Keep candidate claims as evidence, not as facts

A tech stack the candidate says they ran. A metric they say they moved. A project they say they led. A reference who says the candidate was excellent. Each of these is an assertion until someone checks it, and the strength of a hire depends on how many of the load-bearing ones were actually verified.

The failure is not usually a lie. It is drift. A skill inferred from a job title becomes a skill the candidate has. A result the candidate described becomes a result the candidate delivered. By the debrief, the panel is scoring facts that were never established. Inspector exists to stop that drift by keeping a claimed accomplishment separate from a checked one.

Not every claim needs verification. The ones that do are the claims the offer rests on. If the main reason to hire someone is a platform they say they rebuilt, that claim is load-bearing and deserves a reference or a work sample. If the reason is a skill you watched them demonstrate, the evidence is already in hand.

Argumont helps the panel identify which claims are carrying the decision. Champion makes the strongest case for the candidate. Challenger attacks the reasoning that the case depends on, usually a result credited to one person that belonged to a team, or a seniority that describes a title rather than the work. The Judge should not treat a compelling story as a demonstrated capability.

Remote interviews have made it easier to produce fluent answers with real-time assistance. A candidate can sound authoritative on a topic they cannot actually work in. The reliable defense is not a tool that claims to detect a fake, which is easy to evade and easy to get wrong. It is a process that refuses to promote an unverified answer into a reason for the offer.

Lean on evidence the assistance cannot manufacture: a work sample the candidate walks through and modifies live, a specific decision they made and the tradeoff they rejected, a reference who managed the actual project. Scout looks for corroboration that exists independently of the candidate's own account. A verbal answer that cannot be probed or reproduced stays labeled as an assertion, not a checked skill.

A reference is evidence only to the extent it is independent and specific. A referee who managed the candidate on the project in question, and who describes concrete behavior, is strong evidence. A referee the candidate selected who repeats the candidate's own summary is weak evidence, however warm.

Keep the reference's relationship, recency, and firsthand knowledge visible next to what they said. Inspector will not treat a glowing but generic reference as confirmation of a specific claim. If a load-bearing claim survives only on the candidate's own telling and a reference who was not there, that is an open question, not a settled fact.

A rating without provenance is hard to defend later. For each material claim, record whether it was demonstrated, corroborated by an independent source, or asserted and unverified. That status belongs in the record next to the criterion it supports, so a later reviewer can see what the offer actually rested on.

This is also what makes a hire defensible. A decision where the reasons are specific, job-related, checked, and applied the same way to every finalist survives scrutiny. A decision built on impressions and unverified claims does not, whether the challenge comes from a hiring committee, a disappointed candidate, or a manager asking why the person is not working out.

Evidence often arrives after the interview: a reference call lands, a work sample is reviewed, a technical screen comes back. When it contradicts the case for a candidate, the honest move is to revisit the comparison, not to protect the earlier preference.

Use Challenge the Verdict to run the comparison again with the corrected evidence explicit. The Passport preserves both the earlier reasoning and the revised one, so the panel can explain why the leading candidate changed and what specifically changed it.