Clobber Strategy: Mobility, Tempo and Beating the AI

Updated October 2026

Clobber has the shortest rulebook on the site and one of the least intuitive strategies. Every move is a capture, nobody can pass, and the player who is left without a capture on their turn loses. That turns ordinary instincts upside down: taking material is not the goal, and a capture can be the move that loses the game. This guide covers how the board behaves, how to think about tempo and isolated stones, and how the AI chooses its moves so you can extend a winning streak.

The setup and the one rule

The board here is 5 columns wide and 6 rows tall, and it starts completely full: 30 stones in a checkerboard, 15 teal for you and 15 orange for the AI. You always move first. A move means sliding one of your stones up, down, left or right onto a touching enemy stone, which is removed. Your stone now sits on that square and the square you left is empty for the rest of the game. Diagonal captures and moves onto empty squares do not exist.

To play, tap one of your stones. Every enemy it can take gets a green ring; tap one of those to capture. Tapping a different one of your own stones simply changes the selection. In the opening position you have 49 possible captures to choose from, because every stone is surrounded by enemies.

The game ends the moment the player to move has no capture. That player loses. There are no draws, and our simulated games lasted about 19 moves in total on average, so each decision carries real weight.

Why captures cost you

Every capture does two things. It removes an enemy stone, which takes away the moves that stone had. But it also moves your stone forward and leaves a hole behind it, which can cut your other stones off from their targets. A stone with no enemy beside it cannot move and cannot be taken. It is frozen for the rest of the game, whichever colour it is.

So the real currency is mobility: how many captures each side will still be able to make. The material count shown above the board, teal against orange, says almost nothing about who is winning. A player with fewer stones but more live contacts is usually ahead.

• Corners are weak. A corner stone touches only two squares, an edge stone three, a central stone four. Stones near the middle keep their options longest.
• Holes spread. Each capture creates an empty square, and stones bordering holes lose contacts. Think about which of your stones a hole will isolate before you make it.
• Stranding cuts both ways. When your stone captures, it leaves its old square, so any orange stone that touched only that stone is now frozen. But any teal stone whose only contact was the orange stone you just removed is frozen too. Check both before you capture.

Reading the board as separate fights

After a few moves the position breaks into groups of stones that no longer touch each other. Each group is its own small game, and on your turn you choose which group to play in. The overall winner is whoever still has a move somewhere when the other side runs out everywhere.

The simplest group shows the whole idea. One teal stone next to one orange stone, with nothing else around, is worth exactly one move to whoever takes it first: after the capture a single stone remains with no neighbour. If several of these pairs are scattered around the board, they get used up one per turn, and whoever moves first gets the odd one. Larger groups are harder to count, but the question is always the same: how many moves can I make here, how many can my opponent make, and who runs out first?

This is why the best move is often a quiet-looking one in a group where you have spare moves, rather than an obvious capture that hands your opponent a reply.

Tempo: making the last move

Every contact works in both directions: if your stone can take an orange stone, that orange stone can also take yours. So you never own a move outright. What differs is what is left after each capture. In a straight line of teal, orange, teal, whichever side captures first leaves no contacts behind, so the group is worth exactly one move to whoever gets there first, just like a lone pair.

Late in the game, sort the groups this way. Count the groups that will be used up in a single move; if those are all that remain, the side to move wins when their number is odd. Then look for groups where your capture leaves your stone still touching an enemy while orange's capture would leave nothing. Those are where you gain a tempo, and they are worth saving until the simple groups are gone.

The reverse matters too. If your capture lands your stone next to a fresh orange stone, you have spent a move and given the AI a new reply. When the board is nearly empty, check where your stone ends up, not just what it removes.

How the AI thinks

The AI looks ahead a fixed number of moves and scores each position by its own number of available captures minus yours. Its search depth depends on your current streak:

• Streak 0: 2 moves ahead.
• Streak 1: 3 moves ahead.
• Streaks 2 and 3: 4 moves ahead.
• Streak 4 and above: 5 moves ahead, which is the maximum. If a position is too complex to finish that search within about a second, which can happen on slower phones early in the game, it plays the move its 4-move search chose instead.

When several moves score the same, it picks one of them at random, so the same opening will not always get the same reply. It waits about 0.3 seconds before thinking, then plays.

Two things follow from this. First, because it only looks a few moves ahead, the AI is weakest in the opening and early middle game, when the end is far beyond its horizon. That is where you should play for long-term structure: keep your stones central and connected, and isolate orange stones. Second, when only a handful of moves remain, its search reaches the end of the game and it plays perfectly. You need to be ahead in the count before that point.

In our simulation, a player choosing moves at random won about 27 percent of games against the 2-move AI and about 13 percent against the 5-move AI. A program searching three moves ahead won about 27 percent against the strongest setting, so long streaks require genuinely better counting, not just care.

Managing a streak

Your score is your current win streak. After a win, the card offers Next game, which keeps the streak going, plus a Save button. A loss shows the final streak and a button to start a new run. Pressing New at the top during a run starts again from a streak of zero, so do not tap it by accident. Your best streak is saved on your device.

You can submit your current streak to the Hall of Fame from any win card, not only after a loss. If you have reached a streak you are proud of, saving before the next game is a sensible habit, because a single loss ends the run.

Habits that win more games

1. Before each move, ask what happens to your own mobility, not only to orange's.
2. Prefer captures that strand more orange stones than teal ones.
3. Do not empty the middle early; central stones keep the most options.
4. Once the board splits into groups, count moves in each one and save the groups where you gain a tempo for last.
5. At high streaks, slow down. The AI is at full strength from a streak of four, and every game counts.