A lot of the discussion about using the Turing Tumble parts for large-scale computation focuses on infinite boards and repeating cells. I think it would be interesting to explore more traditional computation, similar to what is shown in the puzzle book. Small boards can perform simple computations, which can be connected to form larger behaviors.
Many of the puzzles use input and output in the form of bits or gearbits. If there were multiple boards, they might communicate by having certain bits that were paired to match each other’s state. Other puzzles require a specific sequence of marbles to be dispensed, or a particular color of marble placed in an interceptor. These signals could be received by connecting the outputs of some boards to the input trays of others, in addition to a “trigger” input that could start the board. Obviously, a ‘discriminator’ part that could tell the difference between red and blue marbles would come in handy as well.
I’m currently designing a simulator that has the ability to wire boards and bits together, plus some other goodies like duplicators, gearramps, and synchronizers. The working title is “Marble Magnus” (although maybe that’s too cheesy). You can edit things on the micro level, in the style of other TT simulators, or on the macro level, using a node editor to connect boards and other logic together. I’ll post some screenshots when I make a working prototype.