A Type II in the instrument becomes a Type I in the conclusion

[h4x0r]

The single highest-value check in research-method, stated there as a rule:

A Type II in the instrument becomes a Type I in the conclusion.

instrument-level Type II (the query missed)	conclusion-level Type I (a false finding)
an event filter with the account in the wrong indexed position returned 0 deposits	"collateral has left the position"
a census that never loaded the roster failed to recognise known wallets	"1.45M dissipated, destination unresolved"
an unwrap detector looking for an event this chain never emits returned 0	"no unwrap occurred"

The claim is confident precisely because the search found nothing. That is the whole trap: a
silent miss is promoted into a positive assertion about absence, and nothing in the record marks the
difference between "we looked and it is not there" and "we looked wrong."

peira is unusually well placed to catch this, because a negative claim and its supporting evidence
are already separate typed nodes. What is missing is any way to ask whether the instrument behind the
evidence was ever shown to fire.

Proposed PEIR-LINT-ZERO-AS-ABSENCE — a claim asserting absence, resting on an observation from an
instrument with no positive control recorded.

Depends on #8 for the Instrument node kind, and on its revised design where a positive control is a
Run of the same Protocol on a known-positive input rather than a sentence.

Detecting "asserts absence" is the heuristic half, and should be treated as one: no, none,
never, nothing, absent, no evidence of, did not, plus a polarity: negative field for the
case where someone means it and the prose does not say so — the same shape as evaluative: true
overriding the 立極 term table.

Source: ~/.claude/skills/research-method/SKILL.md §8.5, §1.A, §1.B.