I think if a maze has an entrance and and exit, and there's some surface tension at the leading edge, when you pour water into the entrance, it'll fill the maze like a breadth first search. as soon as the water can flow freely at the exit it'll drain because there's less resistance at the exit.

pour a little water on a counter and you'll get a round area with little walls at the edge, if it's not clean it'll kinda break down where it's dirty (surface tension doesn't hold up) once it hits the edge of the counter, the surface tension pushes all the water over the edge. if it's a perfectly flat clean surface it'll make a perfect circle till it hits the edge.

so the water won't fill every nook and cranny of the maze, it'll start a new circle at every decision point, till one of those circles goes over the edge.

It would to an extent. The idea is that the maze has an exit, and once the water made it to the exit, it would spill out. The steady spilling out of water would create a current of flowing water all the way back to the water's source, following the "solved" path of the maze. The water didn't intelligently solve the maze, though, but rather the solution emerged out of the simple but massively parallel interactions between collisions of atoms (i.e "weak forces") and gravity.

I think OP was talking about a maze with an exit where the water can drain out, not a maze sealed all around the edges.

It would be a fun experiment to test this with and with an open exit drain at the end of the maze.

Ah OK makes sense now, I was assuming like a kids toy maze where it's covered on all sides. An entrance and exit maze makes a lot more sense.

Surface tension will keep it from spreading beyond a certain point unless the exit is the longest path of the entire maze.