Rapid book catalog

How the scores work

A fast, inspectable first pass—not an expert verdict on an entire book.

Three claims, equal weight

Each book gets exactly three central factual or causal claims. A rough AI assessment assigns 100% for supported, 75% for mostly supported, 50% for mixed, or 25% for weak. The displayed score is their average, rounded to the nearest whole number.

The selection of those three claims is itself a judgment. A different defensible selection could change the score, and this rapid layer does not confirm the exact wording against a specific edition.

Method three-central-claims/0.1.0 · 350 books · 1050 assessed claims · checked 2026-09-01

Source-health snapshot

The automated audit followed every unique citation. It tests whether a URL responds; it cannot tell whether the source is strong enough for the claim.

1431unique URLs
2100citations
235moved URLs
1066Reached
323Publisher blocked the checker
0Confirmed broken
7Timed out or server error
31Network error
4Other response

Publisher anti-bot responses are deliberately separate from confirmed broken links. Timeouts and network failures remain unresolved rather than being called good or bad. Snapshot 2026-09-01 using http-head-get/0.1.0.

Calibration sample

Fifteen books were sampled across five domains to look for obviously implausible score behavior. This is a rough audit, not an independent fact-check.

Broad macrohistory

67 avg

Guns, Germs, and Steel 67 · Sapiens 67 · The Dawn of Everything 67

All three land at 67 because the selected claims mix durable broad patterns with disputed universal or deterministic framing. Passage-level selection review remains the main calibration risk.

Science and medicine

84 avg

The Gene 92 · The Emperor of All Maladies 92 · The Structure of Scientific Revolutions 67

Evidence-rich scientific histories score highly; Kuhn lands lower because two selected universal claims remain contested. The spread is directionally plausible, not proof of accuracy.

Disasters and outbreaks

86 avg

The Great Influenza 83 · The Hot Zone 75 · Midnight in Chernobyl 100

Documentary histories score strongly while The Hot Zone loses credit for sensationalized transmission and symptom framing.

Technology and business

92 avg

The Innovators 83 · The Box 92 · Bad Blood 100

Specific institutional and fraud records score higher than sweeping innovation narratives. This domain may be easier to source than ancient macrohistory.

Social and institutional history

75 avg

The Warmth of Other Suns 100 · The Better Angels of Our Nature 58 · Why Nations Fail 67

Narrow documentary claims score higher than global decline or single-cause institutional theories. Selection and causal-scope judgments deserve expert review.

Known limits

  • These are rough AI judgments, not human-reviewed or passage-confirmed verdicts.
  • Three selected claims cannot represent every assertion, omission, or framing choice in a book.
  • 669 citations reuse a URL already used elsewhere; reuse can be appropriate, but it reduces evidence independence when the same source carries several claims.
  • The score says how the selected claims fared under this rubric—not whether the book is worth reading.
Inspect the data and audit code