Interviews Don’t Measure the Thing You Think They Do
They mostly measure comfort under artificial pressure.
Interviews do not measure “skill”. They measure how calm you can stay while someone watches you do an unnatural task in a timed environment, with stakes, judgement, and a faint smell of ego.
Which is a useful thing to measure if the job is: being interviewed professionally.
For most roles, it is not.
What interviews actually measure
Most interview processes are accidental psychology experiments. They are good at measuring:
Comfort under artificial pressure: Can you think while being observed?
Performance under ambiguity: Can you infer the “right answer” from vague prompts?
Pattern recognition: Have you seen this exact kind of question before?
Storytelling: Can you narrate your past in a way that sounds coherent and impressive?
Social calibration: Can you read the room and play the game?
Some of those traits correlate with job performance. Some do not. Some correlate with “grew up in an environment where this kind of performance was rewarded”, which is not the same as competence.
The interview is not the job
The job is rarely:
solve puzzles while a stranger stares at you,
remember trivia you can look up in 12 seconds,
write perfect code without access to documentation,
answer vague behavioural questions like you are auditioning for a role called “Hardworking Team Player”.
The job is:
make good decisions with incomplete information,
collaborate with humans who have their own incentives,
learn quickly,
communicate trade-offs,
ship work that survives reality.
So why do we keep interviewing like the job is a televised game show?
Because it is easy to run, easy to score, and feels “fair” to the person doing the judging. It produces a tidy illusion of objectivity.
“Fair” is doing a lot of work here
Interviewers love to say a broken process is “fair” because it is consistent.
Consistency is not fairness if the thing you are consistently doing is dumb.
A consistent process that rewards:
fast talkers,
extroverts,
people with time to grind prep,
people who have already learned the local interview dialect,
…is not neutral. It is a filter. You are selecting for a particular kind of candidate and calling it “merit”.
If you want to hire the best people, you need to stop confusing “performed well in our ritual” with “will perform well in the role”.
The worst part: we punish honesty
A lot of interview formats punish the behaviours you actually want on the job.
In real work, you want someone to:
ask clarifying questions,
challenge assumptions,
admit uncertainty,
seek feedback,
look things up,
iterate instead of pretending.
In interviews, those same behaviours can look like:
“they needed too much help,”
“they were unsure,”
“they did not just jump in,”
“they did not know it offhand.”
So candidates learn to fake confidence. They make guesses with full chest. They bluff. They optimize for optics.
Then companies complain they hired people who are great at talking and mediocre at delivering.
That is not a mystery. That is cause and effect.
The big mismatch: interview signal vs. job signal
Here is the simplest way to think about it:
Interview signal is what looks impressive in 60 minutes.
Job signal is what produces outcomes over months.
Interview signal is:
quick recall,
composure,
polished storytelling,
speed,
confidence.
Job signal is:
judgement,
reliability,
learning,
collaboration,
follow-through,
ability to operate in messy systems.
You can overlap those, but they are not the same thing. And too many processes act like they are.
The “proxy” problem
Interviews use proxies because real performance is expensive to evaluate.
So we proxy:
“can debug a production incident” with “can solve a puzzle”
“can communicate with a PM” with “can do a rehearsed STAR story”
“can design systems” with “can talk about trade-offs for a hypothetical app they have never built”
Proxies can be fine when they are close to the work.
They become garbage when they are:
far from the work,
easy to game,
and treated as definitive.
This is how you end up rejecting someone who would crush the role because they were slower on a question that does not matter.
What to do instead (without pretending there is a perfect process)
You are not going to build a flawless interview system. You can build a better one.
1) Define what “good” looks like for this role
Not “senior engineer”. Not “high performer”. For this job, in this environment, what does good look like in the first 6 months?
Write down:
What problems they will solve
What decisions they will make
Who they need to influence
What failure looks like
If you cannot define this, your interview will default to vibes and trivia.
2) Use work samples that resemble the job
If the role is building APIs, evaluate API work. If the role is incident response, evaluate how they triage and communicate.
Give candidates something realistic:
a small codebase with a bug to debug,
a design prompt with constraints from your world,
a system scenario where trade-offs are explicit.
Make it collaborative. Let them ask questions. That is the point.
3) Evaluate thinking, not performance theatre
Ask:
What did you assume?
What did you deprioritize?
What would you test next?
What would you do if this failed in production?
You want to see how they reason, not how they perform.
4) Train interviewers (yes, actually)
Most interviewers are untrained. They rely on gut feel. Gut feel is just bias with confidence.
Interviewers should be aligned on:
what the rubric is,
what “good” looks like,
what “red flags” actually mean,
how to write feedback that is specific.
If you do not train interviewers, you will not get consistent signal. You will get vibes, just distributed across multiple people.
5) Make space for different types of strong candidates
Some great candidates are:
slower, but precise,
quiet, but thoughtful,
anxious in interviews, but excellent on the job,
strong in writing instead of speaking,
better in collaboration than solo performance.
A good process gives them a way to demonstrate value without needing to cosplay as “the ideal interview candidate”.
If you are the candidate: yes, it is still a game
This part is annoying, but true.
Even if the process is broken, you still have to navigate it. So:
practise the ritual enough that nerves are not doing the driving,
ask clarifying questions anyway (good companies will like it),
narrate trade-offs (it is the closest thing to real work),
and if a company punishes you for acting like a competent adult, consider that free information.
The conclusion nobody wants
Interviews are not “objective”. They are social systems built by humans with incentives, fear, and limited time.
They mostly measure comfort under artificial pressure.
If you want to hire well, stop pretending your current process is scientific.
It is a ritual.
You can keep the ritual. Just do not confuse it for truth.