One of the popular games distributed with Windows 95 and later is FreeCell, an interesting variety of the Solitaire card game.
The help file for this game contains the following statement:
It is believed (although not proven) that every game is winnable.
This rather strange claim indicates that someone at Microsoft has a warped sense of humor.
For one, it starts with the unclear "it is believed" postulate. Believed by whom? Microsoft? The author of the help file? Someone who has played the game and kept winning?
The postulate is worded very carefully. Obviously, the every game is winnable statement is false. But if there is at least one person somewhere who believes it true, the postulate makes the full statement true, i.e., it is indeed believed that...
Secondly, there is the although not proven assertion. This is what indicates the whole thing is actually a joke. Only someone who knows that not every game is winnable can state that the belief has not been proven. After all, if every game were winnable, how would you know that no one has proven its truthfullness? What if someone did prove it and did not tell you.
If, on the other hand, you have the proof that not every game is winnable, then you can state with certainty that the belief has not been proven.
Finally, there is the claim that every game is winnable. This statement is false.
How can we prove that not every game is winnable? Simple: All we need to do is find a game that cannot be won.
In the case of FreeCell, one such game would have all kings and jacks at the bottom, and all twos and fours right above them. This game is not winnable, hence not every game is winnable. QED.
Can we generalize this and find a way of proving or disproving the winnability of any card game?
In the days of computers this is quite easy: Simply write a program that plays all possible variations of the game until you either come to a hand that cannot be won (not all games are winnable), ot you play and win them all (all games are winnable).
Naturally, this program will take a while to run (possibly even several days), but it is certainly relatively easy to determine for any kind of card game whether every game is winnable.
Surely, someone at Microsoft knows that, and surely, Microsoft has enough resources to use this brute force approach to prove than not every FreeCell game is winnable. Hence, the whole It is believed (although not proven) that every game is winnable statement had to be a joke.
P.S. I was not aware of game -1 pointed out by post86. This game is a variation of the unwinnable game I listed above, and is a clear proof that the "it is believed" statement was indeed a joke.