Everyone has met the engineer who's sure they're brilliant and isn't, and the quietly excellent one who assumes everyone knows what they know. There's a famous name for the first half of that: the Dunning-Kruger effect. The idea, and its important caveats, matter a lot for how you evaluate technical skill, especially your own.
Key Takeaways
- Dunning & Kruger (1999) found that people with low skill often overestimate it, because judging skill requires the same competence (Kruger & Dunning).
- In their study, people in the 12th percentile rated themselves around the 62nd (Kruger & Dunning).
- The cause is metacognition: incompetence hides itself from the incompetent.
- The effect is real but overstated and contested, so hold it loosely, and apply it to yourself first.
The Original Finding
The 1999 paper by David Dunning and Justin Kruger has one of the best titles in psychology: "Unskilled and Unaware of It" (Kruger & Dunning). Across studies on logic, grammar, and humor, they found that the people who performed worst tended to most overestimate their ability. In one striking result, participants in the 12th percentile rated their skills as if they were in the 62nd. Their explanation is elegant: the very skill you need to do something well is the skill you need to judge whether you're doing it well. So the least competent are robbed of the ability to see their own incompetence, a failure of metacognition.
It has a flip side too. Genuine experts often underrate themselves, because they know how much they don't know and assume others share their knowledge.
Why It Matters in Engineering
Software is full of Dunning-Kruger territory, because so much of the skill is invisible. A junior can write code that "works" and feel like an expert, unaware of the security holes, the edge cases, the maintainability problems, and the better approaches they can't yet see, precisely because seeing them requires the experience they lack. Meanwhile the senior engineer, who knows all the ways it can go wrong, sounds less certain. Confidence and competence can point in opposite directions, which is exactly why interviewing on confidence is a trap.
| Low skill | High skill |
|---|---|
| "This is easy, I've got it" | "Here's what could go wrong" |
| Unaware of what they can't see | Aware of the gaps |
| High confidence | Calibrated, often cautious |
The Honest Counterpoint: The Effect Is Contested
Here's the part the pop-psychology version skips: the Dunning-Kruger effect is real but considerably overstated, and parts of it are contested. Later analyses found that most low-skill people estimate their ability reasonably well, in one reanalysis only about 16.5% of the bottom quarter significantly overestimated themselves, and some researchers argue a chunk of the original effect is a statistical artifact of how the data is plotted (the critique). So the strong claim, "the incompetent are always wildly overconfident," is too strong. The durable, useful version is milder: skill and self-assessment are only loosely correlated, low skill can blind people to their own gaps, and confidence is a bad proxy for competence. Hold the effect loosely, and be most suspicious of it in yourself.
A Concrete Version
Two candidates. One breezes through the interview projecting total certainty, dismisses edge cases as trivial, and has an answer for everything. The other pauses, says "it depends," names the tradeoffs, and admits what they'd need to check. The confident one feels like the stronger hire in the room. On the job, the "it depends" engineer is the one who catches the security hole and the race condition, because they can see the ways things break. Interviewing on confidence would have picked exactly the wrong person.
What This Means for Hiring
The practical lesson is to measure demonstrated skill rather than projected confidence, which is the whole reason for structured, work-sample-based evaluation over gut feel. A confident answer in a room is easy to fake and easy to be fooled by; how someone actually reasons through a real, ambiguous problem is not. That's the heart of structured interviews and why confidence is a poor signal, and it's what our vetting is built to surface. It applies to leaders too: the humility to assume you might be the overconfident one is itself a mark of skill. See available engineers.
Frequently Asked Questions
What is the Dunning-Kruger effect?
A cognitive bias, from a 1999 study by Dunning and Kruger, where people with low skill in an area tend to overestimate it, because judging competence requires the same competence they lack.
Is the Dunning-Kruger effect real?
Partly. The core observation, that skill and self-assessment are loosely correlated and low skill can hide its own gaps, holds up. But the strong popular version is overstated and contested, with some of the original effect argued to be a statistical artifact.
Why does it matter for engineering?
Because much of engineering skill is invisible, so confidence and competence often diverge. A junior can feel expert while missing security holes and edge cases, while a senior sounds less certain. Confidence is a poor proxy for skill.
How should hiring account for it?
Measure demonstrated skill through realistic work samples and structured evaluation, rather than trusting confident answers in a room. How someone reasons through an ambiguous problem beats how sure they sound.
The Bottom Line
The least skilled are often the most confident, because seeing your own gaps takes the very skill you're missing. The effect is real but overstated, so hold it loosely, and take the practical lesson: confidence is a bad proxy for competence. Measure demonstrated skill rather than certainty, in your hires and in yourself.
Roberto Espinoza is CEO of Ruzora, which helps US startups hire pre-vetted senior LATAM engineers in 72 hours. See available engineers.
