Archived catalog overlap: inputs and limits

The archived overlap analysis can now be inspected

Re-running the saved August 2026 analysis on October 15 reproduced 566 matching normalized URL slugs out of 1,220 cleaned RatedBooks slugs, or 46.4%. Of the 846 first captured before BookLooks’ March 23, 2025 shutdown, 565 match; of the 374 first captured on or after that date, one matches.

This is a comparison of archived URL names. Normalization can combine different works, and the cleaning rules discard some short names. A first archive capture is not a publication date. These counts measure overlap in the selected archived sets, rather than copying, report authorship or a complete catalog.

The inputs and calculation are below. The citation check and five-district study are separate; their results are outside this overlap count.

Inspect or reproduce the count

The inputs were preserved in the August 1, 2026 archive work. Download the scripts and JSON inputs into one folder. Run python3 normalize_urls.py.txt to rebuild the normalized sets, then python3 reproduce_overlap.py.txt to repeat the cleaned count. The first script writes rebuilt-sets.json; compare it with the supplied _sets.json before running the cleaned count. Both use Python 3’s standard library.

• RatedBooks archived URL enumeration (opens in a new window)
• BookLooks archived URL enumeration (opens in a new window)
• Normalized joined input (opens in a new window)
• Raw-URL normalization script
• Calculation script

Rebuilding from the raw archived URLs reproduced the saved normalized input exactly. The repeated calculation reproduces the saved cleaned analysis. Its normalized input is linked separately from the raw URL enumerations so the earlier normalization can also be inspected.

The cleaning pass excludes 150 of 1,370 initial RatedBooks names: 40 short or empty names, 40 asset extensions, 32 Wix image names, 20 “copy of” pages, 15 image-height parameters and three image-transform parameters. Excluding short names can also discard real titles; even the cleaned set contains imperfect title matches. These are provisional URL-derived sets.

Keep the other studies separate

The earlier five-report citation check compared report page references with edition records. Its July 31 report versions do not establish what a committee received in fall 2025. A later version or a different edition can matter. The study’s numerical results are not published here as a verified account of the committee’s evidence.

The five-district search covered Queen Creek, Fulton County, Alpine, Phoenixville and West Ada. A district-named page on a vendor’s site does not establish district adoption. Queen Creek’s named board-member account is discussed with its policy record in the Queen Creek entry. The other searches remain bounded searches, with policy and search-access gaps; no adoption rate is calculated from them.