Patents, papers, tools, and working notes
A curated knowledge base for understanding how search systems retrieve, rank, cite, and synthesize information.
Patent Reviews
- Google: Ranking documents based on user behavior signals
How dwell time, click patterns, and refinement behavior feed into modern ranking decisions.
- Microsoft: Generative answer composition from retrieval candidates
A retrieval-then-generate pipeline for producing composed answers with attributable spans.
- OpenAI: Retrieval-augmented response synthesis
A grounding pipeline that pairs retrieval with generation to reduce hallucination and improve citation quality.
Research Papers
- Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
The foundational RAG paper — retrieval as a first-class input to generative models.
- REALM: Retrieval-Augmented Language Model Pre-Training
Bakes retrieval directly into pre-training, so the model learns to lean on external documents.
- Citation behavior in LLM answer engines
An empirical study of which sources answer engines cite, and why.
Tools
- AI Overview tracker — monitor inclusions across queries
A tool for tracking when your domain is included in AI Overview responses across a keyword set.
- Citation log — track which sources answer engines reference
A running log of external sources cited by major answer engines, filtered by topic.
- Entity coverage audit — surface gaps in topical authority
Maps your content against the entity graph a category is expected to cover.
Concept Notes
- What zero-click really means for high-intent queries
Zero-click is not uniformly bad — it depends heavily on where the query sits in the funnel.
- Why citations may matter more than ranks
A case for treating share of citation as the primary visibility KPI in AI-mediated search.
- Designing pages that answer engines want to extract
Concrete formatting patterns that make content easier for synthesis engines to lift accurately.