This is a pet theory of mine, actually a pet hypothesis, since I don't have enough to hang a theory on. I suspect that Windows Solitaire has a function built into it that compels it to let you win a certain percentage of the time. I think if you've had an extraordinarily long run of bad luck, the program reassigns the values of the unexposed cards to give you more favorable chances.
     I started getting this idea in Saudi Arabia, during the Gulf War, when I spent vast amounts of time (happily on the night shift - yay bio clock heaven) playing one hand after another after another. Between scud alerts, that is. Sometimes I was so frustrated at not having won in so long, that I decided to play one last game, and that would be the game I won. This happened often enough that it made me suspicious.
     I was discussing this with Rebar the other day. How would you test something like that? For a long time, I've been refusing to carry my wins all the way through, stopping as soon as all cards were exposed. I call that an obvious win. I have no way of knowing whether the program counts that as a win or not, but I suspect not, because my rate of obvious wins seems to be greater than my former rate of carried through wins.
     Since we'd be trying to measure the luck factor, we'd have to have accomplished accomplices (say THAT three times quickly) who would be unlikely to ever miss a play. Meaning that in the course of play, if a card can be used or moved appropriately toward the winning of the game, it is, consistently. I would want a control, too. A deck of cards and a shuffler. They do make shuffling machines, don't they? I'd need at least three accomplices, one to play Windows Solitaire on a computer the usual way, carrying the wins all the way through, meaning sorting out all the suits on the target spaces so that the cards go bouncing all around. The second person would be on a different computer with the same version of Windows Solitaire using the same settings, but would only play until all cards are exposed. The third person will have the physical cards and shuffler. Each person will play a predetermined number of hands, recording each loss and win (or obvious win in the second case). Then, to make sure all factors are equal, everyone switches places and does the same number of hands, again noting the losses and wins. Then one more round of musical chairs and card playing. Then we crunch the data and see how the ratios of wins and losses compare between the two computers and against the manual deck.
     Whaddya think? A sound method? Got a better one?