Chapter 2 — Further Reading: The Ranking Algorithm

Grouped by the book's three evidence tiers (style bible §7). The point of this chapter is to read everything — even this list — with the tier in mind. Tier 1 is where to anchor; Tier 2 is useful but must be read critically; Tier 3 is this book's own constructed teaching material.

Tier 1 — Verified canonical (anchor here)

• Google, "How Search Works" (Search ranking systems and results). Google's own public explanation of the broad factors it uses — meaning, relevance, quality, usability, context — and its plain statement that it uses "many" signals. The single best antidote to "the 200 factors."
• Google Search Central — "A guide to Google Search ranking systems." Google's own list of named ranking systems (RankBrain, BERT, the helpful content system, reviews system, and more), which is the concrete reason to talk about systems, plural, not "the algorithm."
• Google Search Central Blog — "HTTPS as a ranking signal" (2014). The primary source in which Google both confirms HTTPS as a signal and calls it lightweight, affecting well under 1% of queries. Read it as the model of "confirmed ≠ powerful."
• Google Search Central — Page Experience / Core Web Vitals documentation. Where Google frames page experience as a tiebreaker and states that great content matters more. Ties to Chapter 16.
• Google's announcements of RankBrain (2015), BERT (2019), and MUM (2021). Google's own descriptions of each system, including the figures Google itself cited (e.g., BERT affecting roughly 1 in 10 English queries). Treat Google's characterizations as Tier 1 and any third-party extrapolation as Tier 2.
• Google's 2009 statement that it does not use the meta keywords tag. The canonical debunking of the oldest zombie "factor."
• Public record of United States v. Google (antitrust, tried 2023; ruling 2024). Sworn testimony and disclosed exhibits in which Google's use of click/interaction systems (e.g., NavBoost) surfaced — a rare on-record window into signals Google's public messaging had downplayed. See case study 1.

Tier 2 — Attributed, specifics unverified (read critically)

• Large-scale ranking-correlation studies from SEO tool vendors (e.g., the recurring studies published by Ahrefs, Semrush, Moz, and Backlinko). Genuinely useful large-sample evidence — and exactly the material §2.7 teaches you to read for confounders, reverse causation, sample bias, and the publisher's incentive. Never quote their precise percentages as laws.
• Reputable practitioner commentary on the 2024 Google Search API documentation leak (as analyzed by well-known SEO writers). Informative about what features may exist internally, but the interpretation of each feature's live-ranking role is unconfirmed — a Tier-2 reading of a real artifact, not Tier-1 fact.
• Recorded Q&A and office-hours statements from Google Search Advocates (e.g., long-running public answers that "word count is not a ranking factor," "there's no ideal keyword density," "domain age doesn't help"). Attributed and reliable in direction; treat any implied precision cautiously.
• Independent write-ups of the Exact-Match Domain (EMD) update (2012) and other cases where a believed "factor" was devalued. Useful for the pattern in case study 2; verify specifics against contemporaneous reporting.

Tier 3 — Illustrative / constructed (this book's own examples)

• Rivertown Home Services and the "proposed fixes, sorted by evidence" figure (this chapter's Strategy File). A constructed home-services company used to make the evidence sort concrete. All figures illustrative.
• "The AI Overview that took 30% of the clicks" (Figure 2.1) and "Top pages have 3x the referring domains" (Figure 2.2). Constructed teaching examples; the numbers are illustrative, not measured.

Suggested order

1. Start with Google's "How Search Works" and the ranking systems guide — see, in Google's own words, how short the confirmed list is and how it's framed as systems, not a checklist.
2. Read the HTTPS (2014) post next, specifically for the word "lightweight." Let it install the "confirmed ≠ powerful" reflex.
3. Skim one vendor correlation study with §2.7's checklist in hand — practice the reading before you need it under pressure.
4. Read the coverage of United States v. Google and the 2024 leak together (case study 1) to feel the gap between public messaging and internal reality — and why you weigh evidence tiers, not press releases.
5. Come back to Chapters 4 (meaning/entities), 6 (updates), 22 (links), and 36 (AI Overviews) when you reach them; each pays off a thread this chapter only opened.