How we work
We read original test reports, research papers, official statements and release histories. We ask what actually happened, under what conditions, and what the source does not establish.
A statement is not the same as a successful test. Repeated news stories are not independent confirmation. Unknown means information is missing, not that a capability cannot exist.
What the ratings mean
- Not enough information
- Insufficient inspected evidence for the exact criterion; not evidence of nonexistence.
- Reported, not yet verified
- An attributable original statement; accessible evidence does not establish the capability itself.
- Shown in practice
- A primary physical test supports the scoped property. For evolution, dated release records establish the release sequence, not performance.
- Independently confirmed
- Two independent primary evidence lines for the same system/version and compatible conditions. Reprints and trials within one study do not count.
How much confidence to place in a finding
- Low
- Attribution only, ambiguity or major gaps. For unknown, this is not confidence in absence.
- Medium
- A specific primary basis with important limits or no independent replication.
- High
- Independent converging observations and well-bounded alternatives.
What we still need to know
Bounded catch-up review for 12 October–9 November, including older studies newly located through recent publications. Public protocols and global coverage remain incomplete.
Rules used for this review: 1.0.0
Download the complete research record ↓
About the project & contact
Project by: Maksym Kuzmenko
Have a source, a correction or a different reading of a finding? Get in touch and include a link to the relevant assessment.
max.kuzmenko@gmail.comLinkedIn ↗