All capabilities

Capability standard · Data

Experiment Design

v1.0

Testing a change so that the result actually means something.

Designing a test so that its result means something, including what it cannot tell you.

Reference CAP-DAT-003

What each level asks of you

A verifier marks against these descriptors. Level 1 is free and self-directed; Levels 2 and 3 use credits and a human verifier.

L1

You can state a hypothesis and the measure that would settle it.

Free

L2

You design and run a test with a control, a sample able to detect the effect, and a read-out that states the uncertainty.

1 credit

L3

You set the experimentation standard for a team, including the cases where a result should not be trusted.

2 credits

What counts as evidence at L2

A test you designed, its result, and the limits of what it showed.

Evidence has to be work you did. It is attached to the request and read by the verifier — it is never published.

Role models that ask for this capability

Standard due for review 2027-06-30. Levels are issued against the version current on the day they are earned.