Mines, minesweeper with a cash-out button

With 3 mines on 25 tiles, your first pick is safe with probability 22/25. Surviving two picks is 22/25 times 21/24; each further pick multiplies in the next fraction. The fair multiplier after k safe reveals is simply one divided by that survival probability, and the game pays a touch less than fair, scaling the whole ladder by its published edge, typically around 1% on crypto originals.

That structure has a clean consequence: the edge you pay is the same whether you cash out after one pick or push for the board. Cashing out early is not caution rewarded and pushing on is not greed punished; both are the same bet at different volatility.

The mines are placed by the seeded random generator before your first click, uniformly across the grid. Corners, edges, the tile that has been safe five rounds running: every unrevealed tile carries exactly the same probability of hiding a mine. Patterns in where you click change the experience, never the odds.

The mine count you choose is the real control. One mine makes long, gentle ladders; twenty makes a near-coin-flip for a large multiplier on the first click. Every setting carries the same expected value, so the choice, as in Plinko, is about the shape of the session.

Mines feels different from a slot because stopping is your decision, and near-misses are tiles you almost clicked. That sense of agency is the product. It is worth naming plainly: the agency is real for volatility and stake, and worth zero for expected value, which was fixed the moment you chose the settings.

On crypto originals the round is provably fair: the mine layout derives from a hashed server seed committed before your picks, so you can verify afterwards that nothing moved under your clicks. As always, the verification proves the maths was followed; the maths includes the edge.