Methodology

How a claim earns a spot on the sheet.

Nothing goes live without a receipt. Here's how the sheet is built.

Source tiers

  • Primary research — original datasets, controlled experiments, or crawler logs from the original researcher.
  • Vendor study — research from a company with commercial interest but disclosed methodology.
  • Independent analysis — third-party writeups replicating or aggregating primary evidence.
  • Expert commentary — statements from named practitioners with verifiable expertise.
  • Anecdotal — single-case observations. Included for context only; never a lone source for a fact.

Technical impact score (0–10)

A receipt-weighted estimate of how much the factor moves citation probability in production. 9–10 = category-defining. 5–7 = meaningful lever. 0–4 = marginal or myth.

Confidence (0–100%)

Confidence in the finding based on sample size, methodology transparency, and independent replication. It is separate from technical impact — a low-confidence finding can still describe a high-impact factor.

Publication flow

  1. Research lands via /recommend, forwarded newsletters, or direct ingestion.
  2. Curator (David Berkowitz) verifies the source and normalizes the claim.
  3. Claim is entered as a Master Receipt and mapped to one or more matrix rows.
  4. Rows go live only after receipt + scoring + engine attribution are complete.

Corrections

Every row is versioned. Email corrections@geocheatsheet.com with the row ID and a counter-receipt.