What would count as progress
A measurement approach validated on at least one historical source-level change, recovering an effect that independent historians agree took place.
Why it matters
If funders require measurements that source work cannot provide, useful work may be missed. That is one possible mechanism behind the Symptom Trap hypothesis. Hard-to-measure is not the same as small.
Where to start
Work backwards from changes everyone agrees happened, such as the spread of oral rehydration therapy or the end of smallpox (see Historical Case Studies), and test which measures would have detected them early. Contribution analysis, process tracing and forecasting tournaments are candidate methods.