since the capture-promotion has been declared a win for White, it seems that the position as I stated it could arise only after an obvious blunder by Black. But of course the Black piece need not be a Bishop. One could imagine White pushing his last Pawn to the 7th rank, forking Black's last Rook and Knight.

Also, I've just thought of the following pathological possibility. Suppose that, after 41 moves of a game of Extinction Chess, White's only Pawn is on b7 and Black's only Bishop is on c8. If White then plays 42. bxc8=Q, Black's Bishops are extinct, but so are White's Pawns. So the game is clearly over, but what is the result?