Comments/Ratings for a Single Item

Okay, that's fixed.

Reporting a problem on this preset. Between the image of the board and the options menu there is a clearly intrusive (extraviated) long text which is very cumbersome to have to skip.

What is going on with these two Alice Chess logs?

• fordy101-cvgameroom-2016-149-619
• fordy101-cvgameroom-2016-149-620

Both of them have the same list of moves, and these moves include several instances of the same banned move. These logs should not exist like this unless someone handwrote and uploaded them. I could not duplicate them myself through the normal means of playing the game. I copied the movelist to another log up until the last legal move, then tried to move the Knight from g8. When I tried the partial move that got recorded, it caught it before I could send the move. When I tried each legal move available to the piece, it handled it without any problem. When I tried a move directly to the other board, it stopped me. So, what happened here? How did these two logs with banned moves in the moveslist come to be?

Yes, that's how I'm interpreting the rule.

Perhaps the interpretation is that a move that interposes is "legal on the board it is played" because there is no King on that board. The King remains in check on the other board by playing the move (before the transfer), but that does not make the move illegal on the board without King. So the conditions are:

1) You must play a move that would be strictly legal on the board it is played if that were the only board
2) The move must not end on a square that is occupied on the other board, and after the move you must transfer the piece there
3) After you thus complete a move, the opponent must not be able to capture your King with his reply

This and my prior logical analysis of the first rule settle the matter. A piece may block a check by moving from the other board.

Note that the "legal if strictly legal on the board it is played" interpretation is flawed anyway, because it would allow you to legally put your own King in check: white Ke1, Pe2, black Qe8 (all on same board) would have e2-e4 legal, because before you transfer e4 to e4*, your King is not yet in check.

I was never even considering that interpretation. The two interpretations I'm considering are (1a) A move is not legal unless it is strictly legal on its board, including not leaving the King in check even temporarily and (1b) A move is not legal when, considering the board it moved on as the only board in the game, that move would be considered illegal. Both of these are only necessary conditions on whether a move is legal. With either interpretation of the rule, you could not put your King in check, because there is already another rule against that.

The second interpretation, which is the one you agree with, makes more sense of the clause "on the board where it is played." If we consider this clause to be limiting the domain we're going to look at to determine the legality of the move, this rule is not redundant. But on the strict legality interpretation, the rule amounts to nothing more than "a move is not legal unless it is legal." But that interpretation actually leaves things ambiguous. It is trying to define legality in Alice Chess in terms of legality in Alice Chess, leaving open the question of what is really legal in Alice Chess. 1b defines legality non-recursively by defining legality in Alice Chess in terms of legality in Chess, as played on a single board. So, on 1b, leaving the King in check makes a move illegal only if the King would be left in check on that board before the piece moved to the other board. Whether the King is left in check on the other board does not affect the legality of the move. So, a piece should be able to block a check by moving from the board the King is not on.

Indeed, this is what I meant. That the Fairy-Max derivative that plays Alice Chess does it this way is completely accidental; initially I had misunderstood the rules to be such that moves had to be pseudo-legal on the board they were played, and that you only had to worry about your King being capturable after the transfer. This made it virtually impossible to checkmate a King, though, as he keeps fleeing to the other board.

So I patched it for not being allowed to pass through check. Internally it uses the single-board representation, and a flag in the piece code to indicate on which board it is. That flag is toggled every time you move the piece. When a move attempts to go to a square that is occupied by a piece marked as being on the other board, that square counts as empty and the non-capture move rights for that square are suppressed. Those were basically the only changes that were needed to make it play Alice Chess, except for this passing-through-check rule. So I later patched the code to make the test for royalty of the victim before it checks on which board it is, and if it is on the other board, abort due to King capture only if that square is also the to-square of the previous move (which was passed from parent to daughter node anyway for implementing Berolina e.p. captures).

This catches passing through check on any King move, but there is no test for King capture between making the move N transfering to the other board. A move going over that square would either skip over the square on the board you just left, but would block any moves over that square on the board you just arrived on. To implement the other rule interpretation, it would have to continue moves through the to-square of the previous move, solely for the purpose if there was a King behind it. It does not do that.

Note that the "legal if strictly legal on the board it is played" interpretation is flawed anyway, because it would allow you to legally put your own King in check: white Ke1, Pe2, black Qe8 (all on same board) would have e2-e4 legal, because before you transfer e4 to e4*, your King is not yet in check.

Here is a game I just played out from Pritchard's encyclopedia. In this game, no check is blocked by moving a piece from the other board, though there was the occasional opportunity to do so.

Just to be clear here, you mean "If what Erik Maranus said is true ...", you are saying that Fairy-Max allows a check to be blocked by moving a piece from the other board, and you agree with this interpretation of the rules. Is this correct?

If that is true, the description of the rules as given here ("the move must be legal on the board where it starts") is wrong. Funny enough my Alice-Chess engine seems to accept them, though, as it is a King-capture engine. So it first makes the full move (including the board transfer), and then checks if the reply can capture a King. The only thing special I programmed is that it also triggers on capturing to an empty square if that square has a King on the other board that just moved. (I.e. it is as if the King on moving leaves a 'ghost' on the square through which it passed, and can be captured e.p. there.) This would not flag interposing as illegal, however.

It seems a proper formulation of the rules would be: "the move must be pseudo-legal (i.e. legal when ignoring check) on the board where it is made, but when a King moves it must not pass through check on that board."

I found this written on Ed Friedlander's Alice Chess page: "Erik Maranus pointed out to me (3/19/06) that a piece can interpose to save a king in check by moving from the other board. I had originally misinterpreted this as illegal because the move would not by itself be legal on the other board. However, examining the model game in "Pritchard" satisfied me that he is correct."

I have both of Pritchard's books. So I will look into this later.

I was going through a game on Wikipedia, and I will stop here for now. This goes up to move 11, followed by a checkmate variation that my code is currently recognizing only as check.

Here's a curious comment George Jelliss made in this game: "To give a direct check to the king the checking piece must come from the other board, so it is necessary first to transfer forces to the other board." On one interpretation, this is not true at all. Pieces can't capture by moving to an empty space and transferring to an occupied square on the other board. So I assume they can't check that way either. In the earlier examples of checkmate given on Wikipedia, the check comes from a piece on the same board as the King. Maybe what he means is that a piece must move from the other board to be on a space from which it can check the King.

In the Alice Chess preset I'm testing, I have come to a position where a move I was thinking of is not being allowed by the preset. The Red King is in check, possibly checkmate, depending on the interpretation of the rules. It seems my code is in conflict about this. It is recognizing it as check rather than checkmate, but it is also not allowing any move that could end the check. I was thinking of moving the F6 Knight or the F8 Rook to G8, from whence it would move to g8, blocking the check. The problem with both of these moves is that they leave the King in check until the piece transfers to the other board. I have coded this to be illegal under the understanding that any move leaving the King in check before the transfer is not legal. This certainly makes sense when the piece and King are on the same board. But here they are on different boards. The rule "A move must be legal on the board where it is played" is ambiguous. A loose interpretation is that it refers to a piece's own powers of movement, and a strict interpretation is that it disallows any move that leaves the King in check before the transfer. How have people been interpreting the rule in situations like this?

I'm writing a new Alice Chess preset that will fully enforce the rules. I'm posting the game I was playing to test it, so that it will be here when I continue testing it.