Conflicts of interest policy

The conflict at the root

This journal is published by GMedAC, the company building the marketplace this journal writes about. That is not a conflict this policy manages; it is the condition of the publication. Every article touching GMedAC — what it verifies, how a booking works, what it refuses to hold — is written by people GMedAC pays, reviewed by people GMedAC pays, and published on infrastructure GMedAC controls. No disclosure statement makes such an article independent.

What follows is a limit on authority, not a disclaimer. GMedAC's own account of itself enters the source vocabulary as COMPANY_STATEMENT, the weakest type in the list, and an article at contentRisk: HIGH or CRITICAL may rest only on sources at or above PEER_REVIEWED_STUDY. The company therefore cannot support its highest-stakes claims by describing itself. Where an article does rest on COMPANY_STATEMENT, the sourceIds say so on the page, with the dateRetrieved and the retrievalRoute that record how the statement was actually read.

The same limit binds the questions the company would most like answered. A verification result is published as a summary of what was checked, never as an endorsement, and whether verifying and publishing nonetheless makes GMedAC an endorser rather than an intermediary is an open legal question marked COUNSEL_REQUIRED. An article that resolved that question in the company's favour would be advocacy in the voice of a journal. A jurisdiction-specific claim resting on REASONED_INFERENCE, COUNSEL_CONFIRMATION_REQUIRED or UNKNOWN cannot be published at all — the claim check refuses it — and that refusal bites hardest exactly where the commercial appetite for a published answer is strongest.

The declared field

conflictsOfInterest is declared in two places, because a conflict recorded once is a conflict forgotten. On the article it records the relationship between this piece of work and its subject: who commissioned it, whose product it describes, what the publication stands to gain if a reader believes it. On the person record it records standing interests that travel with the individual — employment by GMedAC, a listing on the marketplace, a paid engagement, a holding, a role in a professional body whose position the article cites.

The honest limit: the field is prose, and no static build can distinguish a complete declaration from a thin one. Its presence is required; an empty list satisfies that check. Its truth cannot be checked. A person who omits an interest defeats this policy entirely, and nothing in the repository will notice.

Reviewers are disclosed, not swapped

A reviewer with a stake in the article's subject is named in reviewerIds and their interest is carried in their person record. The failure this prevents: quietly substituting a clean reviewer produces an article that looks unconflicted and destroys the evidence that a conflict was ever in the room. Where the stake is severe enough that the review alone is worth little, a second reviewer is added and both remain named — the conflicted name does not disappear from the record.

Reviewer identity is load-bearing elsewhere, which is why tidying it is not cheap. A correction file records previousClaim verbatim alongside correctedClaim, reason, editorId, reviewerId, material and dateModifiedChanged, and a material correction with no reviewer is refused. A conflict discovered after publication is handled through that file, or by moving the article to WITHDRAWN, not by editing the page in place: a silent rewrite leaves the reader holding a claim nobody recorded changing. The build reaches only part of that. It refuses a material correction that leaves dateModified unmoved, and one filed with no reviewer, but an edit accompanied by no correction file at all passes unnoticed. Not making that edit is a commitment, not a control.

Professionals listed on the marketplace

A professional listed on the marketplace has a direct commercial interest in demand for their own field. Commissioning that professional to write the explainer that sends readers towards their own category is an advertisement wearing an author byline, and the byline is the part that makes it work. Such an author is not commissioned on their own field without the listing declared in the person record and restated in the article's conflictsOfInterest. An article never singles out a named listing as a preferred choice, and never orders or shortlists professionals — that is a ranking, and ranking belongs to the marketplace, where commission does not influence organic ranking and a test pins that behaviour. No equivalent test governs editorial commissioning. Here the control is refusal and disclosure by hand, which is weaker.

Adjacent to this: the journal carries no advertising, no sponsorship, no affiliate link and no paid placement today. Whether a clearly labelled paid surface may ever exist is an undecided question for the founder, not a settled prohibition. If one is ever created, this document is rewritten before it ships or the claim above becomes a lie.

Disclosure is not absolution

Some conflicts are not curable by declaring them. Disclosure records a conflict; it does not neutralise one. The article is not written where the company's interest would shape the conclusion itself:

• Anything that reads an open launch gate as a settled one. Every gate in the Czech launch matrix is open — 3 RESEARCHED_PENDING_COUNSEL, 7 COUNSEL_QUESTION_OPEN, 2 EVIDENCE_MISSING, zero approved. Excluded means unresearched. It never means permitted, and EVIDENCE_MISSING is not a clearance.
• Anything treating the United States as one corridor. It resolves state by state, and the US material in the marketplace record is corridor discovery rather than a clearance.
• Anything offering the absence of clinical fields as a privacy reassurance. GMedAC holds no clinical record, and whether a booking pairing is Article 9 data under GDPR is unknown and marked COUNSEL_REQUIRED. Reaching for the comfortable reading of an open question is the conflict operating.
• Anything describing the marketplace as live. It is at Phase 0: no checkout, no live payment provider, no deployment.

What this policy does not reach

The controls here are declarative and largely human. The build refuses unsourced jurisdictional claims, stale high-risk articles and unreviewed material corrections; it cannot detect an interest nobody wrote down. The journal has no database, no authentication and no analytics, so no one can measure which article produced which booking — which also means no one can produce evidence that commissioning was never shaped by that. Removing the measurement removes a temptation and the proof at the same time. This is honest, not clean.