Eight skills

The skills

A skill is a decision procedure your agent runs with you: the questions in order, the checks that have to pass, and the exit that says this one cannot be answered. Eight of them across 6 positions on the method map; causation carries three.

Routing

routing-questions

Whether this becomes work at all, and which skill it becomes.

 The floor

defining-metrics

A metric turned into a computation, a source of truth, a registry entry, and a statement of how it will be gamed.

 Description

forming-hypotheses

A what-happened question turned into a localised, sized hypothesis, with the floor checked first and the gap never quoted as the prize.

 Causation

designing-experiments

The four choices that cannot be repaired after launch, with the MDE from the prior store.

 Causation

reading-experiments

Whether the result is a result: SRM, exposure, the sequential bound, CUPED, shrinkage.

 Causation

choosing-causal-designs

The method that matches how assignment happened, and the exit that says there is no comparison group.

 Prediction

building-models

Whether a forecast or a repeated decision belongs to a model, validated out of time, and the holdout that measures its impact.

 The ceiling

writing-reports

The decision rule first, the result last; the belief filed where the next question starts.
Knowledge layer the ceiling: what each decision taught, and the prior it left writing-reports Measurement framework the floor: the definition of success, and whether the number can be trusted defining-metrics Routing routing-questions one decision, one lookup, three questions, one gate Exploratory analytics forming-hypotheses the question, localised and sized can you randomise? Experimentation designing-experiments reading-experiments Causal inference choosing-causal-designs Statistical modeling building-models the decision, made at volume yes no description hypothesis
Where each skill sits. Each name is a link.

One question, end to end

Each skill covers one position on the method map and one failure mode. Together they run one question end to end: it arrives at routing-questions, stands on the floor defining-metrics maintains, gets its method from designing-experiments or choosing-causal-designs, is believed or not by reading-experiments, and is filed by writing-reports so the next question starts smaller.

A what-happened question leaves the path for forming-hypotheses and comes back as a sized hypothesis; a forecast, or a decision made continuously at volume, leaves it for building-models, and the decision comes back for the experiment that measures the model's impact.