‹ All documentation
Evidence reference
The 2020 Google presentation said learning from logs is a main mechanism behind ranking and identified major machine-learned ranking systems as relying on logs.
SupportedSUPPORTEDclm_019ff289-27f4-7bc1-9cf1-7f27a4c324cd
This is a stable reference page for one canonical claim in the Editory evidence base. It exists for traceability and citations; the curated documentation is the readable entry point.
Sources
- Trial exhibit UPX0219: Google presentation 'Logging & Ranking' (US v. Google, 1:20-cv-03010)
U.S. Department of Justice (public trial exhibit; document authored by Google) · primary source · retrieved 2026-08-20
“Learning from logs is the main mechanism behind ranking.” — Slide 'Logs -> Ranking -> $$$' (Bates GOOG-DOJ-16512426 vicinity)
“all major machine learning systems for ranking rely on logs: RankBrain, RankEmbed” — Slide 'Logs -> Ranking -> $$$', second bullet
- United States v. Google LLC Trial Transcript, September 20, 2023 PM
U.S. District Court for the District of Columbia · primary source · retrieved 2026-08-20
“Learning from logs is the main mechanism behind ranking. ... That's what it says, yeah.” — transcript page 1788, lines 12-16
- 7 must-see Google Search ranking documents in antitrust trial exhibits
Search Engine Land · retrieved 2026-08-20
“Google presentation: Logging & Ranking (May 8, 2020)” — Section 5 'Logging & Ranking', link caption
Related claims
- BROADENSA 2017 Google presentation made public as a DOJ exhibit presented user interactions as one of three pillars of ranking, alongside body and anchor information.Under evaluation
- BROADENSRand Fishkin reported that purported leaked Google API documentation contained click-specific attributes such as goodClicks, badClicks, and lastLongestClicks associated with Navboost/Glue-related modules.Under evaluation
- REPEATSGoogle’s current How Search Works documentation says aggregated and anonymized interaction data is used to assess whether search results are relevant to queries.Supported
- NARROWSMore informed search-result interactions can improve training data for models used in ranking and language understanding.Under evaluation
Release: seo-geo-product-v0.3 · Phase Two freeze: P2-CORE300-FINAL-v1.0 · Generated: 2026-08-23T08:01:40Z