Minesweeper Bugs & Oddities

Bugs that have been found in Microsoft Minesweeper

Oops! What happened to these boards?

-No, those pictures aren't fake! These boards were produced using a very easy-to-duplicate glitch; all you need is a mouse with a middle button. Hold down the middle button on a cell and then click, and the cells around the click will stay pressed. This glitch works in all versions of Minesweeper, including the XP version.

-There is very frustrating and obvious timer starting glitch found in the Win 95 and Win 98 versions - it often advances from the first second too quickly, cheating you out of some time! Often the timer will advance from 001 to 002 almost right after your first click. It was fixed in the XP version.

-When your first click is on a mine, it moves the mine to the top left-hand corner. (If that is already occupied, it tries the cell beside that until it has it placed.) This creates slightly biased boards, where the top left hand cells are more likely to contain mines. By using the "Show Mines" option in my version, you can see that my version re-randomizes the first mine if you click on one.

-Sometimes, the Windows version will not stop the timer on one-click boards, even though only mine cells remain. Therefore, these boards seem impossible to finish. For example, take a look at this board I encountered - it should be finished! (However, it did finish itself when I right-clicked on a single mine...) Apparently, this glitch has also happened to other people - Yoni Roll - Matt McGinley 1 - Matt McGinley 2. Notice anything similar about each board? Each board has a mine in the top left hand corner. I believe this glitch has something to do with the above glitch - when your first click is a mine, it moves it to the top left corner. For some reason, the movement of the mine causes Minesweeper to not recognize that the board is cleared on the first turn (most likely a programming oversight). Matt McGinley has a nice web page about this bug.
[Update: Using the below glitch I attempt to prove this theory.]

-By editing the winmine.ini file, you can trick minesweeper into loading custom boards that it normally it wouldn't let you. You can even play a board that has a mine in every square! If you try to click a on a square, the timer starts but it wont open up any squares. In fact, the "button depressed" picture is displayed on the cell, even when you let go of your mouse button. (But the cell remains unclicked... you can get it back to its unclicked picture by clicking on it and dragging.)

-Some people have noticed that some Minesweeper boards repeat themselves! Statistically, this is very unlikely to happen. Even in the beginner mode, there are over 150 billion different combinations. In expert mode, there are over a googol (10100) combinations! (That's more atoms then there are in the universe.) If n reprsents the number of squares on the grid and m represents the number of mines, then:
Number of different boards = n! / [(n-m)!m!]
Using this formula, we can determine that there are exactly:
Beginner boards: 151,473,214,816
Intermediate boards: 104,921,249,538,770,987,879,991,995,968,495,941,124,675,048,800
Expert boards: 560,220,999,337,421,345,429,058,985,775,821,108,059,290,502,723,897,901,281,458,809,527,214,479,570,631,168,198,385,673,295,159,633,481,600
Or in scientific notation:
Beginner boards: 1.51 x 1011
Intermediate boards: 1.05 x 1047
Expert boards: 5.60 x 10104
Matt McGinley has an interesting web page about this, with several examples of even intermediate boards that have duplicates. With the huge number of boards that are possible, there is no way this should ever happen if the boards were truely random. (I wonder if there is a problem with the pseudo-random number generator?) Jon Simonsen has written a good article on the randomisation of boards, and states that there are only a limited number of boards that are re-used by shifting them a certain number of squares.