Four statements that should stay distinct
| Statement type | How to read it | Question to ask |
|---|---|---|
| Observation | A record of something seen or measured within a stated scope. | What was observed, when, and through which method? |
| Inference | A conclusion drawn from evidence and assumptions. | Which assumption or missing fact could change it? |
| Simulation | A result within a constructed model or scenario. | How do its assumptions differ from the actual environment? |
| Unknown | A question the available material does not settle. | What evidence would be needed to resolve it? |
These distinctions reflect Black Atlas’s public emphasis on keeping claim status visible. They are reading aids, not numerical confidence scores.
Keep a small but useful source record
For each consequential claim, preserve the source, relevant passage or record, applicable system and period, and the owner permitted to use it. Link the conclusion to the version of evidence actually reviewed.
Source quality and source independence answer different questions. A first-party page can describe its own service. An independent assessment may help establish an outcome if its method, sample, and conditions support the claim. Neither should be substituted for the other.
Read the same example at three levels
A configuration record says an identity can access a repository. The record is one month old, and the current permission has not been checked.
| Wording | Assessment |
|---|---|
| “The identity can access the repository today.” | Overstates the evidence because the current state is unverified. |
| “The earlier configuration record granted access.” | A bounded statement about what the record says. |
| “Current access needs verification before relying on this path.” | Identifies the missing evidence and its decision consequence. |
Better wording does not repair missing evidence. It keeps the reader from relying on a stronger conclusion than the source supports.
Make contradictions useful
When sources disagree, compare their dates, definitions, system boundaries, and collection methods. Preserve the disagreement until there is a reason to resolve it. An average of incompatible claims can hide the actual problem.
A productive next step names the smallest check that could discriminate between the explanations. If that check is unavailable, narrow the conclusion or keep the question open.
Apply the same discipline to public claims
This site links factual descriptions to primary references, labels its fictional examples, and distinguishes documented capability from demonstrated results. No example here is a client case study or a measured performance claim.
For current deployment constraints, use the main Black Atlas limits page. For how this supporting publication handles sources and corrections, read the editorial policy.
Sources & context
- Black Atlas evidence and provenance ↗Source handling and claim-status distinctions.
- Black Atlas contradiction handling ↗Public explanation of conflicting evidence.
- Black Atlas limits ↗Boundaries on validation and deployment claims.
Sources reviewed 18 October 2026. First-party descriptions establish the source’s stated approach; they are not independent validation of security outcomes. About our sources.