Mental models for hiring

The mental models

• Signal vs noise

  A résumé and a 45-minute interview are mostly noise; the few real signals — a work sample, a reference who will speak candidly, a genuine track record — predict far more than charisma. Design the process to amplify signal and discount polish.

• Base-rate neglect

  A candidate who interviews brilliantly still faces the base rate: most hires into this role work out about as often as the last ten did. Weight the base rate of the role and the source, not just the vivid impression in the room.

• Regression to the mean

  The candidate who dazzled in one session will, on average, be more ordinary on the job — a peak performance regresses. Don't extrapolate a single great (or terrible) interview into a whole career.

• Expected value

  A hire is a probabilistic bet: weigh the range of outcomes — great fit, mediocre, costly miss — not the best case you're hoping for. A structured process raises the odds; a rushed one just raises the variance.

• Second-order thinking

  The first-order question is 'can they do the job?'; the second-order one is 'what does this hire do to the team, the bar, and who applies next?' A brilliant jerk clears the first test and fails the second.

• Opportunity cost

  An open role feels expensive, tempting a mediocre hire to stop the bleeding — but the true cost of the wrong person (ramp, backfill, the damage done) usually dwarfs the cost of waiting for the right one.

• The Peter principle

  Promoting your best individual contributor into management can lose you a great engineer and gain a poor manager — the skills don't transfer. Hire and promote for the job being done, not as a reward for the last one.

• Dunbar's number

  Past ~150 people the informal trust that let you hire on gut and vibe stops scaling — you need structured criteria and process. Match the rigour of your hiring to the size the org has actually reached.

• Goodhart's law

  The moment a hiring metric — years of experience, a degree, a coding-test score — becomes the target, candidates optimise for it and it stops measuring what you cared about. Use metrics as evidence, never as the gate.

Biases that trip up hiring

• Halo effect

  One impressive trait — a prestigious employer, a confident manner, a shared alma mater — casts a warm glow over everything else, so a strong first impression quietly inflates every later judgement of the same candidate.

• Confirmation bias

  Once you like a candidate the interview becomes a hunt for reasons to hire them — softer questions, red flags explained away — instead of a genuine test that could change your mind.

• In-group bias

  You instinctively trust the candidate who feels like 'one of us' — same background, school, or manner — mistaking familiarity for merit and quietly narrowing the team to a single type of person.

• Fundamental attribution error

  You credit a candidate's wins to their character and their setbacks to bad luck (or the reverse for one you've cooled on), when the situation — a great team, a lucky market — often did the heavy lifting.

• Anchoring bias

  The first strong candidate, or a stated salary number, sets an anchor every later candidate is judged against — so the order you interview in, not just the merit, shapes who looks best.

• Overconfidence effect

  Interviewers vastly overrate how well a short conversation predicts on-the-job performance; the confidence that 'I can just tell' is exactly where the most expensive hiring mistakes come from.

• Contrast effect

  A mediocre candidate looks like a star right after a weak one, and a strong one looks ordinary right after someone brilliant — the sequence distorts the score. Judge each against the bar, not the last person.

• Survivorship bias

  You only ever see how the people you hired worked out — never the strong candidates you rejected who would have thrived — so your 'proven' instincts are trained on a censored sample.

Mental models for other work