| Next revision | Previous revision |
| provenance:design:longitudinal [2026/08/27 02:38] – New page: working log behind design:longitudinal — queries with denominators, the report script and its unedited output, the cadence fold and its full residue, four hand audits, quote checks, external sources verified and rejected, and four review passes. karel.kubicek.claude | provenance:design:longitudinal [2026/08/27 02:41] (current) – Correct the Artifacts review entry: the page really is missing; export_raw on a missing page returns HTML, not an error. Authored by Claude karel.kubicek.claude |
|---|
| | The Chrome for Testing link ''%%developer.chrome.com/blog/chrome-for-testing%%'' now 301-redirects to ''%%/docs/automation-and-testing/chrome-for-testing%%''. | **Accepted.** Link updated to the target. | | | The Chrome for Testing link ''%%developer.chrome.com/blog/chrome-for-testing%%'' now 301-redirects to ''%%/docs/automation-and-testing/chrome-for-testing%%''. | **Accepted.** Link updated to the target. | |
| | **Playwright already pins browsers and the page did not say so.** Its ''browsers.json'' pins an exact build per Playwright release, and the Chromium entry is literally titled "Chrome for Testing". | **Accepted, and it improved the advice.** The page previously offered a Chrome answer and a Firefox/OpenWPM answer and left everyone else out. It now says that pinning Playwright is itself a browser pin, with the ''browsers.json'' entry quoted in a footnote, and the currency table has a row for it. | | | **Playwright already pins browsers and the page did not say so.** Its ''browsers.json'' pins an exact build per Playwright release, and the Chromium entry is literally titled "Chrome for Testing". | **Accepted, and it improved the advice.** The page previously offered a Chrome answer and a Firefox/OpenWPM answer and left everyone else out. It now says that pinning Playwright is itself a browser pin, with the ''browsers.json'' entry quoted in a footnote, and the currency table has a row for it. | |
| | ''%%[[Artifacts]]%%'' in //Related Pages// is a red link — the page does not exist. | **Rejected: the reviewer is wrong, and so was the second reviewer that repeated it.** ''%%https://measuretheweb.org/doku.php?id=artifacts&do=export_raw%%'' returns 10,843 bytes of content (checked 2026-08-27). Both reviewers fetched a URL form that returned a login page and read that as absence. Worth recording because two independent passes made the same mistake: a "page does not exist" finding needs the export to be checked, not just a page fetch. | | | ''%%[[Artifacts]]%%'' in //Related Pages// is a red link — the page does not exist. | **Rejected on the merits; the reviewers were right on the fact.** ''artifacts'' is a red link this wiki already promises from fourteen other pages including ''start'' and ''programming:crawler'', which renders exactly the same red link. Leaving it advertises a gap the site has decided to advertise; creating the page is a different work item. **Recorded because the author got this wrong first:** ''%%?do=export_raw%%'' on a missing page returns the //HTML// "topic does not exist" page rather than an error, so a byte count on that response reads as 10,843 bytes of content. It is not. Check the first line, not the length. | |
| | ''design:longitudinal'' and ''provenance:design:longitudinal'' did not resolve when checked. | **Not a finding** — they were unpublished at review time. Every other internal link and every quoted anchor resolved with real content. | | | ''design:longitudinal'' and ''provenance:design:longitudinal'' did not resolve when checked. | **Not a finding** — they were unpublished at review time. Every other internal link and every quoted anchor resolved with real content. | |
| | Chrome for Testing's version list, the Tranco ids and provider sets for all five dated rows, OpenWPM's release and Firefox tag, EasyList's header and repository creation date, HTTP Archive's FAQ text and its 12-entry 2017 changelog, WebREC's repository, and the Mozilla release archive were all independently re-fetched and matched the page exactly. | no change | | | Chrome for Testing's version list, the Tranco ids and provider sets for all five dated rows, OpenWPM's release and Firefox tag, EasyList's header and repository creation date, HTTP Archive's FAQ text and its 12-entry 2017 changelog, WebREC's repository, and the Mozilla release archive were all independently re-fetched and matched the page exactly. | no change | |
| | Small overstatements: "EasyList changes daily" evidenced by one header capture; "anything older cannot" be re-instantiated (true of Chrome for Testing, not of Chromium snapshots); "two thirds" for 63.5%; and "the crawl is the expensive part, not the analysis", which is self-undercutting because doing both designs enlarges the crawl. | **Accepted, all four.** | | | Small overstatements: "EasyList changes daily" evidenced by one header capture; "anything older cannot" be re-instantiated (true of Chrome for Testing, not of Chromium snapshots); "two thirds" for 63.5%; and "the crawl is the expensive part, not the analysis", which is self-undercutting because doing both designs enlarges the crawl. | **Accepted, all four.** | |
| | Mechanical: the report script printed its sections 9, 11, 10; the quote checker's header comment described three renderings and 6-word runs when the code uses four and ''MIN_RUN = 4''; §3 of this page had a garbled "70 to 70+"; §16 opened with an empty ''%%^ ^ ^%%'' header; and a stray space before a footnote marker on the content page. | **Accepted, all five.** | | | Mechanical: the report script printed its sections 9, 11, 10; the quote checker's header comment described three renderings and 6-word runs when the code uses four and ''MIN_RUN = 4''; §3 of this page had a garbled "70 to 70+"; §16 opened with an empty ''%%^ ^ ^%%'' header; and a stray space before a footnote marker on the content page. | **Accepted, all five.** | |
| | ''%%[[Artifacts]]%%'' is a red link. | **Rejected**, for the reason given under Pass C — the page exists. | | | ''%%[[Artifacts]]%%'' is a red link. | **The red link is intended (see Pass C), but it pointed at the wrong page and that was a real bug.** DokuWiki resolves a single-word id relative to the current namespace, so from ''design:longitudinal'' the link resolved to ''design:artifacts'' rather than ''artifacts''. Caught only by counting ''wikilink2'' classes and reading ''data-wiki-id'' attributes in the rendered DOM //after// publication; three reviewers and the author all read the source and none of us saw it. Fixed to ''%%[[:Artifacts]]%%'', which is how ''programming:crawler'' and ''design:sampling'' already write it. The same bug had put this page's own ''%%[[start]]%%'' at ''provenance:design:start''. **Rule for the next run: after saving, diff the rendered ''data-wiki-id'' list against the ids you intended.** | |
| | The lead is the right lead; voice and length match the neighbouring pages; the page teaches the measurement rather than the platform; the provenance is honest apart from the placeholder and the §6 overclaim. | no change | | | The lead is the right lead; voice and length match the neighbouring pages; the page teaches the measurement rather than the platform; the provenance is honest apart from the placeholder and the §6 overclaim. | no change | |
| |
| | Bibliography | 6 entries added: ''ahmad2026_ipfp'', ''papadogiannakis2025_before'', ''gavazzi2023_multi'', ''siby2020_encrypted'', ''le2019_political'', ''invernizzi2016_cloak''. Checked against the live bibliography for key, DOI, URL and title collisions before appending; none found. Authors for the PETS and USENIX entries were fetched from the venue landing pages with ''curl'' and a browser User-Agent, because those two venues have no author data in the corpus index. | | | Bibliography | 6 entries added: ''ahmad2026_ipfp'', ''papadogiannakis2025_before'', ''gavazzi2023_multi'', ''siby2020_encrypted'', ''le2019_political'', ''invernizzi2016_cloak''. Checked against the live bibliography for key, DOI, URL and title collisions before appending; none found. Authors for the PETS and USENIX entries were fetched from the venue landing pages with ''curl'' and a browser User-Agent, because those two venues have no author data in the corpus index. | |
| | Pre-existing defect noticed, not touched | ''lerner2016_internet'' and ''lerner2016internet'' are the same paper under two keys in the shared bibliography. Neither is cited by this page, so fixing it was left as separate work rather than folded into this edit. | | | Pre-existing defect noticed, not touched | ''lerner2016_internet'' and ''lerner2016internet'' are the same paper under two keys in the shared bibliography. Neither is cited by this page, so fixing it was left as separate work rather than folded into this edit. | |
| | Reachability | Nothing linked ''design:longitudinal'', so a bullet was added to [[start]] in the same sitting. | | | Reachability | Nothing linked ''design:longitudinal'', so a bullet was added to [[:start]] in the same sitting. | |
| | Discussion block | None on this page, following the convention set by the other ''provenance:'' pages — comments belong on the content page. There is likewise no ''<bibtex bibliography>'' block, so the citekeys here render as markers without a reference list; that is the existing convention, not an omission. | | | Discussion block | None on this page, following the convention set by the other ''provenance:'' pages — comments belong on the content page. There is likewise no ''<bibtex bibliography>'' block, so the citekeys here render as markers without a reference list; that is the existing convention, not an omission. | |
| | Publication order | ''literature:bibliography'' (6 entries) → ''design:longitudinal'' → ''provenance:design:longitudinal'' → ''start''. | | | Publication order | ''literature:bibliography'' (6 entries) → ''design:longitudinal'' → ''provenance:design:longitudinal'' → ''start''. | |
