Paul:

Both you and Suzanne have raised several questions about the rules of Solitaire as I have implemented them in my Solitaire 1.0 game (or "research tool" as I like to call it). You seem surprised that you can't transfer cards from the ace piles once they've been placed there, and Suzanne is used to taking every third card from the deck without starting over after every pass.

You're questions spurred me to do some much-needed research into the history of the game. After perusing a shelf full of books on the subject at my local bookstore, I have discovered that there are literally hundreds of variations on Solitaire, and that not even the experts agree on the precise rules of each variation.

We seem to be playing a cross between the two most popular variations, Canfield and Klondike (some books list these as synonymous, but Hoyle insists on a distinction). Klondike is played exactly as I have implemented it except that the player is allowed only one pass through the deck. Canfield was originally played for money and is different in several respects: foundations which start with cards other than the ace, only four tableau piles instead of seven, a separate "stock" pile for refurbishing the tableaus, and our method of moving through the deck in packets of three.

In none of the games I surveyed was it permissible to remove cards from the ace piles once they had been placed. According to Hoyle "on the aces you may build up in suit and sequence, and to win the game you have to build each up to the king. Once played on these foundations, a card may not be withdrawn to help elsewhere."

The method of moving through the deck in threes we owe to Canfield: "Turn up cards from the hand in packets of three, placing them in one waste pile, face up. You may play off the top card of the waste pile onto foundations or tableau, thus releasing lower cards one by one. After you have exhausted the hand, pick up the waste pile without shuffling, turn it over, and there is your new hand. You may run through the hand as many times as you wish, until play comes to a standstill." Again, I was unable to find any variation in which Suzanne's "circular" deck was permitted.

[In the original casino, players were allowed to move only once through the deck one card a time (like Klondike) or to move three times only through the deck in packets of three. Modern Canfield allows unlimited passes through the deck.]

So I believe the rules as I have implemented them are as close to standard as we are likely to get. This is not to say that the variations Paul and Suzanne have suggested are any less valid than all of the other variations that have evolved over the years. In fact, I have been toying with a variation of my own, adapted to the abilities of a computerized game.

What do you think of this: instead of moving through the deck in the traditional packets of three, the computer arranges the deck along the bottom of the screen in PILES of three (with the final one, two, or three cards in the rightmost pile). The player is free to draw from the top of any pile, but as soon as (s)he does so, a marker will appear below that deck. The player can then draw from any pile to the RIGHT of the marker; when this happens the marker will shift over to indicate the rightmost pile touched in this "pass" and cards to the left are rearranged into piles of three. When the player draws from a pile to the LEFT of a marker, all cards to the right of the marker are instantly reshuffled into piles of three and the marker is reset.

This sounds rather bizarre, but I think would be easy to grasp when implemented on the computer screen. And although it seems like a considerable departure from the classic game, it is in fact not a departure at all. The marker method simulates exactly what happens when we go through a deck in the traditional packets of three. The only difference is that the deck is spread out so that we can see the sequence uncovered so far instead of trying to remember it. A player could pull precisely the same cards from the deck under the old method (as long as there is no limit on the number of passes) by noting each card uncovered as it went by and then planning out each pass ahead of time. A player with a good memory already does something like this.

This variation occured to me as I was thinking about my plans for "robot solitaire" (in which different algorithms are tested directly through automated play). Of course, I wouldn't allow my robotic players to cheat (to use knowldege of hidden cards in making their decisions), but the question arose, should I allow them to use knowledge of cards uncovered during passes through the deck?

Most human players do not remember these cards in any detail, but in theory they could. Sometimes, for example, when I have a hole in the tableau and a red king appears early in my pass through the deck, I might wait because I know that a black king will turn up toward the end of the deck. If I took the trouble to jot down each card encountered during my first pass, I could anticipate at least part of the sequence which would unfold during the second pass. Eventually I could deduce the entire sequence and use that information to make better decisions.

My piles of three method would serve as an equalizer that would give all players this same opportunity without forcing them to take notes or rely on memory. It would change the nature of the game. I think it might make it more interesting, more challenging in the sense that we would have more to think about. It might also slightly raise our chances of winning.

What do you think? If I receive any encouragement I will create a new version of our Solitaire game and release it in a future issue. We could name this variation "Archipelago."

See this issue's installment of Mr. Wizard for more fast-breaking solitaire news.