Get a Clue!
Even before I began designing my Anansi's Maze puzzle, I'd been thinking about how to create puzzles that intrinsically relied on the transparency of the pieces. My inspiration was a relatively unusual and little-known sub-genre of mechanical puzzles, sometimes called overlapping puzzles, in which the pieces have openings or transparent sections and your goal is to find a way to stack up those pieces and form a picture from the intersection of their transparent bits. The first puzzle I know of in this family was released way back in 1900, but there's been a steady trickle of examples ever since.
I got one such in the Puzzle Exchange at one of my first International Puzzle Parties. It consists of six octagonal pieces of transparent acetate, each laser-printed with a gray-scale image; if you stack up the pieces just right, the gray bits combine and darken and you end up with quite a nice picture of a dog, if I recall correctly. Mostly, I remember it being really difficult.
Still, the notion had stuck with me. I wanted to play in that design space, but I also wanted to make a puzzle that wasn't so tough to solve. I had the idea that I could make it easier by sharply reducing the number of layers, maybe using only two or three. To keep it from becoming trivial, I could break each layer into multiple pieces, so that you'd have to assemble the layers themselves before you could stack them up.
So far, so good, but then I got ambitious: what if you could assemble the layers in more than one way? What if you could form either of two different pictures from the same pieces, depending on which assembly you built? This whole story got me pretty excited: this could be a really cool puzzle! Now I just needed to actually design such a thing...
And there the idea sat, more-or-less unmoving, for almost two years.
The problem was, I had no idea how to go about creating this kind of a puzzle. Unlike many of my designs, I couldn't see any way to write software to help me search for a puzzle that would match my story; one of my key design tools had been stripped away from me!
I finally picked up the idea again late last winter, when I was trying to come up with a new mechanical puzzle for use in this summer's Microsoft Intern Puzzleday. There wasn't anything magical about the process, I just dug in, started drawing potential pictures, overlapping them, and looking for interesting area intersections. It was a very incremental, iterative design journey, one of the most difficult puzzle-design efforts I've been through. Even after I'd finished the artwork, what I'd thought of as the hard part of the process, the design went through five different prototype and test-solving iterations before I finally hit on the right combination of cleverness, accessibility, and clarity of solution.
In the end, ironically, the puzzle was completed too late to be used in Puzzleday, but I did use it as my Exchange at IPP 30 in Osaka later this past summer. I had barely enough copies made then to satisfy the Exchange rules, with just a few left over at the time for selling. By the time I finished with building those, I knew that I wanted to do yet one more, fairly minor design iteration before putting the puzzle up here on the website. What with one thing or another, it's taken me a while to do that iteration, but now it's done, and I'm quite happy with how the puzzle has turned out. I hope you'll enjoy it too!