Making Explainable Minesweeper
27 comments
·July 6, 2025MostlyStable
mtlmtlmtlmtl
I looked at the code for this once. It's sort of a hack. It generates a random puzzle, tries to solve it with a deterministic backtracking(IIRC) solver that detects situations where it's stuck. If it is stuck, it shifts some mines around in the place where the solver got stuck, according to some heuristics, and keeps solving. It keeps doing this until the entire puzzle has been solved, verifying that the puzzle can be solved without any guesswork.
I will second the recommendation. As someone who's wasted too many hours on minesweeper, it radically changes the game. Because I know there's a logical solution, I'm able to find patterns that I never found in the original, because I'd often assume it was just one of those guesswork situations. It's similar to how chess puzzles are easier than an equivalent position during a game. Because you know there's a neat solution, you're more willing to put in the legwork of searching for it. Learned a lot about the game simply by playing Simon's version.
anewhnaccount2
I wouldn't describe it as a hack necessarily. This is called rejection sampling, and it's just fine as long as the rejection rate is low enough.
mcmoor
I tried some no-guess variants but somehow there's a kind of sterile feeling to it. For some reason it makes it more boring compared to vanilla variant. It of course excels when I'm getting frustrated at guesses and 50/50, but otherwise I play the normal minesweeper more.
eviks
Unfortunately it doesn't have proper controls and cell sizes (doesn't expand to the screen, so remain small), so it's easy to mistap on a wrong cell or mistap instead of mishold. Basically, controls aren't optimized for a touch screen
gavinsyancey
I made something like this a while back: https://github.com/g-rocket/minesweeper
zabzonk
> anyone who likes puzzles on their phone
it's also available on windows and android - very good.
kevindamm
It's also available in HTML/js/wasm, so practically all networked devices:
https://www.chiark.greenend.org.uk/~sgtatham/puzzles/js/mine...
However, the minesweeper puzzle is the one that causes a big wart in the API, which part of me has wanted to refactor for years. (it's for the reason given in GP comment, the puzzle isn't generated until first click; it's the only one with this behavior and could be reworked to have the clicked area be a parameter to creation, but then it would have a slightly quirky main()-equivalent function.. IMO worth it for the cleaner API)
Simon's puzzles have overall been an inspiration, some good reading if you're into algorithms and applied graph theory and don't mind that it's in C.
Unearned5161
Nice article, I wanted to point out though that your example board at the beginning has no such ambiguities or 50/50's. Here's a recreation of it so you can see for yourself, if you open it up from the bottom left as you did, the 1 2 1 leads to the board being solved rather quickly:
https://agustinfitipaldi.com/minesweeper?seed=eyJ3aWR0aCI6OS...
greentec
Yes, that's right. I'll have to update my blog with additional information about this. Thank you!
rtpg
Many many years ago I remember playing a free minesweeper that had, among other things, hexagonal maps as an option, but the big thing was that it guaranteed that you would never need to guess. You could always figure out the answer, and I believe there was a mode where if you clicked in a way where you couldn't have deduced the answer it would put a mine in there to punish you.
Anyone know what minesweeper software I'm thinking of?
bigstrat2003
Apart from the mine theme (and punishing you by putting in extra mines), that sounds like Hexcells. Very reminiscent of minesweeper, and you never have to guess.
rtpg
what I'm thinking of is, I believe, much older. I remember playing it before 2010 at least based on the machine I was playing it on. And it was distinctly "windows" styled: it looked like windows minesweeper!
GLdRH
In minesweeper online there's also a "no guessing" mode which, in my opinion, is much more interesting than the normal mode. It means that if you guess in a situation where you could have deduced the mines, you always get a mine. Conversely, if you really are in a guessing situation, you will never get a mine. I'm pretty sure the game calculates the unknown part of the map after each click anew.
null
greentec
Hello Hacker News!
Thank you for your interest in my previous post. This time, I've written a blog post about the game and the process of creating it.
In the original Minesweeper, there are inevitable 50/50 moments where you have to rely on luck. In the game I created, 'Explainable Minesweeper,' I eliminated these guessing situations. However, I also prevented the maps from becoming too easy! How? By using logical deduction, you can solve puzzles that initially appear to be luck-based. The blog post explains the process in more detail.
nojs
The first example is not ambiguous. You can solve it starting at the bottom left
red369
Do you mean the first screenshot of Windows Minesweeper (the left half of ems02.png)?
Could you explain how? I haven't played a huge amount of Minesweeper, but I can't see how that can be solved without risk/luck.
nojs
On the left side, the 1 at the bottom means that the third from the bottom must be a mine (since the 2 can't point to the bottom two). The 1 touching that mine then excludes a bunch more, and that's sufficient to solve the rest.
red369
Thanks - I finally saw it just as you replied. I agree that it looks like that mine layout is the only possible one which satisfies the vertical 1,2,1 boxes. I need to play more :)
greentec
Thank you. I didn't see that!
npinsker
For what it's worth, the way 14MV does hints is probably by just throwing the board into Z3 (https://github.com/Z3Prover/z3) or some other constraint solver. Microsoft has already done all the hard work for you.
jsnell
I doubt it. A guaranteed-solveable minesweeper scenario isn't just about the global board setup, but about what information is available when, and what order the solution is gone through.
Bombe[0] is to my mind the definitive exploration of this concept. The tagline is "Minesweeper, but you only solve each situation once", which you do by writing these kinds of deduction rules with a fairly painful visual programming language. (You can't write an invalid rule: the game will detect the logic error and present you with a counter-example.)
You then let the computer churn through it's list of 100k scenarios idle-game style, until you bump into a board that can't be solved with the rules you provided, and you have to figure out what new rules to write.
As the game progresses, you'll unlock ways of parameterizing the rules in various ways, as well as various variant rules.
null
npinsker
You're right, I misspoke a little bit -- 14MV is able to do this because (unlike this game) it does not reveal all givens to you at the beginning.
shkkmo
The exclusion of patterns that involve more than 2 numbers is a pretty huge caveat that should be mentioned earlier and more clearly. When I was playing a lot of minesweeper, larger levels tended to require solving larger patterns most of the time. If you exclude those solutions, your estimate of how often you are required to guess is going to be pretty inaccurate.
joshka
xyzzy shift+Enter... ;P
Simon Tatham's Puzzles app has a minesweeper version that both A) generates on first click so you are guaranteed to never hit a mine on the first click and B) every board is 100% solvable with no need to guess. I have no idea what method it uses. I believe that _every_ puzzle in it is 100% solvable with no need to guess. It's also add free and completely local. It's a great app for anyone who likes puzzles on their phone.