This is a collection of Turn Off variants in which the pattern of light changes correspond to the moves of various fairy chess pieces. In the standard Turn Off that comes with Zillions, when a light is selected, it and all orthogonally adjacent lights change state. This corresponds to the Wazir fairy piece, which can move to any orthogonally adjacent piece. Variants included in this file feature the Wazir, Fers, Dabbaba, Ferdaba, Alfil, Squirrel, Camel, and Zebra. A description of the moves of each piece is included in the "game description" section. Each game starts with all lights on, and the goal is to turn them all off.
 
Curiously, the Zillions version of Turn Off does not include a variant that starts with all the lights on. It turns out that this is a very difficult puzzle. In my experience Zillions has been unable to find the solution on the higher settings. Set to "stupid" it will sometimes find a solution by accident but usually after several hundred moves.
 
So the challenge is: Find the efficient solution to the Wazir 5x5 Turn Off puzzle. You can cheat all you want. The first person to send me the correct answer at dtroyka@justice.com gets a certified smiley emoticon. :-)
 
Wazir Challenge is a companion game to Royal Turnoff.
 
UPDATE: Robert Gauss and Jeff Roy both solved this puzzle the week it came out. We all agree that the shortest solution is in 15 moves. This solution is now included with the game in a Solutions folder, which you can also access through the "Show Solution" option in the Help menu.
 
The order in which lights are selected is irrelevant because the state of a light -- off or on -- depends solely on how many times it or a neighboring light has been selected, not on the order of selection. A priori it is clear that the optimal solution cannot be over 25 turns. Selecting a light twice is the same as not selecting it at all (the second click undoes the first click), so an efficient solution cannot contain multiple selections of the same light.
 
Jeff Roy provided an interesting proof that the largest number of turns in the efficient solution to any solvable 5x5 Wazir puzzle is 17 turns. (By efficient I mean that all doublets have been subtracted out and that each light is selected either once or not at all.) He identified a "return pattern" of 16 distinct lights that, when selected, returns the board to the starting pattern so that no change in board state occurs. This pattern is a2, a3, a4, b1, b3, b5, c1, c2, c4, c5, d1, d3, d5, e2, e3, and e4. To summarize the proof: Suppose there is an efficient path of 18 moves from an initial state to a final state. A player could then add the 16 lights in the return pattern, which would create a list of 34 moves leading from the initial state to the final state. In this list of 34 moves there must be at least 9 doublets because there are nine more moves than there are lights on the board. (There can be no triplets because each pattern is composed of distinct selections.) If these doublets are subtracted out, you arrive at the same board state in 16 or fewer moves, which is a contradiction. Ergo, there is no such thing as an efficient path of 18 moves.