The Engine Turn — A Second Game From the Playbook

A methodology is only proven when it works across multiple projects. Can the same process build a different game, fast, without me carrying it by hand? Rift Fighters was the first cycle. This is the second.

I took a second, unrelated card game, The Engine Turn, from nothing to a playable, legible build in a single day, by loading my design-process playbook into an AI thread as standing context and letting it run the same phases: requirements contract, core-loop model, playable prototype, legibility pass. The playbook encodes the operating rules, the dependency order, and the decision trees that the first game taught me. Here I was testing the playbook itself.

It held. The build reached "designer playtesting a legible build" the same day, and the expensive early rounds were, exactly as the playbook predicts, almost entirely comprehension work rather than balance. The automated natural-play drivers the process mandates caught real design flaws (a dead initiation branch, a stall, a deadlock from a new gate) before any human playtested them.

A second playable game in a day, and proof that the method is not a one-off. The value is not this particular game. The playbook transferred to a new problem, and that transfer is the whole thesis of how I work.