The most confident person in the interview is often the one who knows the least. I did not believe that for years. Early on, when I first started sitting on the hiring side of the table, I rewarded fluency. The candidate who answered fast, named every tool, and never once paused felt like the safe bet. It took a couple of hires that did not survive contact with the real work to make me ask what I had actually been measuring.
There is a name for what was fooling me. In 1999, two psychologists at Cornell, David Dunning and Justin Kruger, ran a series of experiments and found that the people who scored lowest on a test were the most confident about how well they had done. The less someone knew, the less equipped they were to recognise that they did not know it. Skill and the ability to judge that skill turn out to be close to the same skill, so when one is missing, the other goes blind with it.
The effect runs the other way too, more softly. In that same study, the strongest performers slightly underestimated where they ranked. Once something has become easy for you, it is genuinely hard to believe it is hard for anyone else. So the expert hedges and the novice declares, and an interview, of all things, is built to reward the one who declares.
Confidence can be performed, and a vocabulary of buzzwords can be assembled the night before. Competence is slower and quieter, and almost impossible to fake under a real question, which is the same thing that makes it easy to miss in forty-five minutes. At some point I stopped trying to measure what a person knew and started trying to measure how they knew it. I rebuilt the conversation around three things a CV can claim but only a real exchange can reveal: the depth of the skill, the mindset underneath it, and whether the person makes a team better or quietly worse.
Depth Does Not Announce Itself
The first thing I look for is depth, and depth almost never arrives as a headline. Anyone can say they know microservices, Kubernetes, the cloud. The words cost nothing. So I do not ask whether they know them.
What I am listening for is not the happy path. It is the texture of having been there: the thing they tried first that did not work, the layer where the real problem turned out to live, the moment the logs said one thing and the truth was another. I have spent twenty years watching systems fail in ways no documentation predicted, in payments and in search and in retail, and when someone has genuinely owned an incident from the first alert to the post-mortem, you can hear it in how they tell it. They slow down at the hard part instead of speeding up. They remember which detail mattered. A person who only read about the topic does the opposite, and the gap is unmistakable once you are listening for it.
The Mindset You Cannot Rehearse
Skill without ownership is a liability with good posture. So the second thing I want to see is what a person does when something breaks and no one would blame them for walking past it.
Nobody answers that one with a slogan. Either there is a real story, with a real cost and a real choice inside it, or there is not. The engineer who took responsibility for a mess they did not make is telling you how they will behave at three in the morning on a problem with no obvious owner. That is the line between someone who completes tasks and someone who owns outcomes, and it is most of what I mean by seniority. I keep coming back to a quiet belief that there is always a way through, given the right skillset and the right mindset. This is the question that shows me whether a person actually lives that or has only read the poster.
Disagreement Without Wreckage
The third thing is the hardest to test and the most expensive to get wrong. A brilliant engineer who corrodes the people around them is not a net positive, however good the code. And everyone says they are a team player; the phrase has been worn smooth of meaning. So I do not ask it.
What I am watching for is whether they can hold two things at once: a strong opinion about the right answer, and genuine care for the person on the other side of it. The toxic version wins the argument and loses the room. The passive version keeps the peace and lets a bad decision through. The people I want did neither; they argued hard for the better idea and left the relationship intact, sometimes stronger for it. I learned the worth of that the slow way, by once being the kind of lead whose team agreed with him out of fear rather than conviction. A room that cannot safely disagree never surfaces its best ideas, and surfacing the best idea is the whole reason to gather a team in the first place.
What True Potential Sounds Like
Put those three conversations together and a pattern appears, the same one Dunning and Kruger found from the other direction. People with real potential do not tell you everything is simple. They acknowledge complexity, because they have actually met it. They talk about their mistakes without flinching, because the mistakes are where most of what they know came from. Their confidence is the kind you earn by having been wrong and recovering, which is a very different material from the confidence of never having been tested.
That combination is what I mean when I say I am hiring for IQ, EQ, and SQ at the same time, and not only the first one. The raw skill is the part the interview was always good at testing. Ownership, and the ability to disagree without leaving damage, are the parts that decide whether a strong engineer raises the level of everyone around them or slowly lowers it. The same trap waits on a team where everyone already has fifteen years: the surest voice is not always the one worth following. The people I look hardest for do not just fill the role. They make the team better than the sum of the names on it, and they tend to do it with humility rather than noise.