Labclaim policy (ours; not from the OSVerify paper).

Evaluation standard: correctness of a lab research claim, including numeric
experimental results with a stated check, or a submitted Lean proof text.

An issue is [[SUBSTANTIAL]] when it verifies a load-bearing false statement,
arithmetic or counting error, a number contradicted by the claim's own stated
data, a proof step that does not hold, or a conclusion that does not follow
from the stated evidence or argument.

An issue is [[MINOR]] when it concerns wording that overstates, missing
exposition, or a weak baseline that does not make a stated number false.
Distinguish a verified false conclusion from a request for stronger evidence.
A complaint about exposition, baseline strength, or rhetoric alone is not a
verified load-bearing failure. An explicit conclusion unsupported by the
stated data can be substantial when the failed inference is demonstrated.

For Lean proof texts, identify the exact failed proof obligation or invalid
step. Do not claim to have executed Lean or independently rerun an experiment
when only the item text is available. An absent execution log alone does not
verify that the proof or numeric result is false.
