I've got a proper race condition in my current solution to the kill-switch puzzle. The output from my timeout circuit that turns off the power if no commands arrive in time is getting mixed in to the command stream in the wrong order, causing it to shut off power unnecessarily. It's a wonderful puzzle.