Published findings
Cases where a widely republished jurisdictional claim did not survive contact with the primary source. Subjects are unnamed. Every source is cited and dated so the finding can be checked independently.
We publish findings, not subjects. Every organisation described here was contacted before publication. The point is the failure mode, which is structural and general, rather than any individual company's lapse.
004 — Three published crisis helplines were dead
Health resources · Canada · 2026-08-02
An automated verification pass across published provincial problem-gambling helplines found three numbers still in wide circulation that no longer connect to the service named.
In one province the operating organisation had lost its government funding the previous year and was reported to be winding down. In two territories the dedicated line had been consolidated into the provincial 811 service, and the superseded numbers no longer appeared on any government page.
All three numbers remained published on operator sites, affiliate sites and directories, having been copied forward from source to source without anyone re-reading the source.
Why this one matters. A wrong bond amount costs someone a resubmission. A wrong crisis number costs someone the call. This is the clearest case we have that content decay is not merely an SEO problem, and it was found by scheduled re-verification rather than by anyone reporting it.
003 — Four vendors, four answers, one question
Surety bond · Texas · 2026-08-03
Whether Texas imposes a state-level contractor licence bond is answered differently by four national vendors ranking on the same query.
Two state correctly that there is no statewide general contractor bond requirement and that the state licensing body's specialty trade regime turns on insurance rather than bonding. One states the opposite outright. A fourth attributes the requirement jointly to state, city and county in a formulation that cannot be acted on.
Texas is not an obscure jurisdiction. It is among the highest-volume queries in the vertical. The divergence exists because the correct answer is structurally awkward — no state-level requirement, a municipal patchwork beneath it, and a separate state regime for specialty trades — and awkward answers are where maintained content fails first.
Municipal positions beneath the state-level answer remain unverified in our holdings. Texas spans several hundred local permitting regimes with no common publication standard. We report the state position and mark the rest unresolved.
002 — A page set that contradicts itself and links to the contradiction
Firearms retail · Washington · 2026-07
A national retailer publishes two pages addressing the same state-level question. They give opposite answers. Each links to the other in body copy, so a reader following the site's own navigation encounters both positions in sequence.
Neither page is dated. Neither cites a source. There is no indication which is current.
This is the most diagnostic finding in our holdings, because it cannot be attributed to research error. Someone got the answer right at least once. What is missing is reconciliation across a content set that outgrew the point at which any single person held the whole of it.
001 — Nine words
Firearms retail · Connecticut · 2026-07
A retailer's Connecticut page stated that suppressor use while hunting was permitted. The governing section of the Connecticut General Statutes is one sentence.
No person shall use any silencer on any firearm when hunting. Conn. Gen. Stat. § 26-75
We include this one because it sets the floor. If a nine-word prohibition can be published in reverse on a commercial page in a regulated category, the assumption that jurisdictional content is broadly reliable does not hold, and the cost of checking is lower than the cost of being wrong.