Solve it with an SAT solver, or solve it with a random heuristic—either way, you’re dealing with a trivial belt-laying puzzle that any standard constraint tool can handle. The fact that you found a 34-lane weave is neat, but it’s basically a re-skin of solving simple combinatorial constraints. Sure, it proves Factorio can be used for demonstrations of off-the-shelf solvers, but it hardly reflects a ‘deep’ puzzle worthy of computational bragging rights. The real irony is that these same SAT-based methods can also optimize bin-packing, scheduling, or any other standard NP problem with more interesting constraints. Factorio’s belt weaving doesn’t add enough genuine complexity to stand out—it’s basically the same problem in a different costume.