Sample frameDay 263·Hiring · Select

Morning. You've never met a candidate you couldn't read. That's precisely the problem.

Hiring.

The Interview Illusion

5 min read·Apply by lunch

The question

Think about the last person you interviewed and turned down — can you name the specific thing they said that changed your mind, or was it that the conversation never quite lifted, and if it's the second, what exactly did you learn that morning about the candidate rather than about the room?

The idea

Jason Dana, who teaches at the Yale School of Management, ran a set of experiments with the Carnegie Mellon psychologist Robyn Dawes and Nathanial Peterson. Undergraduates interviewed other students and then predicted their grade average for the coming semester. Some interviews were ordinary. In others the person being interviewed had been told to answer at random, deciding each yes or no by a rule about the letters in the question rather than by anything true about themselves. Nobody was warned, nobody noticed, and the interviewers came out reporting they'd got a sense of the person; only after the questionnaire were they told that half the pool had been random and asked to guess which they'd sat through. The arithmetic is the part that hurts. Predictions built on a student's prior grades alone tracked the real result at .65. Predictions made after an interview managed .31. The conversation didn't merely fail to add — it diluted what was already there. The paper ran in Judgment and Decision Making in 2013, and Scott Highhouse of Bowling Green State University had explained five years earlier why a finding like that changes almost nobody's process: a method that removes your read on people feels like a verdict on your judgement. Day 70 named the mechanism. An interview is the ladder of inference at its most expensive — whatever a candidate says, you'll build it into a story, and coherence feels exactly like insight.

The framework

01
Put a number on the file first.
Day 62 gave you the instrument — a prediction written down before the result arrives. Point it at a candidate. Read the file — the scorecard, the application, whatever evidence of real work you have — and write three things down before the interview is even booked. A number out of ten against each scorecard outcome. The single fact about this person you most expect to turn out to be true. And the one thing you'd have to see or hear that would change your mind. That third line does the work, because it's the only part of the prediction an hour of conversation can honestly test. The number is a calibration check on you rather than a score for the file — it never enters the total the offer eventually gets decided on. Dana's result is frightening only if the interview is your first estimate.
02
An interview allowed to overturn everything will do exactly that.
Decide in advance which of those pre-interview numbers a conversation may legitimately move, and by how much. Specific evidence against a named outcome should move it — they claimed to have run the migration and can't describe a single decision they made. Warmth, fluency, shared references and the sense that they got you shouldn't, because those are precisely what a random-answer interview also produces. Write the permitted movement down before you sit down, in points, per outcome. Without that number the hour will overturn the file, and it'll do it inside the first ninety seconds.
03
Two directions, the same hour.
Day 72 gave you the move — build the strongest case against your own conclusion, out loud, before the money moves. Point it at your notes. Straight after the interview, write the two sentences summing up what you learned about this person, then write two equally convincing sentences arguing the opposite impression from the very same hour. If the second pair comes easily, and after an unstructured conversation it nearly always does, you didn't learn something about them; you built something around them. Three minutes, and it kills more bad hires than any clever question you could have asked.
04
Give the conversation a job it's genuinely good at.
None of this means cancelling the interview. It means demoting it from measurement to the things it does well. A conversation is how a good candidate decides about you, how the awkward specifics of the role get explained, and how you find out what they still need to know before they'd commit. Those are real jobs and they deserve the hour. What they don't deserve is a score that outranks everything else in the file, or the sentence that quietly ends most hiring debates: I just had a really good feeling about them.

Members only · 4 principles + template + AI mentor

364 more frameworks are waiting.

The Interview Illusion is Day 1 of 365. One framework every morning for a year — across strategy, sales, negotiation, leadership, and more.

$1 /day

Billed $365/yr · cancel any time

365 frameworks

12 topics · templates · AI mentor

Reserve your place — $365/yr →

One payment a year · the whole library · no auto-renewal.

More Hiring frameworks

All 19 hiring frameworks →