Search the work,
find the economist.

A higher match means stronger documented fit to your search within the evidence we have indexed. We verify sources, attribution, and inclusion criteria, but do not assess each paper’s rigor, originality, importance, or overall quality. A match is a discovery lead, not a quality endorsement.

University prestige, journal rankings, and citation counts receive no ranking weight. Institution and journal names remain searchable text, so searching those names can affect results. The selected context can also reward the presence of matching published work, verified current work or professional service, or the recency of matching work. These adjustments do not grow with the number of works.

To support wider discovery, the default general-discovery order favors institutions and career stages not yet represented when relevance scores are within 6% of the highest remaining score. A person’s score uses their strongest matching document, with no bonus for additional matching documents. Coverage of distinct parts of your query can still improve the fit. Verified official-profile research can establish inclusion without publication-level evidence.

These safeguards do not make the results independent of prestige. Publishing and indexing practices, unequal online documentation, and our incomplete coverage can still affect whose work is found. More indexed works also create more opportunities to match. We have not yet established how much institutional concentration remains after accounting for topic fit. Follow the linked evidence and consider relevant people beyond the first results.

The current method (v0.23) follows these steps:

1. Normalize the query. Search lowercases the text, removes accents and punctuation, ignores generic discovery words, and reduces a small set of common English endings so close word forms can match.
2. Interpret concepts and context. Exact phrases from a reviewed concept vocabulary activate their approved aliases. Phrases such as “seminar,” “working paper,” “editor,” “recent,” or “current work” can select a limited ranking context when the context control is set to automatic.
3. Apply filters first. Research topic, career stage, institution type, region, doctorate year, method, evidence depth, and work-status filters determine which economists are eligible before their evidence is scored.
4. Score two evidence channels. The literal channel uses BM25 keyword relevance and an exact-phrase bonus; a multiword query normally needs at least two distinct term matches unless the full phrase occurs. The concept channel expands activated concepts to reviewed aliases, applies BM25, and adds a concept-overlap bonus.
5. Fuse the channels. Literal and concept document rankings are combined with reciprocal-rank fusion, using 1 ÷ (40 + rank position), plus a small bonus when an exact reviewed alias appears in both the query and evidence.
6. Aggregate evidence to economists. A person’s score starts with their strongest evidence document, then receives up to a 10% adjustment for coverage of distinct query terms. Additional matching documents receive no count bonus. Exact-name and complete surname/name-token matches receive separate boosts. Up to four supporting evidence items are retained for the result explanation.
7. Apply the selected context and ordering. Seminar ordering can reward verified current-work evidence; editor/referee ordering can reward the presence of matching published work and reviewed service evidence; research-activity ordering can reward newer relevant work. The default discovery order broadens only close-score groups—within 94% of the group leader—to reduce repeated institutions and career stages. Users can instead order by matching indexed-work counts, newest relevant work, or name. Count sorts reflect incomplete indexed coverage and do not measure research quality.
8. Return and explain the matches. For a reviewed-concept query, direct concept matches are retained along with other results scoring at least half as highly as the top result. The site reports the full match count; the 12, 25, 50, and All controls change only how many cards are displayed.

The method ranks documented research fit, not overall quality. It never uses university prestige, citation counts, journal rankings, or sensitive personal characteristics.