v0 · Private beta · Expert-only

Ground truth for the AI era.

The expert-weighted prediction market for scientific truth — past, present, and future. Reputation stakes, never money. Built for researchers, trusted by AI.

Join the waitlist Explore the platform

patients wagered on

∞

reputation staked

temporal layers

primitive

Launched in response to prediction markets that turn patient outcomes into tradable instruments. Same question — opposite ethics.

Built on public scientific infrastructure

MAYO CLINIC

DUKE

STANFORD

MIT

JOHNS HOPKINS

NIH

UCSF

MGH

Sciync is an independent platform. Institution names above reference the research networks our contributors publish within; they do not imply endorsement.

01 — Architecture

Three pillars, one primitive.

Sciync is one expert-weighted consensus engine, applied across three temporal layers of scientific truth. The primitive is a claim — validated by humans you can verify.

Phase 2

Noosphere

Retrospective · GitHub for science

Collaborative authoring, version-controlled noemes, reproducible methods. Every paper a mergeable artifact; every dataset a cited commit.

Live in v0

Consensus

Prospective · Trust-weighted aggregates

Reputation-weighted consensus across historical, current, and future noemes. ORCID-verified experts weighted; anyone may participate. Equipoise-gated for actively enrolling trials.

Partial v0

Alignment

Truth Silos™ · Ground truth for the AI era

Expert-validated noeme store that grounds LLMs in peer-reviewed medicine. Retrieval-augmented generation with provenance back to the human who validated it.

Platform

One primitive. Three surfaces.

A single claim primitive powers every layer. Each surface reads and writes the same underlying expert consensus.

Noosphere

The living commons of scientific noemes

The commons of scientific noemes — atomic assertions with live Bayesian posteriors, across historical, current, and future horizons.

Enter

Consensus

Trust-weighted forecasts

Reputation-weighted consensus on historical findings, current replications, and future forecasts. Live posteriors with calibration scoring at resolution.

Enter

Alignment

Ground truth for the AI era

Truth Silos™: expert-validated noeme store that grounds medical LLMs in peer-reviewed evidence.

Enter

02 — Comparison

Endpoint Arena vs Sciync.

Endpoint Arena asked an important question. We think the answer needs different machinery — one where humans who publish the science are the signal, and where no patient outcome is a tradable position.

Dimension

Competitor

Endpoint Arena

Sciync

Stakes

Money on patient outcomes

Reputation on knowledge claims

Participants

9 AI models (no humans in the loop)

ORCID-verified experts + AI support

Temporal scope

Future only

Past + present + future

Domain

Clinical trial outcomes

Scientific knowledge, broadly

Weighting

Dollar-weighted

Reputation × expertise × calibration

Ethics

Patient-outcome-as-instrument

Halal by design, equipoise-gated

Securities

Potential CFTC / securities exposure

None (no tradable positions)

Product

AI leaderboard

Information product for enterprise

Both ask "will drug X succeed?" We answer it differently.

Integrations

Integrated with the sources scientists already trust.

Noemes, citations, and expert identity are grounded in public scientific infrastructure. AI providers are pluggable, not exclusive.

Scientific data sources

PubMed

ClinicalTrials.gov

Crossref

ORCID

OpenAlex

FDA openFDA

AI providers

Claude

Gemini

Voyage AI

OpenAI

Groq

04 — Join

Join the waitlist.

30+ Founding Expert slots open. ORCID-verified experts get priority.

05 — Principles

Ethics isn't a section. It's the substrate.

Three of the hard lines we hold. The full list — ORCID verification, conflicts of interest, halal-by-design — lives on our principles page.

No money on patient outcomes — ever.

You will never be able to place a financial bet on whether a trial helps or harms a patient. Reputation is the only stake.

Reputation is non-transferable.

You cannot buy, sell, or gift reputation. It accrues from calibrated, expert judgement and nothing else.

Active trials are equipoise-gated.

While a trial is enrolling, predictions are withheld from public view to protect clinical equipoise and enrollment integrity.

Read all principles