PM: Are you ever concerned that a maze is too hard or not hard enough?

AG: Not really. When I first start toying with a potential new type of maze, my main aim is to discover the depth and breadth of its 'maze-space' - and find out whether the space exhibits the right sort of connectivity to create good mazes. Maze-space is my term for the total number of distinct states (or potential 'nodes') in the maze. Most of the time this value be assessed relatively simply mathematically, but just how much of the space can be reached and successfully exploited is less obvious. The fun bit for me is weaving together the maze-space and discovering just how devious the resulting puzzles can be.

If I find the maze-space and its connectivity is good, I shrink the over-all dimensions of the maze, creating something more compact but still challenging to solve. My aim is not to create a puzzle that takes particularly 5 minutes or 5 hours to solve, but one that packs the biggest punch for its apparent size.

PM: What do you consider to be your best creation so far?

AG: I think this now has to be Plank-puzzles. Initially I had no idea how successful these puzzles were going to be, and I struggled to design even mediocre puzzles on large grids. The maze-space was interesting I knew - but I found it hard to weave into good puzzles. It was a slow (but intensely satisfying) process discovering how much could be squeezed onto smaller and smaller grids.

Plank-puzzles has been taken to market as a mechanical puzzle called River Crossing by US company ThinkFun. Some people will now know my name through this one puzzle, rather than through my website.

1 | 2 | 3 | 4 | 5