A research question.
An inspectable system.
Predome connects source material, defined research roles, computational tools and persistent records. The workflow makes each handoff explicit, including the points that still require a reviewer or further implementation.
Follow the evidence
through the system.
This diagram explains the intended workflow. Select a stage to distinguish current tools, human responsibilities and planned forecasting features.
Give every claim a source.
Available in the research workspace- Input
- Allowlisted public registries, published reference datasets and authorized uploads.
- Output
- Dated source files, provenance, numeric checks and visible information gaps.
Public-source collection and validation are operational. The public website publishes selected snapshots with their collection time.
The source travels
with the claim.
A result needs more than a paragraph. Keep source files, collection time, data provenance, task parameters and generated artifacts connected to the research record.
The current workspace collects allowlisted public sources, validates numeric inputs and retains original uploads. Public case pages publish selected reference material with explicit boundaries.
Collaboration and comparison
are different operations.
Specialized roles can collaborate on a research draft. A fair model comparison requires each model’s first judgment to be made without seeing the others, using the same evidence and outcome definition.
The independent comparison workflow is planned. Existing sequential model-drafting adapters require server configuration, and a successful live model call has not been verified.
A test needs
a declared boundary.
Source integrity, numeric consistency and reproducible calculations can be checked by software. Scientific meaning, causal interpretation and the adequacy of a study require evidence and qualified review.
A computational success does not establish a therapeutic effect. A plausible hypothesis does not become a completed experiment.
Preserve the original.
Explain the change.
The intended forecast protocol preserves original submissions and records revisions as separate versions. An official outcome is evaluated against rules set before the forecast.
Research storage exists today. Dedicated forecast freezing, independent timestamping, adjudication and prospective performance evaluation remain in development.
Open-source research references
DiscoveryRSI is a research reference for traceable hypotheses, executable tasks and revision history. It is not an enabled backend in this preview, and this product does not claim to train an RSI model.
Explore the work
at its current stage.
Follow a public reference case or inspect the existing authorized research workspace.