If you know how the pieces move and still lose, it is almost always one of three things: you walled too early, you walled without counting, or you built something that cost you more than it cost them. This page is about those three.
New to the game? Start with the rules instead.
MazePawn has exactly one number worth tracking: the difference between the length of your shortest route and the length of theirs. Call it tempo. If your route home is eleven steps and theirs is fourteen, you are three ahead, and in a pure walking race you win by three.
Almost every decision follows from that number:
This is not a metaphor — it is what MazePawn's own engine computes. Our post-game review scores positions in tempo rather than in chess-style centipawns, so a bar reading +2.4 is literally saying "about two and a half moves ahead in the race". If you want to know whether your instinct about a position was right, the review will tell you in the same units this page is written in.
Every wall is a turn. So a wall that adds one step to your opponent's route has gained you precisely nothing: they lost a step, you lost a turn, and the count is exactly where it started. A wall needs to buy you two or more steps of tempo before it is worth the turn it cost.
Before you place one, ask two questions in this order:
The second question is the one people skip, and it is the one that loses games. Walls belong to nobody once placed. A wall you built to divert your opponent sits on the board for the rest of the game, and if your own route home crosses it later, you have paid for their detour twice.
In the first few turns your opponent has not committed to a route. The board is open, every column is available, and a wall placed now is simply walked around for free — you spent a turn and a permanent resource to achieve nothing.
Walk first. Walls get stronger as the board fills and as your opponent's options narrow, which means the same wall is worth more later than it is now. The one exception is when your opponent has already spent turns committing to a specific lane; then the wall that closes it is worth playing immediately, before they get further down it.
A single wall is usually worth one or two steps, because there is a way around either end of it. The reason walls feel weak to new players is that they are being placed alone.
Real cost comes from placing a wall so that the detour around it runs into something — the board edge, an existing wall, or the wall you are about to place next. Two walls forming a corner or a short corridor can add five or six steps where each of them alone would have added one. Look for places where your opponent's detour has only one exit, and then consider closing it.
Be careful doing this near your own route. Corridors are the easiest way to wall yourself in, and MazePawn will not let you finish one that seals either player completely — so you can find yourself with a half-built structure you cannot complete and cannot remove.
Ten walls is fewer than it feels. A player who has spent all ten has no answer left: from that point on they can only run, and their opponent can build against them freely, knowing nothing can be built back.
The last walls are also the strongest ones, because by the time they are played the opponent is close to home and committed, and there is no cheap detour left. Trading your final walls for one or two steps in the middle game is usually a bad exchange for that reason.
Standing in front of your opponent does not stop them. They jump over you and gain a square, so blocking with the pawn is worse than doing nothing.
Going the other way, the jump is a free step when you can arrange it. If your route and theirs are going to cross anyway, meeting them head on is worth a tempo, and it is one of the few ways to gain a step without spending a wall. Remember the diagonal case: if a wall or the edge sits directly behind them, you jump to either side instead — which occasionally lands you somewhere better than the straight jump would have.
Play the computer, then read the review. MazePawn grades every move you played against what the engine would have chosen and shows the eval in tempo, so you can see the exact turn where the race flipped — usually a wall that felt clever and cost you two steps.
The puzzles are the other half: they put you in a position with one right answer, which is the fastest way to train the "count the race" habit that everything above depends on.