Automating a game of automation - The Signal State

Описание: Automating a game of automation.

The Signal State is a puzzle game where you design signal-processing circuits using predefined blocks like splits, biases, attenuators, and sums.

After playing the game, I had a slightly cursed idea:
what if we could automate the design itself?

In this video, I give a high-level overview of a solver I’m building for The Signal State. Instead of manually placing blocks, the solver takes semantic intent — what the outputs should be in relation to the inputs — and automatically synthesizes a valid design using the game’s constraints.

We walk through the full pipeline:

Describing outputs as mathematical relationships
Lowering those expressions into actual game blocks
Deduplicating structure using symbol interning
Building a DAG and netlist
Accounting for fanout and splits
Evaluating cost (modules, cables, time)
Using Pareto optimization to return optimal tradeoffs instead of a single “best” solution

This video focuses on the big picture.
In the next videos, we’ll start implementing each part step by step — in code.

If you enjoy thinking about systems from first principles, compilers, solvers, or just automating things that probably shouldn’t be automated, you’ll enjoy what’s coming next.