How the game is actually won

Not all points are equal

The board has twenty-four points and sixteen possible mills. Because sixteen mills of three points fill forty-eight slots, and there are twenty-four points, every point on the board lies on exactly two mills. There is no square that belongs to three. So the usual advice to “take the point that is on the most mills” means nothing here: they are all the same.

What separates the points is how many neighbours they have, and that is wildly uneven. Four points — d6, b4, f4 and d2, the midpoints of the middle ring — touch four others. Eight points touch three. The remaining twelve, the corners of each square, touch only two. A corner piece has two ways out, and if both are covered it is furniture.

This matters more than it looks, because the second phase of the game is about movement and the third is about not being trapped. Pieces you put on corners during placement are pieces you may not be able to use later.

Do not rush the first mill

Beginners place to make a mill as fast as possible, take a piece, and feel ahead. They usually are not. A mill formed in the first four moves announces exactly what you are doing, costs you the flexibility of the pieces now locked into it, and buys one piece from an opponent who still has eight in hand.

The stronger habit is to place so that you threaten two mills at once. A single threat gets blocked. Two threats that share a point cannot both be blocked, and the piece your opponent spends blocking one of them is a piece not developing their own position.

The running mill

This is the single tactic worth learning properly, and most explanations of it are wrong. Two mills that share a piece are nearly worthless: moving the shared piece breaks both at once, so the outgoing move closes nothing, and you take a piece only on the return — assuming the shared piece has anywhere to go at all, which in a crowded position it often does not.

What you want is two mills that share no piece. Each is missing exactly one point, and those two missing points are adjacent. One piece then shuttles back and forth between them, completing a different mill on every single move. Five pieces build one. An opponent who cannot occupy one of the two points, or break one of the mills, is losing a piece per turn and the game is already over.

The defence is to see it two moves early. Once both mills are built and the shuttle square is empty, there is usually nothing left to do — occupying the square costs you a piece to a mill you cannot prevent.

Count moves, not pieces

You lose immediately if you have no legal move, however many pieces you still own. This is not a rare technicality: on a board where twelve points have only two neighbours, suffocation is a normal way to lose.

So in the moving phase, the number to watch is not the piece count but how many legal moves each side has. Seven pieces with two moves is a lost position against five pieces with eight. If you are ahead on material, trading pieces helps you; if you are ahead on space, taking away your opponent's last exits wins on its own, and you never have to form another mill.

Three pieces, and flying

A player reduced to three pieces may move to any empty point rather than only an adjacent one. That changes the endgame completely: a flying side cannot be blocked, and it can answer any threat from anywhere on the board. Three against three, with both sides flying, is usually a draw.

The practical consequence is that if you are grinding an opponent down, the moment to be careful is when they reach four pieces, not three. Taking them to three hands them the flying rule; leaving them on four, immobile, does not. Local games and games against the computer let you turn flying off, which is worth doing once to see how much of the endgame it was holding up.

The mistakes that decide most games

• Taking a piece from wherever is convenient. Take the piece that is one move from completing something, or the one whose removal leaves its neighbours stranded — never the one that is simply easiest to see.
• Filling your own corners during placement, and finding in phase two that half your pieces have nowhere to go.
• Breaking a mill for no reason. You get to reclose it, but only if the point you left is still empty when you come back.
• Blocking every threat. Sometimes the right answer is to build a bigger threat and let them have the mill.
• Not looking at what your opponent threatens after your move — the most common way a running mill gets built against you is that you spent three turns making your own.