Falsification & the kill-criteria gate

Generation-first tools happily produce hypotheses. CoSci is falsification-first: it models the discipline of the scientific method as an explicit gate. This is the hinge of the whole system.

Competing mechanisms, not one guess

For a sharpened question, CoSci pushes you to lay out competing mechanisms — rival explanations that make different predictions. A single favored hypothesis is a story; competing mechanisms are a test.

The killer experiment

The killer experiment is the one whose outcome distinguishes the mechanisms — where each candidate predicts something different, so the result rules some out.

The kill-criteria gate

Here’s the rule that makes it rigorous:

An experiment cannot be marked Ready until it has hard kill-criteria and named artifact risks.

• Kill-criteria — the specific results that would falsify the hypothesis. If you can’t state what would kill it, you don’t have an experiment yet.
• Artifact risks — the ways the experiment could produce a misleading result for reasons unrelated to the hypothesis, named up front.

This is a methodology product, not a search box: the gate forces the part of experimental design that’s easy to skip and expensive to get wrong.

See it fire in the worked example, where a designed experiment passes the kill-criteria gate before a proposal is written.