What Changes When Prototypes Cost Almost Nothing

In 2023, building a credible software prototype meant weeks of work by people who were hard to hire. By 2025, a working prototype of most software ideas could be produced in days. Inside this studio, several of the systems in the Registry went from written hypothesis to functioning product in under two weeks.

The cost curve is not flattening. Each traversal of our operating cycle has been faster than the one before it, and the constraint that slowed each traversal was never construction. It was deciding what to construct, and interpreting what happened after launch.

When prototypes cost almost nothing, the prototype stops being evidence of commitment and becomes a unit of inquiry. You no longer build to demonstrate that you can. You build to find out whether you should.

This inverts a habit most builders carry. The traditional discipline was to protect scarce engineering time by thinking carefully before building. The new discipline is almost the opposite: think just enough to form a testable hypothesis, then let a cheap build answer questions that analysis cannot.

Three things follow. First, the number of traversals matters more than the perfection of any single one. A studio that runs eight honest experiments learns more than one that polishes two.

Second, killing weak concepts early becomes a skill rather than a failure. If building is cheap, the expensive mistake is not the prototype that dies — it is the prototype that lives too long because someone was reluctant to read the evidence.

Third, merely having an idea is worth less every quarter. Ideas were never scarce, but the gap between idea and artifact used to protect them. That gap is closing. What remains scarce is the judgment to select which ideas deserve a traversal, and the discipline to interpret what reality says back.

The Registry is our structural answer. Every system we run is positioned on the same six-stage cycle, and archived systems stay visible with the distance they traveled. The point is to make the cost of an experiment legible — and to make abandoning one feel like recorded progress rather than erased work.

Whether validation keeps pace. Cheap building tempts teams to skip the validation stage entirely, reasoning that the build is the test. Sometimes it is. But a build tests the solution, not the problem — and problems are still expensive to misunderstand. The open question is where the line sits between questions a prototype can answer and questions it quietly begs.