RHTP Navigator
Start your Atlas

Trust

Public corrections log.

When we get something wrong, we fix it in the open: the date, the page, what was wrong, and what is right.

A map you can act on has to be honest about its own mistakes. Every correction we make to a published fact is recorded here, permanently, with enough detail that you can re-check the work.

Current verification queue

Our final August 9, 2026 national pass checked 812 citations across 602 unique source URLs. It successfully re-verified or re-fetched 743 citations. No live quote mismatch or confirmed dead link remains in the published corpus.

  • 0 quote mismatches and 0 confirmed 404s remain after primary-source reconciliation and link replacement.
  • 181 of 193 cited binary documents are held locally with byte and extracted-text hashes. All 181 have extracted text.
  • 126 held anchor checks found 113 verbatim matches, 13 layout-only wording differences with every figure present, and 0 absent figures.
  • 30 bot-walled sources and 8 transient or other request failures remain fail-closed. Twelve binary documents could not be newly acquired.

These are known verification gaps, not automatically known factual errors. If resolving one changes a fact readers previously saw, that correction moves into the permanent log below. The methodology page explains the fail-closed rule.

Corrections

14 corrections, newest first. Most recent: August 11, 2026.

  1. Oklahoma Rural Regional Reorientation: the $20 million pool, on 17 pages

    Was
    The pool was labeled a live solicitation, posted now, and on the sample page as one you could apply to.
    Now
    The same figure carries the status from the state record, currently closed with awards pending, rewritten on every build.

    The state record moved that window to closed on August 9 and the site kept selling it as open for two days, because the sentence beside the figure was hand-written prose repeated across seventeen pages while the status lived in the data. A reader could have chased a pool that had already closed. The label is now read from the same timing record the capture pages use, so the claim expires when the evidence does.

  2. Vermont: rural demographics and hospital risk

    Was
    13 rural counties, 419,260 rural residents, 13 rural inpatient hospitals, and 13 hospitals at risk.
    Now
    Those four RHTP-specific fields are unreported. The 419,260 Census nonmetro population remains visible only as a separately labeled comparison.

    The state publishes a rural definition, a statewide hospital total, qualitative population language, and projected operating losses. Those facts do not directly publish the four values we had labeled. Derived arithmetic and Census nonmetro geography cannot substitute for the state RHTP denominator, and projected losses are not a closure-risk count.

  3. Indiana: RHTP rural population

    Was
    1,846,221 rural residents from a January statewide presentation.
    Now
    No state-defined RHTP rural population is shown. The current state page still confirms 64 fully rural counties, and the Census nonmetro population remains separately labeled.

    Indiana retired the presentation and no current primary source republishes that population figure. The value was removed instead of carrying forward proof that readers can no longer inspect.

  4. Arizona: behavioral health and SUD expansion target

    Was
    Ability to stabilize 250 to 450 crisis episodes annually.
    Now
    300 to 600 encounters per year across all service areas after launch.

    The current approved budget narrative publishes the newer operational output. Held-document verification found that the earlier target no longer appeared in the authoritative source.

  5. Maryland: Mobile Health allocation

    Was
    $23 million shown as a separate Year 1 Mobile Health allocation.
    Now
    No separate Mobile Health allocation is shown; the amount is withheld pending a current primary-source allocation.

    The current Maryland portal still identifies Mobile Health as an activity but no longer publishes the prior $23 million line. Keeping the number would imply current proof that no longer exists.

  6. Virginia: DMAS and VDH operating leadership

    Was
    Cheryl Roberts as DMAS Director and R. Christopher Lindsay as VDH Chief Operating Officer.
    Now
    Steve Ford as DMAS Director and Matthew Ahern as VDH Chief Operating Officer.

    Current Virginia organization and leadership sources show both offices changed. The routing records and their primary-source links now reflect the Q3 2026 leadership roster.

  7. Federal scoring rubric: source link on every state record

    Was
    The scoring-rubric proof linked to a retired CMS-hosted NOFO PDF that now returns 404.
    Now
    The proof links to the current Grants.gov opportunity 360442 record.

    The scoring factors and state data did not change, but the document path readers used to verify them was dead. A broken proof link violates the same provenance promise as a mismatched quote, so we corrected it across the corpus and logged it here.

  8. Colorado: program leadership

    Was
    Kim Bimestefer listed as the sitting Executive Director of the Colorado Department of Health Care Policy and Financing.
    Now
    No Executive Director listed. The program office and the RHTP program director remain.

    She left the department on April 10, 2026, so listing her as the sitting director was not merely stale, it was wrong, and a reader could have tried to reach the wrong person about a funding decision. We are not naming a successor yet: reporting names an acting director, but no state source we can retrieve says so, and we would rather show you nothing than show you a guess. This is the same rule that leaves at-risk hospital counts empty for 32 states.

  9. Maryland: the source quote behind the award figure

    Was
    The award of $168,180,838 was shown with the supporting quote “Maryland received $168 million in federal funding in December 2025, renewable over five years, supporting 18 state-designated rural areas across the state.”
    Now
    The same award, shown with the quote “Maryland $168,180,838”.

    The award figure was correct and is unchanged. The sentence we showed as its source was not: it was a summary written by us, and it appears nowhere in the federal announcement it was attributed to. The real wording was on that same page the whole time. We are logging this even though no number moved, because the promise on every figure here is that you can check it against the document, and for this one you could not.

  10. Funding window status, 24 state pages

    Was
    77 funding windows were labeled Applications open, each with a pulsing live indicator, across 24 states. $196.8M in five-year funding was attached to them.
    Now
    15 are open. 62 were resolved to closed against the closing dates the states themselves published. The earliest had closed on March 27, 2026, most of the rest in June.

    We recorded each window's status on the day we read the source, then rendered it in the present tense indefinitely, so a window that shut in June still read as live in July. Status is now resolved against the calendar every time we publish, and it moves only toward closed, never toward open, because being late to call something open costs you far less than being wrong that it still is. Any status we asserted more than seven days ago now carries the date we asserted it and loses the live indicator, so you can see the difference between what is happening and what we last checked.

  11. Texas: program leadership

    Was
    Cecile Erwin Young listed as Texas Health and Human Services Executive Commissioner.
    Now
    Stephanie Muth.

    Muth was appointed in May 2025 and our record was an older snapshot that we never re-checked. Confirmed against the agency's own leadership page and two 2026 releases from the Governor's office about the program. State leadership turns over mid-program and a contact list is only useful if it is current.

  12. Ohio: rural population and funding per rural resident

    Was
    4,400,000 rural residents. $46 per rural resident.
    Now
    2,125,291 rural residents. $95 per rural resident.

    The 4.4 million figure was carried without a source that actually stated it, and it was roughly double the real count. We replaced it with the Census American Community Survey nonmetropolitan population, which doubled the per-resident figure. This is the largest single error we have published, and it is the reason every number on this site now has to name the document it came from before it can render.

  13. Arkansas, California, Louisiana, Michigan, Pennsylvania: funding per rural resident

    Was
    California $85, Louisiana $189, Arkansas $161, Pennsylvania $81, Michigan $102 per rural resident.
    Now
    California $274, Louisiana $287, Arkansas $176, Pennsylvania $110, Michigan $104.

    Each of these came from the state's own narrative, where rural population is usually rounded to the nearest hundred thousand and sometimes counted more broadly than the federal definition. Rounded denominators are fine in a state's own document and useless for comparing states to each other, which is what this site is for. Every denominator is now the Census American Community Survey nonmetropolitan count. California moved twice, from $103 to $85 on May 29 and then to $274 here.

  14. Missouri: funding by initiative

    Was
    Five initiatives carried five-year allocations: $57,225,309 for Provider Transformation, $56,159,145 for Digital Backbone, $52,040,578 for Regional Coordinating Networks, $17,769,474 for Rural Workforce, $8,791,933 for Alternative Payment Models.
    Now
    No per-initiative allocation. Missouri shows its award and its initiatives without a dollar split.

    Those are Year 1 figures, not five-year ones. They come from Missouri's Year 1 Fund Distribution table published on April 7, 2026, whose six pillars sum to $216.2M, which is a single year of Missouri's award. Calling them five-year understated each initiative by roughly a factor of five. Missouri has not published a five-year split, so we publish none rather than infer one.

Spotted something

If a fact looks wrong, tell us and we will check it the same day. Write to hello@rhtpnavigator.com with the page and the figure. If you are right, the fix and the credit land here.

How we verify in the first place is on the methodology page. How every fact carries its source is on the provenance page.