The computational complexity of Minesweeper

We show that the Minesweeper game is PP-hard, when the object is to locate all mines with the highest probability. When the probability of locating all mines may be infinitesimal, the Minesweeper game is even PSPACE-complete. In our construction, the player can reveal a boolean circuit in polynomial time, after guessing an initial square with no surrounding mines, a guess that has 99 percent probability of success. Subsequently, the mines must be located with a maximum probability of success. Furthermore, we show that determining the solvability of a partially uncovered Minesweeper board is NP-complete with hexagonal and triangular grids as well as a square grid, extending a similar result for square grids only by R. Kaye. Actually finding the mines with a maximum probability of success is again PP-hard or PSPACE-complete respectively. Our constructions are in such a way that the number of mines can be computed in polynomial time and hence a possible mine counter does not provide additional information. The results are obtained by replacing the dyadic gates in [3] by two primitives which makes life more easy in this context.

💡 Research Summary

The paper conducts a thorough investigation of the computational complexity of the classic puzzle game Minesweeper. It treats two distinct decision problems: (1) determining whether a partially revealed board admits at least one consistent placement of mines, and (2) locating all mines with the highest possible probability of success. The authors extend earlier work by R. Kaye, who proved NP‑completeness for the solvability of partially uncovered square‑grid boards, by showing that the same hardness holds for hexagonal and triangular grids. Their reduction encodes Boolean variables in individual cells and uses the “number of adjacent mines” clue to simulate logical gates. Crucially, instead of the dyadic gates used in previous constructions, they employ only two primitive gadgets—an AND‑like gadget and a NOT‑like gadget—to build arbitrary circuits. This simplification works uniformly across all three grid types, establishing that the solvability problem is NP‑complete regardless of the underlying lattice.

The second line of inquiry addresses the probabilistic optimisation problem: given a board, what is the maximum probability with which a player can uncover every mine? The authors design a board in which the player first selects a square that, with 99 % probability, has no adjacent mines. If this “lucky” start succeeds, the player can, in polynomial time, reveal the entire Boolean circuit hidden in the board. The circuit’s output determines the exact mine configuration, so the player’s subsequent task reduces to guessing the correct assignment to the circuit’s inputs. This guessing problem is equivalent to the PP‑complete problem MAJ‑SAT (deciding whether a majority of assignments satisfy a formula). Consequently, finding the mines with maximal success probability is PP‑hard. Moreover, by embedding quantified sub‑circuits (∀∃ structures) the authors create instances where the optimal success probability is infinitesimally small; these instances are shown to be PSPACE‑complete. Thus, depending on the probability threshold, the Minesweeper optimisation problem jumps from PP‑hard to PSPACE‑complete.

An additional contribution is the observation that the total number of mines can be computed in polynomial time from the board layout, meaning that a built‑in mine counter does not give the player any extra advantage. The paper therefore isolates the intrinsic computational difficulty of the game from auxiliary information.

Overall, the work paints a layered complexity landscape for Minesweeper: (i) the existence‑check for a consistent mine placement is NP‑complete; (ii) the problem of locating all mines with the highest achievable probability is PP‑hard; and (iii) when the optimal probability may be arbitrarily small, the problem becomes PSPACE‑complete. By replacing complex dyadic gates with two simple primitives, the authors achieve a cleaner, more general reduction that applies to multiple grid geometries. Their results demonstrate that a seemingly simple puzzle can embody the full spectrum of classical complexity classes, offering a compelling bridge between recreational gaming and theoretical computer science.