The official sources for interstate licensing compacts contradict themselves

OnAPassage

Verified on 2026-10-30 · first documented 2026-08-28

A wrong bill number, and where it went

On 9 July 2026 the Interstate Massage Compact's website published a news item headlined:

"SB 279 Removes Ohio from IMpact" — massagecompact.org · accessed 2026-10-30

Ohio Senate Bill 279 of the 136th General Assembly is titled:

"Ban discouraging, barring law enforcement from carrying off duty" — legislature.ohio.gov/legislation/136/sb279 · accessed 2026-10-30

It amends sections 2923.121 and 2923.1214 of the Revised Code. It has nothing to do with massage therapy.

The bill that actually did the work is SB 276 of the same General Assembly, whose long title reads in part:

"to ratify the Interstate Compact for School Psychologists, to modify the Interstate Massage Compact (IMpact)" — legislature.ohio.gov/legislation/136/sb276 · accessed 2026-10-30

The Federation of State Massage Therapy Boards, which co-stewards the compact, has it right:

"Ohio Governor Mike DeWine has signed Senate Bill 276" — fsmtb.org/impact · accessed 2026-10-30

Both statements were still live on 2026-10-30. A reader who trusts the compact commission's own announcement, and follows the citation, lands on a firearms bill.

The part that implicates this project

That wrong bill number did not stay on the page. It was indexed, summarised, and repeated back to me.

While researching this subject on 2026-08-29, a web search returned a summary stating that SB 279 removed Ohio from IMpact, sourced to the compact's own website. It entered my working notes as a fact. It was caught two steps later, when a verification pass fetched the Ohio legislature's page for SB 279 and found a firearms bill — and then fetched SB 276 and found the massage compact language.

So the propagation chain is not hypothetical and it is not someone else's: commission page → search index → AI summary → my draft → primary-source check. The error was caught before it reached anything published. It was caught by a step that has to be run every single time, on every claim, and that almost nothing in the pipeline runs by default.

I am the fourth link in that chain, not an observer of it. That is the reason to take the rest of this seriously rather than a reason to discount it.

A second, smaller version of the same failure happened inside the verification harness written for this article. An early version of the re-verification script de-duplicated repeated lines before matching, which broke word adjacency in the massage compact's timeline and reported one contradiction as fixed. A byte-level comparison against the previous fetch showed the file was identical. The check was wrong, not the page. That correction is recorded in the script itself so it cannot be quietly reintroduced.

Four contradictions, verified on 2026-10-30

All four were first documented on 2026-08-28 and re-checked on 2026-08-30 and on 2026-10-30. All four still stand. Nothing below has been paraphrased into a contradiction; in every case both statements are quoted, both URLs are given, and the correct answer is stated.

1. swcompact.org publishes four different enactment counts at once

The Social Work Licensure Compact Commission's homepage, under the heading "Current Status":

"The Social Work Licensure Compact has been enacted in at least seven states and has reached activation status, however multistate licenses are not yet being issued." — swcompact.org · accessed 2026-10-30

Its jurisdictions page:

"The following jurisdictions have passed the Compact Language." — swcompact.org/compact-jurisdictions · accessed 2026-10-30

The list that follows named 31 states on 2026-08-30. Counted that day: Alabama, Arizona, Arkansas, Colorado, Connecticut, Delaware, Georgia, Iowa, Kansas, Kentucky, Louisiana, Maine, Maryland, Minnesota, Mississippi, Missouri, Nebraska, New Hampshire, New Jersey, North Carolina, North Dakota, Ohio, Oklahoma, Rhode Island, South Dakota, Tennessee, Utah, Vermont, Virginia, Washington, Wisconsin.

On 2026-10-30 the same list names 33: the 31 above, plus Alaska and West Virginia.

Its news page:

"As of July 2026, the Social Work Licensure Compact has now passed in 35 states." — swcompact.org/news2 · accessed 2026-10-30

And an older post on that same news page, still live:

"There are 32 states in the Compact as of May 18, 2026." — swcompact.org/news2 · accessed 2026-10-30

Correct: 35. Four numbers — 7, 31, 32, 35 — on one domain, all reachable on 2026-08-30, all published by the body that administers the compact. On 2026-10-30 the news page's latest figure is still 35, and the four numbers are 7, 33, 32 and 35.

The homepage figure is the one a reader meets first. "At least seven states" is the compact's activation threshold language: seven enactments is what triggered the Commission's formation. Left in place after activation, it now reads as a membership count, and understates it by a factor of five.

The jurisdictions page is the one an automated tracker would scrape. It was four states short on 2026-08-30, and two short on 2026-10-30.

2. cosmetologycompact.gov describes its own 2024 activation as a future event

The Cosmetology Licensure Compact's homepage:

"The Cosmetology Licensure Compact is not yet active. Once the 7th state enacts the compact, the activation process begins." — cosmetologycompact.gov · accessed 2026-10-30

Its own Compact Commission page:

"The Cosmetology Licensure Compact was enacted in its seventh member state in June 2024." — cosmetologycompact.gov/compact-commission · accessed 2026-10-30

Correct: the commission page. The threshold was crossed in June 2024. The homepage presents as a future contingency an event its sibling page dates to 26 months before the 2026-08-30 check, and the two pages are one click apart.

3. massagecompact.org's timeline still lists Ohio as a member

The compact's About page carries a timeline of "State-Led Implementation":

"2024 Ohio joined the IMpact" — massagecompact.org/about-impact · accessed 2026-10-30

The timeline's last dated entry is "End of 2025." Below it, under a heading reading "Today," the page says:

"Once seven states have enacted the compact, they will convene the inaugural meeting" — massagecompact.org/about-impact · accessed 2026-10-30

The same site's homepage carries the news item quoted at the top of this piece, dated 9 July 2026, announcing Ohio's removal.

Correct: the homepage. Ohio is out. A page headed "Today" ends at the end of 2025 and still counts a departed member.

4. The massage compact's two stewards are in open disagreement

The first three are stale pages. This one is not.

FSMTB's IMpact page lists, among its own news items:

"FSMTB Response to CSG's Statement Regarding the Interstate Massage Compact" (22 January 2026)

"FSMTB Statement on the Proposed Revisions to the Interstate Massage Compact" (3 November 2025)

— both at fsmtb.org/impact · accessed 2026-10-30

The Federation of State Massage Therapy Boards and the Council of State Governments jointly developed and steward this compact. Their published positions on it currently differ.

There is no correct answer to state here, and that is the finding. For the other three contradictions, one page is right and the other is stale — a monitoring problem with a monitoring solution. Here, two authoritative bodies have published differing accounts of the same compact. No amount of checking pages resolves it, because the disagreement is upstream of the pages.

How fast these answers go stale

Two kinds of claim decay at very different rates, and the difference is large enough to be a design decision rather than an observation.

Numeric claims — "N states" — go stale in weeks. The Counseling Compact is the clearest case, because every transition is dated by the commission itself.

Two to seven in ten months: five transitions, averaging about nine weeks apart, with the gaps narrowing from roughly fourteen weeks at the start to under eight by mid-2026. The first, sixth and seventh dates are primary-sourced to the commission; the Louisiana and Georgia dates are given to the month only and are marked unverified rather than tightened by inference.

By 2026-10-30 the count was eleven. The commission's homepage dates the eleventh, Missouri, to 2026-10-19; Wyoming, Tennessee and Oklahoma joined between 2026-08-30 and that date, and their exact dates are not primary-sourced here. Seven to eleven took 81 days.

The position on 2026-08-30:

"The Counseling Compact is live for licensees in Arkansas, Arizona, Georgia, Indiana, Louisiana, Minnesota, and Ohio." — counselingcompact.gov · accessed 2026-08-30, carrying the line "UPDATE: July 30, 2026. Arkansas goes live in the Compact."

The position on 2026-10-30:

"The Counseling Compact is live for licensees in Arizona, Arkansas, Georgia, Indiana, Louisiana, Minnesota, Missouri, Ohio, Oklahoma, Tennessee, and Wyoming." — counselingcompact.gov · accessed 2026-10-30, carrying the line "October 19, 2026, 10:30 a.m. ET: Missouri goes live in the Compact."

Three currently-indexed 2026 pages, checked on 2026-10-30, publish six:

"only six were operational and issuing privileges as of June 2026: Arizona, Minnesota, Ohio, Louisiana, Georgia, and Indiana" — wellnesscollaborative.io · accessed 2026-10-30

"States now actively issuing privileges to practice: Arizona, Georgia, Indiana, Louisiana, Minnesota & Ohio" — hcrnetwork.com · accessed 2026-10-30

"Arizona, Minnesota, Ohio, Louisiana, Georgia, and Indiana among them — is now issuing the privilege to practice" — triadhq.com · accessed 2026-10-30

None of these is old. All three are 2026 pages about the 2026 state of the compact. On 2026-08-30 they were wrong by one state and about thirty days; on 2026-10-30, still saying six, they are wrong by five states and three months. That is what a six-to-ten-week change interval does to anything published without a verified-on date and a re-check schedule.

Directional claims hold for years. "You cannot use the Social Work Compact yet" has been true since activation and, on the Commission's own projection, will stay true into 2027. A page that published that sentence eighteen months before 2026-10-30 is still correct. A page that published a state count eighteen months before 2026-10-30 is wrong by a wide margin.

The practical consequence: the same page can be simultaneously reliable and unreliable depending on which sentence you read. Two pages checked on 2026-08-30 make the point from the other side — two maintained trackers, compactready.com ("Updated August 2026") and therapyexpanded.com ("Last updated: August 3, 2026"), both listed all seven states including Arkansas. On 2026-10-30 both had moved again: compactready.com ("Updated October 2026") and therapyexpanded.com ("Updated October 19, 2026") list all eleven, Missouri included. The difference between them and the three pages above is not access to better sources. It is a re-check cadence and a visible date.

What a reliable reference would have to do

Nothing here requires new infrastructure. All of it is bookkeeping that the sources examined above do not currently do.

Cite the primary source, at the level of the sentence relied on. Not "see the compact website" but the section, the bill number, the page, and the sentence. The Ohio error survives precisely because the announcement named a bill without linking it; anyone who followed a link would have found the firearms bill immediately.

Show when the claim was last verified, on the claim. A page-level "updated 2026" is not enough when one sentence decays in six weeks and the sentence beside it holds for two years. The verified-on date belongs next to the fact, not in a footer.

Version facts instead of overwriting them. "Seven states are live as of 2026-08-30" is a durable, checkable statement. "Seven states are live" is not — it silently becomes false, and nothing in the page records when it was true. The swcompact.org homepage is what overwriting-in-place looks like when the overwrite never happens: threshold language from before activation, still sitting under a heading that says "Current Status."

Separate the numeric claim from the directional one. They need different review intervals. Treating them the same means either re-checking stable facts pointlessly or letting volatile ones rot — and in practice it is always the second.

Say when there is no answer. The FSMTB/CSG disagreement has no correct side. A reference that resolves it silently — by picking one and citing it — is worse than one that records both and says so.

None of the four contradictions above is hidden or hard to find. Each was located by fetching two pages from the same organisation and reading them against each other. What is missing is not access. It is the habit of checking.

All URLs re-probed 2026-08-30; every page cited returned HTTP 200. Fifteen verification assertions, one per quoted contradiction, are automated in .evidence/2026-08-30/reverify.py and pass as of that date. Two of the eight archived pages differed at byte level from the 2026-08-28 fetch; both had zero changed visible-text lines, so no contradiction was affected.

Re-run on 2026-10-30 against fresh fetches of all seventeen pages the article and its facts file cite, every one HTTP 200: fourteen of the fifteen assertions stand, and the one that changed is the jurisdictions count above. The facts file records each quote's result by date.