How to Beat the Impossible Freecell Game

On Windows, Freecell game #11982 is impossible to beat. But there is a way, albeit unorthodox.

Open up Freecell and hit F3 (or Game -> Select Game) and type in 11982. Hit Control-Shift-F10, then move the 2 of hearts to the 3 of spades and voila, you won the game. At least, that’s what your game statistics will say.

Beat #11982

Update: If you think you can beat this game, check again. This is the starting layout.

Freecell game 11982

The FreeCell FAQ declared game 11982 to be impossible after exhaustive research. They wrote, “11982 has now eluded solution by probably thousands of human solvers, and at least eight independent computer programs I am aware of (most of which are designed to search exhaustively for a solution), and I am confident in calling it impossible.”





  1. Re Game No. 11982, which you say is unwinnable. I just played it and won the first time I tried it! Played in the usual way, without the “trick” you described.

      » Comment by Harriet on February 14, 2009 @ 2:38 pm
  2. Harriet: Are you sure? #11982 has been proven to be unbeatable. You probably mistyped the game number.

      » Comment by Dan on February 14, 2009 @ 3:32 pm
  3. Yep, if Harriet comes back with step by step instructions, then I would believe her. I’ve been trying to beat this one for awhile. I haven’t given up hope yet. It always seems like I am one cell short ;)

      » Comment by Ross on June 25, 2009 @ 2:46 pm
  4. I’ve won every game from 1 to 11981 and tried for two mo. to win 11982. Don’t believe it’s possible in the conventional way.

      » Comment by SKW on July 3, 2009 @ 7:27 am
  5. #11982 is not PROVEN to be unbeatable (extremely hard to prove something like that), but is BELIEVED to be.

      » Comment by Frank on October 22, 2009 @ 7:21 am
  6. @Frank Correct.

      » Comment by Dan on October 22, 2009 @ 10:26 am
  7. Game is impossible, i tried it. the only way to win is by cheating or glitching

      » Comment by Cody on June 18, 2010 @ 3:40 pm
  8. http://freecellgamesolutions.com/ds/?g=1982&p=aiq

      » Comment by hossi on June 23, 2010 @ 1:19 pm
  9. @hossi That’s 1982, the impossible game is #11982.

      » Comment by Dan on June 23, 2010 @ 2:59 pm
  10. Before realizing the game was impossible, I came across it at random. I can’t give up on a game without winning, so naturally, I played this hand over and over again. Never won. Is this hand still proven to be impossible? I see that one user posted he/she beat it on first try. Really??

      » Comment by Jennifer on November 30, 2010 @ 4:55 pm
  11. @Jennifer No one has ever beaten the game and shown how it was done. I can guarantee Harriet didn’t beat it on the first try. She played another game thinking it was 11982. If you do beat it, you’d immediately become a Freecell legend.

      » Comment by Dan on November 30, 2010 @ 5:04 pm
  12. How about help on 25450?

      » Comment by Ed on July 8, 2011 @ 10:16 pm
  13. I have completed 16710 games in a row except 11982 keep going back to 11982, findly went on line to see if possable,glad I did after 500+
    attempts over 2 years time. thanks people

      » Comment by ralph on August 6, 2011 @ 7:16 am
  14. @ralph That’s true dedication.

      » Comment by Dan on August 9, 2011 @ 8:54 am
  15. The picture above with the 11982 game won is fake.
    Try to open the game, than hit ctrl+shift+f10 , hit abort, than move a card and the cheat is done.
    I also tried 11982 but no succes,
    best regards,
    tibanu

      » Comment by tibanu on December 7, 2011 @ 1:18 pm
  16. @tibanu Did you even read the post?

      » Comment by Dan on December 7, 2011 @ 2:59 pm
  17. I play freecell evryday @ work and i beat it with no cheats, it is very easy, i have come across some that were more difficult than that.

      » Comment by iplayzharder on December 9, 2011 @ 6:07 am
  18. @iplayzharder You played a different game. If you don’t believe you did, what was your solution? Wambooli couldn’t solve it and offered a $25 prize for someone else to solve it, but no one could. The FreeCell FAQ declared it to be impossible after exhaustive research. I quote, “11982 has now eluded solution by probably thousands of human solvers, and at least eight independent computer programs I am aware of (most of which are designed to search exhaustively for a solution), and I am confident in calling it impossible.”

      » Comment by Dan on December 9, 2011 @ 8:56 am
  19. If you were able to beat it, how would you ever prove it? Are you expected to remember each move and be able to repeat it?

      » Comment by Champswest on February 18, 2012 @ 6:53 pm
  20. @Champswest The easiest way would be to record the screen as you play the game.

      » Comment by Dan on February 19, 2012 @ 8:17 am
  21. hi have yall ever tried #-1 that i think is impossible!!!!

      » Comment by raldo on February 28, 2012 @ 12:34 pm
  22. I did not know that many people had already written about game no. 11982. I came across this game and was trying for two days, every time hitting the impossible position. Perhaps the reason is that four number sets are piled one on another in this set. They are Black 5, Red 9, Black 4 and Black 7.

      » Comment by Nagaraju on April 14, 2012 @ 4:00 am
  23. I’ve won games 1 thru 21605 to date but never won 11982. Is it still proven unbeatable?

      » Comment by Sam Williams on May 21, 2012 @ 9:39 pm
  24. @Sam Yup :)

      » Comment by Dan on May 21, 2012 @ 11:10 pm
  25. I doubt that any computer program searches exhaustively; I think that would take far more computational power than anyone has. But it still probably has no solution. What are the odds that there is a solution or class of solutions but only when done by one very precise and unexpected sequence of events? Usually there are a LOT of ways to do them, that’s why it’s so easy to stumble onto one of them.

      » Comment by fred on May 22, 2012 @ 3:01 pm
  26. @fred Most programs search exhaustively and at least eight independent programs were unable to find a solution. A game like FreeCell isn’t all that complex and with Moore’s law we have access to more and more of it. With today’s processors, a hundred million combinations can be tested in seconds.

      » Comment by Dan on May 22, 2012 @ 4:40 pm
  27. I have played every game up to 27730 sequentially and the ony one I couldn’t beat was 11982

      » Comment by RJGWho on June 12, 2012 @ 10:58 am
  28. @Fred Computer’s can be extraordinarily exhaustive at searching. All it takes is for someone with expert level skills at scripting, to create an Algorithm that would explore every possible move, every certain pattern, and every combination in and there of in a single scripting program. Even if there were a million different types of ways to attempt to solve it, a program can run through all of them. Trust me. Hackers do it every day to crack codes, algorithms, passwords.

      » Comment by JayOhhEeh on June 13, 2012 @ 8:33 pm
  29. Impossible FreeCell Game # 11982 other way

    http://www.youtube.com/watch?v=QZrnB0BgKA0

      » Comment by cool on July 22, 2012 @ 6:28 am
  30. We should start a collective fund that is payable should someone be able to beat it legitimately and prove it. $25 isn’t enuf, get it up to like $1000000 and we would get some attention!

      » Comment by rknell85 on July 27, 2012 @ 5:57 pm
  31. @rknell85 That’s a great idea. I would be willing to put up $1,000 for anyone who can beat that game using only legal moves (no tricks).

      » Comment by Dan on July 28, 2012 @ 10:36 am
  32. Just ran freecell 11982 through a computer program designed to beat these type games. After 20,000 attempts and no success, I pulled the plug. I believe the game is unbeatable,the only free cell game I’ve ever found that can;t be beat and I;ve been playing since 2000.

      » Comment by cattrick on January 17, 2013 @ 8:06 pm
  33. @cattrick Thanks for adding even more evidence to confirm our theory that game 11982 is unbeatable.

      » Comment by Dan on January 18, 2013 @ 8:56 am
  34. I wonder if these programs include moves where you would take from the foundation pile to help complete a black red black red pattern. As far as I know this is a legal move in freecell, albeit a bit unorthodox.

      » Comment by william on March 7, 2013 @ 2:10 am
  35. @William As far as I know the programs are extremely thorough and try all possible legal moves.

      » Comment by Dan on March 7, 2013 @ 2:14 am
  36. I won it after a few ‘f7′ undos to the beginning. The first time took me 40 minutes. The second time, ten. How do I get my $1000? LOL.

      » Comment by Keith on March 25, 2013 @ 7:16 pm
  37. @Keith That’s fantastic! Just send me the steps to beat the game. They can be written out, a screen recording, using Microsoft’s Problem Steps recorder, or whatever else works. I’ll gladly send you $1,000 (I’m not kidding).

    Update: As of 10 Apr 2014 Keith never replied to my request.

      » Comment by Dan on March 26, 2013 @ 11:38 am
  38. Obviously, one can easily place the cards into an unwinable hand; that’s trivial. Since unwinable hands have been proven by example to exist, it seems logical that a pseudo-random deal would, eventually, produce such a hand. If I were designing the game, instead of dealing randomly, I would start from a completed game and make a pseudo-random sequence of legal backward moves to arrive at the starting position. Such an algorithm could not generate an unwinable game. If it were able to generate #11982, then it would be solvable.

    At least one would be able to say with certainty that all games generated would be winable.

      » Comment by Steven Smith on April 6, 2013 @ 11:21 am
  39. @Steven Smith (#38) (commenting on my own comment)

    I should have thought about that one more deeply. One gains nothing by working backwards… it’s exactly the same complexity. Obviously, I can generate *some* board state by inverse plays. From there, my “deal” algorithm would have to solve the issue of getting all 52 cards onto the board; I’m guessing that this would prove to be equally as difficult as generating a solution in the conventional manner.

      » Comment by Steven Smith on April 7, 2013 @ 6:30 pm
  40. I’ve been playing FC for years and have even created a log of hard games. I’m stuck on #1941. Just want to know if anyone has won it. Usually I’ve gotten the solution at most 10 tries. I’m stuck after 10 moves on this one.

      » Comment by Will on April 24, 2013 @ 5:24 am
  41. @Will Here’s a video of how to beat that one. http://www.youtube.com/watch?v=1J2LOCVQhuI

      » Comment by Dan on April 24, 2013 @ 10:27 am
  42. Thanks Dan I’ll try a little longer before I view the answer. I’ll let you know which way I went.

      » Comment by Will on April 24, 2013 @ 11:15 am
  43. @Will That’s a tough one. I tried several times, then gave up and looked at the solution.

      » Comment by Dan on April 24, 2013 @ 12:59 pm
  44. I guess this is the one of the few I’VE BEEN LOOKING FOR!!!!!!(:-))
    Talk to you later.

      » Comment by Will on April 24, 2013 @ 5:07 pm
  45. I’ve just wasted the last two hours trying to beat this Freecell puzzle. I finally decided to search the net and see if anyone else has had as much difficulty with this game as I have… I wish I knew the game was “unwinnable” two hours ago. lol :)

      » Comment by Nat on May 12, 2013 @ 2:46 pm
  46. @Nat I know the feeling. I was stuck on the game for quite a while myself :)

      » Comment by Dan on May 12, 2013 @ 8:17 pm
  47. If it is possible to beat, there’s a very small chance that someone might solve it not knowing that it was thought to be impossible
    I doubt that will happen though

      » Comment by KoreusZ on August 18, 2013 @ 6:47 pm
  48. @KoreusZ I don’t deny there’s a minuscule chance, but it’s so small it’s hard to take it seriously.

      » Comment by Dan on August 18, 2013 @ 11:11 pm
  49. I’ve been playing Freecell eversince and just encountered this game today. In the pretext that all freecell games are winable, I just couldn’t solve this one. So I went to the Net only to find out that there are lots of guys who find it impossible to solve.

      » Comment by Manny on September 1, 2013 @ 11:00 am
  50. What does Microsoft, or more specifically the original author say on the subject? Did they generate the game sequences by working in reverse? If not, how did the exclude other “unwinnable” combinations and still retained 11982? I am a frecell addict as well, with a win percentage in the 90’s and have hite triple digit win streaks. I remember tangleing with 11982 a couple of years ago, but after the destruction it was doing to my my win percenage, I gave up and forgot about it till just now.

      » Comment by Birdsong on September 14, 2013 @ 4:19 pm
  51. @Birdsong I don’t know of any official comments from Microsoft or from the developers on how they generated the games. Sorry about the damage to your win percentage ;)

      » Comment by Dan on September 14, 2013 @ 10:51 pm
  52. The Freecell that came with Windows XP said that “It is believed (though not proven) that all games are winable.” I agreed with that since I’ve never lost one (thousands and thousands played randomly) until I came to the dreaded 11982 a few days ago. I tried for several hours to no avail. Though it can’t be PROVED impossible, the evidence weighs STRONGLY in that direction. No? One more vote for impossible from me.

      » Comment by Laura on October 22, 2013 @ 8:50 pm
  53. @Laura That’s interesting about Windows XP saying that. Thanks for the addition.

      » Comment by Dan on October 22, 2013 @ 9:17 pm
  54. There have been a few games that I have tried several times and given up on, only to go back to the following day and beat easily. BUT I have been working on 11982 off and on for maybe 10 years and still no luck.

      » Comment by Carol on November 7, 2013 @ 5:37 pm
  55. @Carol Thanks for yet another vote that the game is unbeatable.

      » Comment by Dan on November 7, 2013 @ 7:53 pm
  56. Thanks all, so glad I thought to look on line – I shall now give up with 11982 – though maybe just one more try :-)

      » Comment by Anne on November 9, 2013 @ 5:24 am
  57. @Anne No harm in trying ;)

      » Comment by Dan on November 10, 2013 @ 11:11 am
  58. play at 100% @1200 games but hit 11982 and wasted several days!
    Glad someone said I had not lost my stuff. moving on at at 100% any more

      » Comment by phil on November 25, 2013 @ 10:17 am
  59. @phil Wow, 1200 games!

      » Comment by Dan on November 25, 2013 @ 11:12 am
  60. with the aces so deeply buried instinct says this would be a very difficult game at best. i shall try it without any hope of winning!

      » Comment by gor on December 4, 2013 @ 10:01 am
  61. @gor That’s a great attitude. And if you can prove you won, I’ll give you a $1,000.

      » Comment by Dan on December 4, 2013 @ 2:08 pm
  62. Thanks for this post. I was stumped by 11982 for the last two days. A few Freecell versions ago (XP, maybe) I remembered reading that “all games are winnable.” I checked the “About Freecell” section of the game a few minutes ago and noticed the message was gone. Thanks for saving me some sanity. …But now I somehow feel the thrill of Freecell is gone. An unwinnable puzzle always makes you feel cheated. *Smiling*

      » Comment by David Bernard on December 4, 2013 @ 7:41 pm
  63. This is not the Microsoft version of freecell but I think I found one that is not beatable and if you can beat it please send me the way you did it my email is jmullins875@gmail.com… here is a link to my screen shot….. http://grab.by/sMxK
    I have tried to figure this out for about 2 days now and I have had no luck so I am hoping that someone that is better at freecell will not have any issues solving this!!!

      » Comment by J Floyd on December 11, 2013 @ 8:49 pm
  64. Thank You ALL for saving my sanity!!! Got snowed in for a couple of days and had game 11982 from the first moment I sat down to play. I even told my spouse that I didn’t think this was winnable. Now I can walk away without feeling the exasperating all-consuming need to win this one!!!

      » Comment by Pugs on December 16, 2013 @ 11:40 am
  65. @Laura : the problem is that “impossible” is not a political issue, but a mathematical one :-P

    What’s interesting me the most is not is this game is possible or not, but the way games are created.

    Freecell games are obviously created in a pseudo-random way, they’re two obvious way to create a game :
    – the way explained by Steven Smith, which can’t create an impossible game
    – The “stupid-one” : put the card in a random way on the board.

    The second one can obviously create impossible game, and 11982 might be one of them.
    However, it seems it’s the only not-sure-to-be-winnable game, for 32 000 games.

    Then we should look at the probability to create an impossible game with the second algorithm.
    If it’s about 1/32000 or less, then this algorithm could have been used by Microsoft to create freecell game.
    If it’s like 10%…

    I don’t know the solution of my question, however I think there are “many” impossible starts which could lead to a lose.

    According to me, there are 3 options :
    – I’m wrong about the “quantity” of impossible starts, then 11982 could be one of them
    – The Steven Smith algorithm is the good one, then 11982 is theorically possible, but could need like 1000000 plays to be win.
    – A non-trivial algorithm has been used to create freecell game.

    I’ll think about it… ^.^

    Btw, sorry for the mistakes I could have done, i’m absolutly not a native speaker ^^

      » Comment by Hoetre on January 31, 2014 @ 6:22 am
  66. I have been trying for two days to beat 11982. I have a run of 5400+ without a loss. In the end I Googled and found this site. Like so many others, I wish I had looked earlier:). I thought I would be in for a tough time with 11982 when I saw that all four aces were hidden either at the bottom of card stacks,or in the case of the ace of diamonds, one down from the bottom. It makes it nigh on impossible to start clearing the game of cards before they start blocking avenues of play.

      » Comment by Glenno on March 2, 2014 @ 8:42 pm
  67. I used to gleefully watch my own stats mount up until I found by accident that, when a game becomes obviously lost, you can just restart the PC without closing Freecell and, Hey Presto, the game is forgotten and your stats remain untouched. Now that the stats are less important I find each game is more fun for itself. Before I start I can set myself other tasks to achieve, such as making sure the aces line up in a particular order as I free them, or seeing how few cards other than the aces I need to remove before that final rush of cards to victory, (I have once managed only the 4 aces and a single 2). That can make a relatively simple game take on a more challenging aspect.

    Is that just plain too nerdy?

      » Comment by Albie on March 9, 2014 @ 3:32 am
  68. I’ve noticed a lot of comments here over the years saying things along the lines of “this can’t be proven impossible”. This is a mathematical problem, and as such absolutely can be proven impossible. And it has been.

    The simplest, and longest, way is simply to try every possible set of moves at every juncture until you reach a point where there are no moves left to make (or only moves that go in a loop, like moving the same card back and forth), and no solution was found. There are numerous freecell solving programs out there that do this (despite some concerns above about processing power, this kind of task is exactly what computers do well and far within the capacity of even an older computer to do quickly) and, as mentioned in the article, many have been applied. So there is no clever solution waiting that simply hasn’t been found, because literally everything has been tried, multiple times.

    So let’s be clear; this puzzle isn’t highly improbably, likely impossible, or nigh impossible. It is simply impossible.

      » Comment by Adrian on April 10, 2014 @ 3:53 am
  69. @Adrian Without having seen or run the programs used, I can’t say with 100% certainty that it’s impossible, but you’re probably right. However, I’ve never seen incontrovertible proof of your claim. The chances of a solution eluding millions of attempts is certainly low, but I would like to see a mathematical proof that the starting configuration of #11982 makes winning the game impossible.

      » Comment by Dan on April 10, 2014 @ 12:29 pm
  70. @Dan In that case, surely you can easily resolve this by running some of those programs yourself and seeing what happens? The Freecell FAQ linked above mentions several I believe. Otherwise, arguing that “we don’t know it’s impossible because maybe every human and program has made a fundamental error that would otherwise to make it solvable” sounds about as useful as “we don’t know that 2+2=4 because maybe every person and calculator has been making the same error”, without ever actually examining a calculator. It can be considered possible only by dismissing a truly staggering amount of evidence to the contrary.

    I’m half tempted to write my own brute-force solver and put up the results and source code for you.

      » Comment by Adrian on April 12, 2014 @ 10:58 am
  71. I get what you’re saying, but basic math and Freecell gameplay simulation aren’t equivalent. Really, we’re splitting hairs here. I’m just saying there’s a minuscule chance someone missed something, and you’re saying there’s no possible chance of error.

      » Comment by Dan on April 14, 2014 @ 9:34 am
  72. on the question of 11982 if you shuffle a pack of cards you are doing it in such a way that no person living or dead has ever done before or will ever do so in the future taking this in to account would it not be foolish to conclude that there are no more 11982s lurking out there

      » Comment by andy on April 14, 2014 @ 11:02 pm
  73. @Dan But that doesn’t make sense in a limited domain like this. There are an absolute number of possible… you know what? That’s it. My task for tonight (well, OK, might take a few days); write a brute-force Freecell solver, with source (because I guess you need to see the innards to be super certain,) to demonstrate, definitively, that there is no, no, NO solution to this puzzle. Stay tuned.

      » Comment by Adrian on April 15, 2014 @ 2:27 pm
  74. I don’t feel the difference between “possible” and “impossible” is “splitting hairs”, I think it’s a fundamental issue in the discussion. With that in mind, and because I’m a money-where-my-mouth-is kind of guy: https://drive.google.com/folderview?id=0B1j_Jk2qDT_4QnVRWkRrcGdrU0U&usp=sharing

    Welcome to cell11982, a brute force freecell solver, by default set up to test the impossible puzzle, 11982. Three files are at that link. cell11982.py is the python program. Python programs are compiled on-the-fly, which means the program is the source code; you can open it in any text editor and see exactly what the program is doing. Source Explanation is a more detailed explanation of what the program is, how it works, and what it means.

    11982 results.txt is the (rather large) results file. You will probably want to click the download link in the bottom-right corner and view it on your computer, rather than wait for Google’s preview of all 36+ mb to load up. This lists, literally, every possible layout of cards you can reach for 11982, and (if you wish) can trace them back to see what moves were made to get there.

    It doesn’t even get close.

    There are paths that lead to two hearts being sent home, or two diamonds… but these are exclusive. Getting the ace of one puts you in a position where you cannot reach the ace of the other. Clubs and spades? Forget it; you can’t even get started.

    But don’t take my word for it. Check the layout I’m using, peruse the code, run it yourself, browse the results. Set up some solvable puzzles and see if it gets there, or put in obviously impossible ones and see if it throws them out. If something’s been missed, then good news; you only have to check a few hundred lines of code (included comments explaining what’s going on) to find it rather than well over 80,000 individual board layouts, a relatively simple task.

    For this puzzle to be possible, either I’m not correctly testing for every move that can be made on a layout, incorrectly discarded valid moves, or not sending every valid layout to be tested. If I’m not doing any of those things with this program… and anyone can see for themselves whether I am… then all possible minuscule avenues have been covered, and the puzzle is impossible.

      » Comment by Adrian Wood on April 18, 2014 @ 2:23 pm
  75. @Adrian Nicely done. I am really and truly impressed you did this.

    You’ll probably think I’m just being stubborn, but I maintain that you have only proven your program can’t solve the game, not that the game itself is unsolvable. You haven’t proven you’re correctly testing for every move, or not incorrectly discarding valid moves, or that you’re not testing every valid layout to be tested. You claim you didn’t omit anything as a fact, then ask others to prove you wrong.

    In your comment, you were approaching the kind of proof I was looking for by discussing a step that needs to be reached in order to win (like getting the Ace of clubs or spades out), then showing how achieving that is impossible (easier said than done), which could then be used to prove the game is impossible. But I have yet to see such a proof.

    Writing a brute force program is useful, but as a fellow developer, we both know bugs can easily sneak past watchful eyes. (e.g. Heartbleed…). Can you prove your program (or any of the other solvers) don’t have a bug?

    Using Proof by Exhaustion is probably the best approach for proving this Freecell game impossible, but even mathematical proofs have been later proven incorrect.

    After trying yours, I downloaded a few more solvers – FreeCell Pro, Don Woods’ c program, FreecellJSolver and this online version. Of course they all found 11982 to be impossible.

    And as a side note, for anyone who thinks they can solve 11982, download FreeCell Pro and play the game so it will log your moves.

      » Comment by Dan on April 18, 2014 @ 4:28 pm
  76. @Dan But, again, it’s a limited domain puzzle, so we the statement “you have only proven that your program can’t solve the puzzle” doesn’t make sense. Either it can solve Freecell, or it can’t. We know that no legitimate Freecell solution is going to involve moving two cards at once, or having a 5th freecell, or a 9th column, or starting a home cell at King and working our way down. It’s not like heads or tails, where someone may cry “what about the edge?”. In Freecell, nothing is hidden. We know exactly what’s possible, and if we’re testing for that, then we have our answer.

    Bugs can sneak past but, as a developer, you know we can test by example. Oh, sure, if a bizarre input comes into play maybe some weirdness will occur, I daresay you can trip my program up by substituting cards with strings of random characters, but that’s not relevant to this discussion: we have one set, simple layout, and that’s what we’re testing. We don’t have to worry about complicated maths or string handling messing things up, because it’s not here. We know it doesn’t crash out part way because… well, firstly, because I’ve run it fully, and second, because you can too, so it definitely runs well enough to reach the end. The example test board included in the source (an almost complete board) takes less than a second to run, useful to double-check if the program will run to completion (without exhaustively testing each element on the way). As such, we can see that, at least, the system as a whole works correctly sometimes. We’re now into unit testing, seeing if each part of the machine acts the way it should.

    If my program (or any program) can be demonstrated to make each individual type of move possible, for example, then we can cross this off the list; there isn’t a bug causing particular kinds of moves to not be made. It’s possible that, for example, an off-by-one error could mean the system was ignoring the last column, but we can easily identify that. I know that because at an early stage, the program *did* have that same error, and I did catch it. In the results we can see cards are moving to and from every column and freecell area. So that’s one factor we can tick off, we know it’s not ignoring any position. The same way, we know it’s moving to and from the top card of each column, that it can take the bottom card of a column and repopulate an empty one, that it’s making use of all 4 freecells and that it can, at least with hearts and diamonds, move cards home.

    “Ahhh”, I can hear you thinking, “but what if it can’t move ALL cards home? Maybe it’s stopping at 3?” And I fall back to the same reasoning; if the input is valid, the process isn’t falling apart at the extremities, we can see it’s working in some places, and we can’t see any place that it’s wrong, then it’s right. In this case, we can look at card list in the source and see if there are cards missing, or in the wrong order, or check for indexing errors. But that’s all; once that’s all there, we can tick this off, it works.

    Perhaps it’s missing layouts? Afraid not. The way the program works means the list of boards is initially populated by just one; the starting board. Generally, we’d expect an error resulting in the first or last element of the array being skipped or going past the end of the list. But here we know it isn’t going past the end of the list (because otherwise it would crash out every time you ran it), and we know it isn’t skipping the first/last element because, at the very start, it only has one; skipping this would mean the program would never run. And we know it isn’t skipping out on saving elements along the way from way back in that first bug; we can see it saving every move in the results, and we can see moves being made from each layout thanks to the “parent layout” tracker (a late addition in development, but I’m glad it’s there). This is one of the easiest things to check, in fact; just see what moves can be made from the initial layout, and these will be the first set of moves in the results. (You can do the same check at any point, but bear in mind some possible moves may have resulted in a duplicate layout of an earlier one and so be shown earlier, with a different parent).

    Remember: when we’re saying “maybe there’s a bug”, we aren’t just suggesting a bug is there in my program, but that *every single freecell automated solver and human player, ever, has been making the same error, on just this puzzle* (amongst the first… 10,000? Is that correct? I know later version of MS Freecell had higher numbers of puzzles with more unsolvable ones). Likewise, when I say “you can see for yourself”, this wasn’t me saying that it might have bug and I wanted someone else to test it… it was me saying I *had* looked for bugs, weeded them out, but that I don’t expect my word to be good enough.

    But I’m going to go one further here. My email address is right there in the source. So if you still feel the burden of proof is on my end (and I’d argue I’ve submitted plenty at this point), then tell me what tests you’d need to be certain. Do you want a running tally of how often cards are moved from each area, to eliminate the possibility something is being missed? Perhaps you’d like it to track how often each card is moved along with how often each card is on the bottom of a column, to identify if cards are being revealed but never moved? Or maybe you just want full test boards, to prove that it can move every card, and is checking everywhere? Whatever it is, and this is an open invitation, my email is right there. Tell me what proof you need.

    For now though, I am confident in saying once more; this puzzle is impossible.

      » Comment by Adrian Wood on April 19, 2014 @ 3:17 am
  77. Man oh man that’s a comment and a half! :)

    My point isn’t about bugs in your program. I didn’t find any bugs either, or in the other solvers. But even if we had a billion other solvers all declaring the game impossible, a valid solution would prove them all wrong.

    Albert Einstein said, “We can not solve our problems with the same level of thinking that created them.”

    In other words, if our knowledge of FreeCell isn’t as advanced as we thought, and we missed something, then a solution that we had never considered could be found.

    My argument can be summed up thusly.

    Not being able to find a solution is not proof a solution doesn’t exist.

    However, with that being said, I concede FreeCell falls within the bounds of an NP-Complete problem and can therefore be computationally solved with brute force analysis as you have done.

    In short, I too am confident that game 11982 is impossible.

      » Comment by Dan on April 19, 2014 @ 10:12 pm
  78. So far, I’m at 3,000 straight wins without a loss. I really don’t look at the game # so I couldn’t even say if I’ve ever come across game #11982. Will look for it from now on.

      » Comment by Jeff Younger on April 20, 2014 @ 6:07 pm
  79. I have Windows 8.1 and found another one I believe to be unbeatable: #6362856.

      » Comment by WillH on May 2, 2014 @ 10:58 pm
  80. @WillH Assuming this is the same game you’re referring to, my FreeCell solver says it’s winnable.

      » Comment by Dan on May 3, 2014 @ 8:44 pm
  81. I have 4651 wins and 0 losses. came across game 11982 and as always, before I make a move I mentally drill each stack down to the bottom. It didn’t take long before I realized 11982 was going to be a tough one. Now I realize probably impossible. I discovered this forum and it confirmed my belief. My little game is to not use a free cell to win a game. Of my wins I was able to win 62 games (Being a baseball fan, I call them perfect games) that way. And yes, I do have a girlfriend,, play a shitload of golf and have a decent tan.

      » Comment by Alex Giuliani on June 20, 2014 @ 4:39 pm
  82. Thanks to Dan, and to Adrian for his computer-generated proof of the insolvability of 11982 using the method of exhaustion. Freecell is much simpler than chess, which is not solvable in such a way (or any way, for that matter). I have a long list of freecell games that I found “difficult”. For me, 17993 had been the most recalcitrant. However, after about 50 tries at 11982, I consulted the Internet. Thank goodness! One shorthand of the play of the game would be to give the origin and destination of each card. So moving a card from col 1 to the holding zone might be written 10, or moving a card from col 7 to its appropriate suit 79. So a game might begin 59 18 82 and so on. This is clearly a unique and complete way of writing the game. Incomplete but suggestive would be simply to use the column of origin: the same game would start 518 in this notation. Might use 0 for promoting from the holding zone. This latter method would be more in the nature of a hint than of a replay.

      » Comment by Mint Julep on July 7, 2014 @ 9:58 pm
  83. “Clearly” is a kiss of death. A unique notation would differentiate the four places in the holding area, say as a, b, c, d. Sorry for the slip.

      » Comment by Mint Julep on July 7, 2014 @ 10:26 pm
  84. I thought I could beat them all. Not this one. It’s nice to know that somebody else couldn’t beat #11982.

      » Comment by John Boyer on August 13, 2014 @ 10:29 am
  85. I use the android freecell game (on my phone) from http://www.odesys.com and have played every game consecutively starting at game 1 and have won all 11981 games (using undo as needed)…until I hit 11982. Usually, if it takes more than 30 minutes (like one in a thousand games), I look up a solution online. I could see “by inspection” that 11982 was “probably impossible” because of the location of the aces and the stacks above them, and after trying to solve it for about 30 Minutes.

    Your posts here have been very helpful in convincing me to go ahead and ruin my 11981-game winning streak . I don’t think there is any way to restart my phone or shift-f10 or anything similar to work around this issue for this android version of FC, but if anyone out there knows of a way to avoid ruining my 11981-game winning streak, please let me know. I guess I can just “start over” consecutively again at game 1 and remember to skip 11982 when I get to it next time.

    I hope I remember to skip over 11982 next time. I will stop playing for a few days in case a work around gets suggested here.

    Thank you everyone for your dedication!

      » Comment by Albert Einstein Lassiter on August 20, 2014 @ 12:47 pm
  86. @albie I am also a nerd. I managed only two aces and one two.

      » Comment by Tundra Nugget on August 22, 2014 @ 11:25 pm
  87. With no suggestions on how to prevent ruining my 11981 winning streak (playing consecutively numbered games starting with number 1), I will just reset my game statistics and “start over” with game number one, and plan on remembering to skip game number 11982 after I win the first 11981 games again. That should only take a few months or so. Again, I appreciate everyone’s dedication to the game.

      » Comment by Albert Einstein Lassiter on September 18, 2014 @ 11:15 am
  88. I cannot beat #1941

      » Comment by Howard on September 24, 2014 @ 5:54 pm
  89. Here’s a video of the solution: https://www.youtube.com/watch?v=1J2LOCVQhuI

    And the steps to beat it. Numbers = column number, letters = 4 free cells (a-d) and h = the home cell.

    42 87 1d 1c 15 c5 3c 3b 3a 2h
    5h 15 16 37 83 81 c1 7c 73 c3
    7c 7h 71 73 71 21 23 87 5h 53
    52 32 38 28 23 83 82 58 5h 2h
    1h 68 67 17 81 28 c2 1c 71 c7
    2c 26 c2 26 24 3c 38 64 c8 5c
    52 65 45 46 56 54 a5 64 6a 65
    a5 2a 26 45 42 52 54 24 25 45
    42 52 54 a6 24 25 1a 12 a2 1a
    16 45 41 51 54 a6 14 15 45 4a
    41 a1 2a 26 51 52 12 15 a6 25
    21 3a 32 35 31 51 53 21 a1 6a
    62 a2 7a 76 26 27 12 67 6h 1h
    13 15 35 31 51 53 13 15 35 31
    63 6h 51 5h 4h 41 21 72 73 23
    72 76 26 37 a3 4a 32 36 76 87
    81 71 b7 6b 26 d2 3d 63 b6 2b
    a2 6a 36 b3 8b 68 a6 2a d2 6d
    62 d2 6d 64 d4 2d 24 d4 2d b2
    1b 12 b2 4b 84 d8 4d 48 b8 3b
    83 d8 7d a7 1a 17 a7 1a 15 a5
    1a 1h dh 6d 61 d1 6d 38 23 27
    2h 7h 62

      » Comment by Dan on October 8, 2014 @ 12:40 pm
  90. Have been playing for years but just now finding this site. Enjoyed reading all the comments esp regarding the elusive 11982. I mean the impossible 11982

      » Comment by Candice on October 12, 2014 @ 5:49 am
  91. Okay, I’m not at the skill level or competence of you guys, but I’ve been stuck on 6911962 and just can’t see any way of completing it. Can anyone give me a heads up on how to do it?
    Many thanks!

      » Comment by Shaz on October 20, 2014 @ 10:57 am
  92. Here’s a solution from the FreeCell solver:

    Numbers = column number, letters = 4 free cells (a-d) and h = the home cell.

    3h 54 48 4d 3c c3 4c 46 3b 36
    64 6a b4 35 6b 63 b3 a3 5b 65
    67 63 43 4a 68 36 34 a3 63 2a
    26 34 a6 2a 21 c1 51 b5 2c 8h
    23 4b 42 b2 4b c4 1c 51 c5 2c
    32 c3 6c d6 86 8d 84 82 82 8h
    dh 1h 51 5d 51 b4 87 57 5h 5h
    4b 48 45 85 b5 4b 41 c1 7c 78
    74 84 c4 1c 18 c8 2c 23 c3 2c
    25 c5 1c 12 c2 1c 14 c4 2c 24
    c4 1c 12 17 a1 71 7a 4h 3h 71
    c1 71 d1 41 4d 4c dh ch 73 1h
    1d 14 1c 17 18 16 12 a1 81 71
    c1 41 4c 4a 46 d1 a6 c6 3d 35
    3h 5h 6h 3h 1c 1a 16 18 17 12
    14

      » Comment by Dan on October 20, 2014 @ 1:18 pm
  93. Thanks Dan! Very happy now I’ll be able to do it :)
    Have a great day,
    Shaz

      » Comment by Shaz on October 21, 2014 @ 4:22 am


Post a comment










Number of readers
- home
news links
other links

about me
blog archives
docs
entertainment
experiences
funny lists
humor
intellectual
interests
music
opinions
photos
web designs
tools
software
webmaster help

Become a Patron


Most Popular Pages


Sign up
Enter your email address to be notified of new blog posts



Search blog archives

Calendar
February 2009
SMTWTFS
« Jan Mar »
1234567
891011121314
15161718192021
22232425262728







Random quote
Better to write for yourself and have no public, than to write for the public and have no self. -- Cyril Connolly

(See all the quotes)



Say hi on








(?) Choose theme:  X X X X X X
Page loaded in 0.2615 seconds