Show HN: DfsWord – a puzzle that turns words into graphs

I accidentally invented dfsWord while developing a force-directed graph layout algorithm to turn relationships among data into an easy-to-understand picture.

As a quick way to generate test data, I started using words: each letter became a node, and consecutive letter pairs became edges. This worked particularly well because I could always easily verify whether the system had accurately represented the relationships between the letters by tracing the correct spelling of the word through the diagram.

The words have many different shapes. For example, "Antarctica" looks like a kind of tiny Sierpinski triangle. "Utilitarian" like a rocketship from the 1950s. "Llama" is a straight line, but can also spell out the word "mammal".

I often wound up showing these "word graphs" to people to explain what I was trying to do. Inevitably, I'd be in the middle a sentence and the person would interrupt: "Hey, can we put in my name and see what comes out?"

Finally, after a chemist friend enthusiastically noted "Wow, 'Austria' looks sort of like benzene," I decided that maybe there was more to the word diagrams than just their value as test data.

Since people -- including non-chemists -- enjoyed trying to work out the word from the network of letters in the diagram, I thought it would make a good game. I added a clue, and voila, dfsWord was born.

Right now the site allows you to view and solve puzzles, submit new ones, and just try things out in the "Playground." I'll be running new daily puzzles for at least a month or so, and hope to have a basic newsletter set up soon that will send out that days puzzle. I'd love to hear HN's thoughts, suggestions, and any puzzles you care to share.