Each experiment needs a stable identifier, owner, date, hypothesis, protocol or command, code and environment version, input identity, raw artifact links, measured result, interpretation, status, cost, and the decision that followed. Missing fields remain explicit.
Record enough context to reproduce the run boundary
The ledger should identify the exact question and the conditions under which it was tested. Store the repository commit, dependency or container version, configuration, random seed when relevant, input dataset identity, hardware or service, command or protocol, start and end time, and operator. Store a stable link to the full raw configuration file instead of duplicating selected values in free-form text.
Give every raw output a stable location and content identifier when the storage system supports one. Keep logs, metrics, screenshots, model responses, intermediate files, and cost records distinct from the summary. A result without its raw artifact cannot be checked, while an artifact without the experiment identifier cannot explain why it exists.
Connect observation, interpretation, and decision as separate fields
The observed result states what the run produced under the recorded conditions. Interpretation states what the team believes that result means and should name uncertainty or alternative explanations. Decision states what changed afterward: another run was authorized, an approach was held, code was promoted, or work stopped. Keeping these fields separate prevents a later summary from turning an interpretation into a measurement.
Status values should distinguish completed, failed, aborted, invalidated, superseded, and inconclusive runs. Add a reason and source for each change. The final linkage runs in both directions: a decision can reveal the evidence behind it, and an experiment can reveal which decision, if any, consumed its result.
Where the service stops
Reality Contact, LLC implements evidence organization and retrieval, but does not certify scientific validity, reproducibility, research integrity, regulatory compliance, safety, authorship, or the correctness of any experimental conclusion. The buyer approves the record schema and storage boundary, verifies scientific interpretations, controls access and retention, and decides which evidence supports the next experiment, publication, or funding choice. This is technical record organization and implementation; it does not replace legal, medical, scientific, safety, compliance, or professional advice. A complete ledger improves traceability but does not certify that a result is valid, reproducible, safe, ethical, or suitable for publication or regulation.
Sources: MLflow Tracking documentation; DVC experiment management documentation.