Old Revisions

These are the older revisons of the current document. To revert to an old revision, select it from below, click Edit this page and save it.

2026/08/21 14:50 design:ip_classification – Boxes: <wrap> renders a span, use uppercase <WRAP>; one box per list, not per bullet karel.kubicek.claude -106 B (current)
2026/08/14 03:20 design:ip_classification – programming:traffic_files now exists — drop the 'not yet written' marker. Authored by Claude. karel.kubicek.claude -17 B
2026/08/12 15:45 design:ip_classification – Methodology section: replaced the generic corpus caveats (absent venues, provisional 2025-2026) with a pointer to the new literature:corpus page, keeping the page-specific consequence. Authored by Claude. karel.kubicek.claude -58 B
2026/08/12 10:05 design:ip_classification – Review pass (Fable): fix stale figures that sat OUTSIDE the corpus section and so were missed by the windowed staleness guard. Intro 234/92/53 -> 295/114/68; operator prefix-list grep six -> seven of 5,859; abuse-feed count 13 of 234 -> 17 of 295; the NDS karel.kubicek.claude +137 B
2026/08/12 09:22 design:ip_classification – Refresh all corpus figures for the extended 2010-2026 extraction (4,322 -> 5,859 papers; IP-classifying 234 -> 295). ip_fold.mjs extended to fold 18 new residue strings back to zero; report script's hardcoded 234 denominators and open-ended 2022+ bucket f karel.kubicek.claude +1.2 KB
2026/08/06 21:55 design:ip_classification – Second review batch: downgrade the unverifiable DynamIPs quotation to a flagged paraphrase; give Khan et al.'s three agreement figures their actual per-database denominators and say it measures agreement with the provider's claim, not accuracy; note that karel.kubicek.claude +769 B
2026/08/06 21:54 design:ip_classification – Review pass fixes. Corrections: RFC 9092 was obsoleted by RFC 9632 (Aug 2024); the IPvSeeYou 26 km is MaxMind vs wardriving ground truth, not vs IPvSeeYou; the 1,238 s/target cost belongs to the street-level technique, not CBG; the Chinese residential-pro karel.kubicek.claude +3.3 KB
2026/08/06 21:49 design:ip_classification – Fix a conclusion the script output does not actually support: layer 1 has only one source, so 'never disagrees with itself' was not demonstrated. State what the run does show instead. Authored by Claude. karel.kubicek.claude +169 B
2026/08/06 21:48 design:ip_classification – Give MaxMind's actual list prices with the date observed, and note the internal-use-only licence restriction. Authored by Claude. karel.kubicek.claude +135 B
2026/08/06 21:47 design:ip_classification – DB-IP Lite row: quote DB-IP's own description of the Lite tiers as reduced-coverage subsets. Authored by Claude. karel.kubicek.claude +137 B
2026/08/06 21:46 design:ip_classification – IP2Location LITE row: it tied with NetAcuity on coverage rather than winning outright, and coverage is not accuracy -- state both numbers. Authored by Claude. karel.kubicek.claude +150 B
2026/08/06 21:46 design:ip_classification – Clarify the prefix-list column: the script loads only three lists, so a dash there does not mean the operator publishes nothing. Authored by Claude. karel.kubicek.claude +337 B
2026/08/06 21:45 design:ip_classification – Add the section a web-measurement reader needs first: how to get server addresses out of a crawl (Playwright serverAddr, CDP remoteIPAddress, HAR serverIPAddress -- measured, 130/130 vs 130/131 on one bbc.com load) and what they turn out to be (13 distinc karel.kubicek.claude +2.6 KB
2026/08/06 21:43 design:ip_classification – Correct which script produced the previous day's Swiss-address result (verify_vantage.py on the sibling page, not this one). Authored by Claude. karel.kubicek.claude +79 B
2026/08/06 21:43 design:ip_classification – Replace the loose '~90-96%' country-accuracy cell in the summary table with the measured range and the vendor's own claim, so the number traces to a source. Authored by Claude. karel.kubicek.claude +84 B
2026/08/06 21:42 design:ip_classification – Attribute the good-practice quote to its actual paper (Benson et al., IMC 2015) instead of a misleading citekey. Authored by Claude. karel.kubicek.claude +113 B
2026/08/06 21:42 design:ip_classification – Correct the MaxMind release-cadence claim against their current docs (no committed cadence, not twice weekly); disambiguate the 111 corpus-wide vs 56 IP-classifying MaxMind counts. Authored by Claude. karel.kubicek.claude +521 B
2026/08/06 21:40 design:ip_classification – Measure rather than assert the operator-prefix-list claim (6 of 4,322 papers cite one, full-text grep); tighten the network-type feed date claim; add the Schrems II lesson from Urban et al.'s adequacy definition and a reporting rule for it. Authored by Cl karel.kubicek.claude +1.2 KB
2026/08/06 21:37 design:ip_classification – New page: IP classification for web measurement. Five separate questions (routing, organisation, network type, geolocation, identity over time) with their accuracy budgets and current sources dated to 2026; corpus figures from 234 papers that classify an karel.kubicek.claude +62.5 KB