CAIN-42 invariant · E21
D50: benchmark changes invalidate incompatible comparisons
Held · CAIN-42 Evolution 21 -- Governed Open-Ended Intelligence Fabric
Last reviewed 2026-10-01
held niche Benchmarks, coverage & performance
What this rule means
CAIN-42 must always satisfy: benchmark changes invalidate incompatible comparisons. It is one of 155 invariants checked for CAIN-42 Evolution 21 -- Governed Open-Ended Intelligence Fabric. An invariant is a rule the system may never break, whatever an agent or attacker does; the test suite tries to break it across many scenarios and records the result.
Recorded detail
[
{
"comparable": false,
"reason": "BENCHMARK_VERSION_MISMATCH_COMPARISON_INVALID"
},
[
"STALE_BENCHMARK_VERSION"
]
]Check it yourself
Related rules
- D140: a missing dimension is a regression
- D142: benchmarks detect selective reporting and partial metrics
Try CAIN-42 on your own agents
Create a free account and every new account starts with a 7-day trial of the full platform. Or try the sandbox first, with no account at all.
Create a free account → · Try the sandbox · See the whole ecosystem