Retro Chess

Wei-Hwa Huang, 25 Apr 1995

Two years ago I invented along with my friend Robert Au a game inspired
by retroanalysis, which we called "Retro-chess."  Unfortunately, we never
finished a game due to its complexity.  Here are the rules:
Start with an empty chessboard.
Black places his king on any square.  White then places his king on 
any square not adjacent to the Black king.  Play then alternates.
Each move must be a "backwards" move in time in a valid chess game.
The winner is the first player to restore all 16 of his pieces to the 
starting position.
If either player proves that the position is unreachable in a standard
chess game, then the player who has made the last "unmove" loses.  A 
corollary is that a player may try to bluff and create an unreachable 
position, hoping that the opponent will not notice and make an unmove.
If after your unmove your King is in check, it is customary to say "kcehc"
(pronounced "kech") to warn your opponent that he must uncheck your king
his next move.
The Fifty-Move Law:  A player loses if after fifty of his moves he has not
uncaptured any of his opponent's pieces, and not vacated any of his opponent's
starting squares.  In other words, he does not lose if within the last fifty 
moves he has made an uncapture or has moved a piece away from the 16 squares
on the other side of the board.
Perpetual Kcehc:  A player may not Kcehc more than 10 moves in a row.
This may work for a play by e-mail game.  Are there any contingencies that I
have not planned?

Philippe Schnoebelen, 25 Apr 1995

This looks like playing Proca-Retractor ! Except that in Proca games, the
goal is not to restore your men to their starting position (what if your
opponent will not free your home squares ?) but to reach a position where a
forward mate in 1 is possible. That is, the goal is to retract your last
move and mate in one. If you can't mate in one then it is your opponent who
will retract and try to mate in one.
The problem with starting with two lone Kings is to get back your men
without which you can't mate the opponent. Clearly the opponent will not
uncapture them unless forced. 
I remember playing a few such games with Laurent Joudon. A usual trick was
to bring the wh K to a1, retract -1. Kb2xBa1 a2-a1=B+ -2. Kb1-b2 and if
your opponent will not retract b3xa2 then drive the bl.P up to a7, start
the same trick again and eventually you'll get some of your men back. (Of
course all these maneuvers have been discovered by the 1st composers of
Proca-retractors problems, perhaps Zeno Proca himself !!!)
Quite soon interesting legality problems arise, and we had to spend more
and more time between our moves, simply sorting out the position. I
remember it was very funny even when the games were aborted due to lack of
> This may work for a play by e-mail game.  Are there any
> contingencies that I have not planned?
Certainly!  Imagine a World RetroChess Championship, between G. Vorapsak
and V. Dnana. Who could you hire as a valid TD ? Only M. Dualliac could
manage this...

Richard Sabey, 27 Jul 1995

I propose replacing Wei-Hwa's (or should I say Huang's?) perpetual kcehc
rule with the rule that if a player may retract so as to create a position
which has occured at least twice before, with the same player to retract,
and the same set of retractions legal, then he may claim a draw by 3-fold
I think that a player can prevent his opponent reaching the array in any
finite number of moves, merely by blockading his opponent's men's starting
squares. Rules such as Wei-Hwa's "A player loses if after fifty of his
moves he has not uncaptured any of his opponent's pieces, and not vacated
any of his opponent's starting squares." are artificial and don't do enough
to prevent this blockading tactic.
Perhaps it would be better to play Hoeg-retractor. That is, play goes in
the following sequence:
- Black retracts a move,
- White optionally adds a white man whose capture has been retracted,
- White retracts a move,
- Black optionally adds a black man whose capture has been retracted,
  provided that at all times the retractions are legal.
I prefer to ditch the aim of reinstating the array, and go for Philippe's
aim of being in a position to retract a single-move and then mate in 1. RA
is still very much involved, as a player may still challenge the legality
of either his opponent's last retraction or the choice of uncaptured man.
You lose Philippe's tactic of retracting Kb2xBa1, but you gain a far
greater freedom to get your men back. I haven't tried it, but perhaps
there's a risk of brutal mates with lots of queens on the board. In this
case, make an additional rule that each retraction must be consistent with
there never having been any promotions. So each player may not add more
than one queen.