I once heard someone say that every game of Freecell can be defeated. I’m pretty good at the game (about 85% success), but I have no way of knowing if that particular claim is true. I can imagine some severely degenerate card layouts that would be difficult to say the least, and I don’t replay games I’ve lost because I find myself repeating most if not all of my previous moves. Has anyone else ever heard this claim? Does anyone know if it’s true?
The claim (made in the help file of the game itself) is that all of the 32000-or-so layouts that come with the game have a solution. It is easy to construct layouts that have no solution.
There’s a note in free cell help that states “It is believed (although not proven) that every game is winnable”.
I usually play a game over and over until I win. I’ve had to bail on a few, but generally there’s always a way to win.
Man, y’all are quick! Thanks for the help. I guess I never looked at the help file. If they’ve only got 32000 layouts, I suppose it’s quite possible to develop others that aren’t on the list that are unsolvable. That would be one booger of a proof though.
After 932 consecutive wins, I was close to 933. Then I made a “cute” move (I wanted to get the cards a certain way for aestheic purposes), but I was too tired (no, not from playing over and over, but from a long day at work) to notice that I was one space short. SACK!!!
Actually, a lot of people have reasearched the game extensively. According to this FAQ and several others,
one of the MS games is unsolvable - #11982:
http://members.aol.com/wgreview/fcfaq.html
Once you get the hang of it, it’s easy to win most of the random deals you’re handed, so I imposed a “scoring” system on myself - your “score” is the number of cards still left when the program concedes the win and plays the game out for you. This means you can try to win while deliberately leaving an ace buried to the end somewhere, for instance, and it becomes a challenge as to whether you can get away with that or not. A “good” game is where you can get away with leaving a PAIR of aces buried to the end, and get a score of 48 (all that’s up when you uncover the two aces is the other two aces and their deuces which freecell automatically played up for you). 50 is possible with at least one of the MS layouts.
That FAQ seems to call my little addition “flourishes”.
At least part of my problem is that I play it as throwaways - I don’t spend enough time thinking so I lose due to “cute” moves ala Johnny L.A. Still, I don’t know if I’m up to 900 straight wins even at my best (have to stop playing while drinking beers). Of course, now I’ll have to try the “unbeatable” one. Thanks again y’all. You guys are so freakin cool!
Oh, and here’s a picture of the intended “insoluble”
one, in case MS decided to rearrange the numbers in
some version (I don’t think they have):
I’m not sure if game 11982 has actually been mathematically proven to be unsolvable. The last I heard, it hadn’t been solved, which isn’t necessarily the same thing. Of all the possible games that have been analyzed, at least 100 games have so far defied solution. Only one of these is included in Microsoft’s 32,000. (There are 52!=8*10[sup]67[/sup] possible games).
There was an earlier thread FreeCell
Game 11982 has been conclusively proven unsolvable by exhaustive search. There’s a freeware program called FreeCell Pro which can do the solution checks in a surprisingly short period of time. All of the other Microsoft games are known to be solvable, however. There’s also at least one layout which has been mathematically proven unsolvable by a human being, but it’s rather contrived, and is known to not be in the Microsoft set and presumed not to be in the FC Pro set.
As to flourishes, I seem to remember reading of a game that allowed for a full 52-card flourish, but I can’t remember if it was from the Microsoft 32k games, or one of the 4G games accessible in Freecell Pro.
By the way, on a moderator note, can we keep this thread to the factual discussion? If you want to brag about your high scores or consecutive games, take it to IMHO.
Bibliophage,
Mea culpa … didn’t search for the thread. I knew that 32000 was **way[\b] too few possibilities for the card layout. You know, that would be a pretty cool job though.
“Yah, I research possibilities for solving Freecell”.
If anyone knows if they’re hiring …
Ricken-fricken codes.
I missed your previous post Chronos. Thanks for the info.
I didn’t know you were not supposed to do them over as it didn’t count. I have had to work weeks on a game to find the solution. I am trying to do them in sequence and am just past 4k now. I will now have to refrain from jumpinmg up to that 11,000 something game {already trying to forget the # } and hope that by the time I’m there I will be grounded in the game well enough to get that one to. I’m a persistent devil… LOL Q? If a machine can firuge out a game, why can not a human? Or is that based on the one try, no mistakes, all the way straight to a victory solution?
3½¢
In addition to the impossible randomly generated hand 11982, the Microsoft game also has two contrived impossible games that can be attempted by picking game numbers negative 1 or negative 2.
According to bibliophage:
You are counting games with the same columns in a different order as different, since you say there are 52! possible games. So that one impossible game should count as 576 impossible games, and the two contrived hands give another 1152 impossible hands. There are plenty of impossible hands: even if one every 32,000, 52!/32000=2.5E63
BTW, a quibble with the counting of 52! as the number of possible games - that is true if you count every permutation as a game, but you can rearrange the columns and have a game for which the identical solution would apply, simply playing out of different columns. You can also switch the positions of all the pairs of red suit cards and/or black suit cards, or map both red suits to the black suits and vice-versa. The same solution would apply with a translation of suits. There may be other ways to trivially reduce the number that I’m not thinking of off the top of my head. There’s still a lot, though.
Oh, hell, I’ve been simulposted! Well, at least Manlob didn’t mention that you can translate suits (not ALL translations though, since you have to follow the black-on-red rule in play).
Try selecting games -1 or -2 . Those are definately not winnable.
FWIW, I have done some freeware games for the HP 200LX palmtop PC, and FreeCell was one of them. In my implementation, I didn’t have access at the time to the random shuffle algorithm that MS used, so I created my own with one million possible deals. (I’ve since added an option for an MS-numbered deal).
The user community found one of them that they couldn’t solve, and it was verified by the FreeCell Pro people that it was not solvable. The guy who wrote FreeCell Pro told me that this was the first non-solvable randomly generated game he had ever seen besides the MS 11982 game.
People have already pointed out that unwinnable deals are most certainly possible. Here is somebody with way too much time on his hands proving a theorem saying that “any deal satisfying the following five conditions is unwinnable.”
He also shows a diagram of a deal that is clearly unwinnable; where the hell are you going to go from that position?
The FreeCell Solution:
Shift-Ctrl-F10 then choose “Abort”, select card, and you will win!