Principles.

Sciync does not accept, hold, settle, or redeem money against any prediction, validation, or claim. There are no tokens, no share classes, no prize pools, no cash leaderboards.

Reputation is the only instrument. It is scored from calibrated expert judgement over time, weighted by specialty and track record. It is not a security, not a commodity, not currency.

The platform has no mechanism — monetary or reputational — to stake on whether an identifiable patient or cohort lives, dies, recovers, or relapses. All claims must be framed against population-level, peer-reviewed endpoints.

This is a mechanism-level commitment, not a moderation rule: the primitives that would make outcome wagering possible are not implemented. Medicine's incentive structure breaks the moment a market is placed on patient outcomes, and our design refuses that failure mode at the schema layer.

Clinical equipoise — genuine uncertainty in the expert community about which arm of a trial is superior — is the ethical foundation of randomized research. Publishing a live, weighted expert consensus during enrollment can collapse equipoise and bias recruitment.

For any claim linked to a trial with enrollment status RECRUITING or ACTIVE_NOT_RECRUITING, public consensus is suppressed, votes are sealed, and only aggregate diagnostics (participation count, specialty spread) are visible until the trial reaches COMPLETED or TERMINATED.

To receive expert weight in the consensus aggregate, a validator must link an ORCID, verify an institutional email, and demonstrate a publication record that touches the specialty in question. Authorship is cross-checked against PubMed, ClinicalTrials.gov, and NIH RePORTER.

The integrity of the consensus signal depends on the integrity of the weighting inputs. A smaller verified expert body produces a stronger signal than a larger unverified one.

Each validation carries its own declared conflicts: industry funding, advisory roles, equity in sponsors, family relationships with PIs, prior work on the endpoint. Disclosure is a structured field, not a free-text afterthought.

Conflicts don't disqualify — they contextualize. Readers see the weighted consensus and the conflict-adjusted consensus side by side.

Anyone — researchers, clinicians, companies, journals, AI systems, students, members of the public — may create claims, submit validations, comment, and subscribe. The input funnel is deliberately broad.

Expert weighting in the consensus mathematics is the gated concern, not participation. ORCID-verified credentials, domain proximity, and calibration history raise a vote's weight in the aggregate. They are never required to contribute a vote.

Reputation cannot be sold, gifted, inherited, delegated, or tied to a position someone else takes. It is a property of the person, not of the account.

Sub-accounts, shared credentials, and institutional amplification are grounds for suspension. We enforce one-human, one-reputation, cryptographically where possible.

Disputes, claim-framing challenges, and equipoise calls are reviewed by a rotating panel of ORCID-verified experts, with public minutes.

We will publish aggregate moderation statistics quarterly: claims rejected, grounds for rejection, appeals upheld. Moderators disclose conflicts like every other expert.

All weighting functions, calibration adjustments, and consensus-threshold rules are open-source and documented. Readers can recompute the consensus from the raw validations.

Individual votes are attributable only to verified experts who have consented to public attribution. Everyone else votes under a stable pseudonym that accrues reputation without revealing identity.

Sciync collects the minimum data required for the system to function: email, ORCID, affiliation, specialty, validations, and rationale. There is no cross-site tracking, no behavioural profiling, and no third-party data brokerage.

Every claim and validation carries a provenance chain back to the validating expert and the source publication. Account deletion removes attributions; the underlying claim persists, reattributed to the anonymous pool so the record of scientific reasoning is preserved.

Sciync is designed for experts in any country. We localize specialty taxonomies, verify credentials against regional equivalents of ORCID, and respect data-residency expectations of researchers working under GDPR, HIPAA, and comparable regimes.

Where a jurisdiction classifies our reputation system as a regulated instrument despite our design, we withdraw that feature from that jurisdiction rather than compromise the non-monetary commitment.