Conceptual illustration. Shapes, colors, and motion are not measured data.
The research atlas connects claims, sources, models, datasets and tools. Each connection needs a record of origin and transformation. A catalogue downloaded from a survey, a mask derived from an image and a synthetic point cloud are not interchangeable inputs. The W3C provenance framework distinguishes entities, activities and responsible agents, while FAIR principles address making research objects findable, accessible, interoperable and reusable. The proposed Hill implementation can use these ideas in a lightweight form: every displayed result should identify its input version, processing step and evidence status, so a reader can follow the chain without assuming that a polished visualization is observational data.
Keep raw inputs separate from derivatives
Preserve original files and record their retrieval location, date, version, checksum, license and relevant access restrictions. Save cleaning and transformation steps as code or explicit instructions rather than overwriting the source. A cropped image, normalized table or resampled time series receives its own identifier linked to the original. For simulated data, record the generator, parameter values and seed. Mark generated conceptual artwork as illustration. This distinction is particularly important for the Alpha 1 model, whose visual particles and spirals must never be mistaken for a measured catalogue of surrounding universes.
Attach provenance to claims
A claim should link to the specific table, figure, equation or passage that supports it when such detail is available. Record whether the source was checked through its abstract, full text, official documentation or a reproduction. A DOI resolving successfully verifies an identifier, not every inference made from the paper. Likewise, a source about material memory does not support a claim about consciousness or cosmic inheritance merely because the word memory appears in both. The relationship between source and claim should be stated in plain language and reviewed independently where consequential.
Propagate corrections
When an input or interpretation changes, identify every dependent output. A corrected calibration may affect a figure, a comparison tool and a public summary. A paper’s correction may alter a parameter but leave the project’s general research question intact. Publish a dated change note describing the scope rather than silently editing away the history. The proposed export manifest should make these dependencies machine-readable where practical. Contributors should be able to determine which version they used and whether a newer correction changes their calculation or only its presentation.
The mathematical lens
A compact provenance chain; agent, date, version and method metadata belong to each relevant step.
Turn it into a question.
- Question
- Can every numerical or evidential claim be traced to a specific version?
- Method
- Audit a sample of claims backward to raw inputs and forward through the transformation chain.
- What would count against it?
- Missing versions, overwritten inputs or unsupported source links break the chain.
Provenance tells readers what a result is made of.
Follow the original work.
These sources support the mechanisms and limits discussed here. They do not endorse Hill’s framework. Notes distinguish research proposals, reviews, and observations.
Continue the investigation
Edition 02 · 18 October 2026 · AI-assisted synthesis. See editorial method.