™A ranking needs a defined context
A possible next investigation depends on available data, model fit, unresolved constraints, time, cost, rights, controls, and the endpoint being pursued. The private workspace shows why an option rose or fell, displays missing information and refusal states, and allows a researcher to disagree.
A rank is not a prediction of experimental success. Before any computational result is described as a scientific prediction, canXcer still requires a versioned method, reproducible parameters, controls, applicability limits, uncertainty, and validation for the stated question. A literature hit or AI explanation does not meet that standard.
Keep predictions separate from measurements
The outcome ledger preserves a prediction before a later lab result is entered. Positive, negative, and inconclusive measured outcomes remain distinct, and a correction creates a visible new record rather than rewriting the original.
This comparison can help evaluate a method over time, but it cannot establish accuracy before enough appropriate outcomes have been measured and reviewed.
What is available today
Approved project researchers can run an eligibility check, inspect deterministic candidate ranks and their complete calculation manifest, save immutable sessions, respond, append a distinct researcher-authored alternative, and compare revisions. Reviewers can append accountable reviews, and owner or administrator oversight shows bounded operational counts.
The engine does not execute a method, authorize a physical experiment, change permissions or a Research Signal Score, create evidence, publish a result, recommend treatment, or claim that an investigation will succeed. Formal eligibility and ranking work without AI providers. Optional AI explanation is not part of the formal decision or score.