Writing
Substack essays and research notes
I write about the gap between a trustworthy-looking output and the evidence needed to trust it. Published essays live here as canonical copies; the external Substack destination will be added only after its URL is verified.
Why the label “verified” does economic work even when the underlying test is undefined.
Why Provenance Beats Freshness
Forthcoming
A current number without a source receipt can be less useful than an older number whose lineage is inspectable.
Pre-Registration Outside Academia
Forthcoming
What product and data teams gain by fixing hypotheses and decision rules before results are visible.