Today I tried to get SK combinators to evaluate using mostly only string replacements. I think I did ok. I think this might be a step toward building a functional or symbolic programming paradigm in a relatively simple Turing machine or other simple system. I think it's an alright visual way to see them evaluate. I might try and implement this in this Turing complete "mechanical" puzzle game called Opus Magnum some time. It'd be interesting to be able to watch it compute.
The notebook here