# The Open Source Board Engine Ecosystem, Layer by Layer


The Open Source Board Engine Ecosystem, Layer by Layer

A board engine is three layers: rules, referee and search. How the open source ecosystem fits together and where Gomoku and chess engines live.

The short version first, then the parts that usually get skipped.

Splitting a board engine into rules, referee and search makes every repository legible at a glance. Most projects are one layer plus the minimum of the other two, and reading them as three parts tells you which one you are actually looking at.

What actually works, in order: Start from the rules layer because it constrains everything above it, then check whether the search is plain minimax with pruning or something stronger, and read the referee last since its job is mostly to stay out of the way.

The numbers worth knowing: Most repositories are a rules layer plus a search layer, with the referee reduced to a few legality checks. Web projects add two constraints: render on a canvas, and run inside a browser budget.

Where people go wrong: Judging an engine by how strong it plays. Strength comes from search depth and evaluation tuning, both of which are replaceable. The rules layer is what decides whether the project survives its second year.

A note on defaults. Every tool in this space ships with settings chosen for the average case, and if your situation is not average, the default path is what sends you back to redo the work a month later. Decide what you are optimising for in one sentence, then let that sentence filter the steps. Anything that does not connect to it can wait.

I keep the full walkthrough with the order of operations on my own site, here: https://yiboardgame.com/blog/open-source-board-engine-ecosystem