The Patent That Ranked Every Web Page: PageRank, Filed 1997
Note on this format: This memo records what I found at the patent URL. Full text, Claim 1, and forward citations have not yet been confirmed. Verified facts only; speculation is labeled as such.
Why Dig This
Google's search ranking is built on a simple, strange idea: the importance of a page is determined not by how many times it says the word you searched, but by how many other pages point to it — weighted by how important those pages are. That recursive logic is in a 1997 patent filed by Lawrence Page. As LLM-based search and graph-structured RAG systems grow, understanding how "importance in a linked structure" was first formalized provides a useful baseline for comparison.
Patent Basics
- Patent number: US6285999B1
- Title: Method for node ranking in a linked database
- Filed: 1997 (exact date: not confirmed from full text)
- Inventor: Lawrence Page
- Assignee: The Board of Trustees of the Leland Stanford Junior University (later transferred to Google)
- Primary source: Google Patents (URL confirmed; full text unread)
- Legal status: Expired; transfer details not confirmed
Core Content (Abstract and Wikipedia-Level Information)
The patent describes a method for ranking nodes in a linked database such as the web. Each node receives a rank score based on the number of links pointing to it and the rank scores of the nodes those links come from. The model is formalized as a "random surfer": a user who clicks links at random — where does that user end up most often? Pages the random surfer visits most often rank highest.
Key structural property: the importance score is recursive. A page's score depends on the scores of the pages linking to it, which depend on the scores of pages linking to them. This is solved iteratively.
Claim 1 wording and formula details not confirmed — abstract and Wikipedia-level information only.
Connections to Today (Hypothesis)
| US6285999B1 (1997) | Modern technology | Assessment |
|---|---|---|
| Node importance from incoming links, weighted by linker importance | Node weighting in graph-based RAG | Analogy (structure is similar; purpose differs significantly) |
| Recursive importance computation | PageRank-referencing LLM evaluation research | Similar (same problem orientation) |
| Random surfer probability model | Probabilistic graph traversal broadly | Similar |
These are pre-full-text hypotheses. Assessment will be revised after Claim 1 review.
What's Unconfirmed
- Claim 1 verbatim text
- Exact filing date
- Transfer timeline from Stanford to Google
- Forward citation count (scale of influence on modern research)
- Published academic papers connecting PageRank to LLM search design
Next Action
Confirm Claim 1 and forward citations via Google Patents. If forward citations include LLM or RAG research, the connection to modern search design has firmer grounding.
Sources:
- Primary patent: US6285999B1 on Google Patents
- AI & ML Patent #1 (full note): Amazon item-to-item collaborative filtering US6266649B1 (1998)