Data Sources
Why we settled on OpenAlex for metadata over the alternatives.
The trade-offs we accepted, and the ones we didn't.
Mar 8, 2026 · 6 min
The trade-offs we accepted, and the ones we didn't.
Keyset pagination, query budgets, and what we'd do differently.
How fields phrase novelty, and why it matters for matching.
Why we kept a number at all, and when it's allowed to lead.
What broke at scale, and the queue design that fixed it.
The linguistic markers that predict early citations.