Maze Escape — a free browser maze game against the clock

Every run builds a brand-new 15×15 maze, drops you in the top-left corner and asks one question: how fast can you reach the green square in the bottom-right? The clock does not start until your first move, so you get as long as you like to study the walls before you commit — and the moment you swipe, it is running until you are out.

The thing that makes it play well on a phone is that one swipe is not one step. Swiping runs you down the entire corridor in that direction and stops you only where a decision actually exists: a fork, a dead end, or the exit. The route out averages about ninety squares, and a player who never takes a wrong turn covers it in fewer than a dozen swipes — the steps you skip were never choices in the first place. Behind you the game leaves a faint breadcrumb trail on every square you have stood on, so a corridor you have already burned looks different from one you have not. Free, instant, no download, and your best five times stay on your own device.

How to play

• Swipe up, down, left or right on the maze — or press the arrow keys / WASD on a desktop.
• You slide the whole corridor, following its bends, and stop at the first fork, dead end or the exit. A wall right in front of you is a no-op, not a wasted move.
• Keep your finger down and keep swiping to chain moves — each swipe past the threshold fires immediately, so you can draw a route in one gesture.
• The faint trail marks every square you have visited. It is the difference between a maze you can read and one you keep re-walking.
• Your time in seconds is the score, and lower wins. Moves are counted beside it but never ranked.

Why keeping one hand on the wall always gets you out — and why it is slow

The maze here is what maze people call perfect: exactly one route between any two squares, and not a single loop anywhere. It is built by randomised depth-first search — carve into a random unvisited neighbour, back up when you are boxed in — which grows a spanning tree over the 225 squares. A tree of 225 nodes has exactly 224 edges, so the finished maze has exactly 224 open passages and no more. That single property is what the classic wall-follower rule rests on. Put your left hand on the wall beside you, never take it off, and walk: because there are no loops, the walls form one continuous connected surface, and tracing it drags you past every square in the maze. You physically cannot circle forever, because there is no circle to be trapped in. The exit is guaranteed. This is also exactly why the rule fails in mazes with loops — an island of wall that touches nothing else will happily walk you round it all day.

The guarantee comes at a price, and the price is the whole reason this game has a clock. A wall-follower walks the outline of the tree, which means it enters and leaves every dead-end branch it meets — and these mazes carry around two dozen dead ends apiece. In the worst case it covers each of the 224 passages twice, roughly 448 steps, to finish a route that averages about 92 squares. It is the algorithm you want when you are inside the maze with no view of it — a robot, or a person in a hedge maze. You are not in that position. You can see the whole board, and spending your seconds walking dead ends you could have spotted is the most common way to post a bad time.

What you should be doing instead is the trick called dead-end filling, and it is worth learning because it is almost effortless by eye. Find any square with exactly one opening — a dead end — and mentally plug it. That turns its neighbour into a dead end, so plug that too, and keep going. Repeat until nothing is left to plug, protect only the entrance and the exit, and in a perfect maze what survives is precisely the solution path, nothing more. On screen this is a scan, not a walk: your eye picks off the stubs in a second or two and the long spine that is left is your route. That is also the honest reason the breadcrumb trail is here. It is a hand-rolled version of the same method — every square you shade is a square the answer does not go through — and the players with the fastest times are the ones filling dead ends with their eyes before they ever fill one with their feet.

More games

More puzzles that reward planning over reflexes: 15 Puzzle and Sokoban. Or browse all games.