When a hiring loop gives one outside calibrating interviewer veto power, what tradeoffs does that create?

## What the veto is buying A weighted, blocking scorecard in the hands of an interviewer outside the hiring team is a deliberate transfer of authority away from the people closest to the need. That only makes sense if the people closest to the need have a predictable bias — and they do. An understaffed team judges candidates against the gap in front of it. The org needs candidates judged against a standard that outlives this quarter, this roadmap and this team. Nobody lowers a bar on purpose; it slips through reasonable exceptions, each defensible in isolation. Giving the standard a dedicated owner with no stake in the seat, and giving that owner the power to stop rather than merely to advise, is the difference between a stated standard and an enforced one. The secondary gain is a **record**. Because the role is expected to justify a block in writing, the organisation accumulates written reasoning about what its bar actually is — far more useful than a rubric nobody applies. ## What it costs **Concentrated judgement.** One person, one write-up, one taste. Where a panel averages out idiosyncrasy, a veto amplifies it. A calibrator with a private preference — a favoured way of framing tradeoffs, a discomfort with a certain background — becomes an unwritten standard nobody can see. **Latency and throughput.** Trained calibrators are a scarce, cross-cutting resource. Loops wait on their calendars, and in a hiring push the scarcity binds precisely when speed matters most. **Ownership.** A manager who cannot hire the person they want, twice, starts treating the loop as something done to them. Disengaged managers write thinner scorecards, prepare candidates less, and quietly route around the process. **Rubric leakage.** Any consistent standard eventually becomes a preparation target. Some of that is healthy; some of it converts the round into a performance test that rewards the well-coached. **Relocated bias, not removed bias.** A veto does not launder judgement. It moves it from many people to one, which can help if the one is trained and audited and hurts badly if they are not. ## What keeps the weight legitimate The useful question is not whether a veto is good but what makes a specific blocking write-up defensible. Four things do most of the work. 1. **Evidence, not impression.** The scorecard should quote what the candidate said or did, name the standard applied, and let a reader who was not present check the reasoning. An unreviewable impression should never be able to block. 2. **Training and rotation.** A pool large enough that calibrators shadow, are shadowed, and can be compared. A permanent single gatekeeper is a person, not a process. 3. **An escalation path.** A block that can never be revisited is a taste. A block that can be revisited with new evidence — an extra round, a different problem, another calibrator — is a standard. 4. **Divergence review.** Compare calibrators against one another. If two calibrators reach opposite conclusions on similar packets at any noticeable rate, the standard is personal rather than shared, and the fix is training rather than more veto power. ## An illustration of the failure In one invented but recognisable pattern, a platform organisation runs six-round loops with one calibrating slot. Over a hiring push, one calibrator blocks a noticeably higher share of candidates than their peers, and the blocks share a theme: the write-ups say the candidate did not show enough depth, but they quote very little of what the candidate actually said. Managers learn the pattern and start scheduling around that calibrator. The standard has not risen; it has fragmented, and the process has acquired a shadow routing rule nobody wrote down. The corrective is unglamorous. Require the citation of evidence, sample the write-ups, and treat a calibrator whose blocks cannot be reconstructed from their own notes as a training problem rather than a strict one. ## Judging the mechanism Speed is the easiest thing to measure and the wrong thing to optimise alone. A loop that closes fast while the bar drifts has traded a visible cost for an invisible one, and the invisible one is paid over years in performance management and rehiring. Conversely, a calibration role that survives only because nobody dares question it has stopped being a control and become an institution. The principal-level judgement is knowing which failure you are currently exposed to — drift or bottleneck — and being willing to say what you would measure before you extend the veto or take it away.