define dunning-kruger effect — let me try
tom asked me last april if i thought he should move in with someone he had been dating for six weeks. i gave him forty minutes of advice. tom is single. tom has been single since 2019. i was wearing a t-shirt that said ‘idiot’ on it, ironically, which is the only way i wear anything.
desk on a wednesday. carla is upstairs at the all-hands and will be there until eleven, give or take a coffee. ninety minutes, no surveillance.
the trouble with this kind of post is that the topic and the author are, regrettably, in the same photograph. let me define dunning kruger effect the way i’d define it for a friend at the corner of a bar, after one beer, before the food arrives. the friend, in this scenario, has the patience of a person who knows me. you, reading this, have been kind enough to grant me roughly the same patience. i appreciate it.
define dunning kruger effect: a pattern in which people who know little about a subject tend to be the most confident about it, while people who actually know the subject tend to hesitate, qualify, and quietly doubt themselves. it is the reason the loudest voice in a meeting is statistically the wrong one to follow. it is also, in this writer’s case, the reason this post exists.
CONFIDENCE. IS. NOT. EVIDENCE. OF. ANYTHING.
the formal definition, one sentence
the formal version, in one careful sentence, is this: the dunning kruger effect is the observed mismatch between perceived and actual competence, in which people of low ability tend to overestimate their ability and people of higher ability tend to underestimate theirs, because the skills required to perform a task well are, often, the same skills required to assess whether a task has been performed well at all. there. one sentence. i had to back up and re-punctuate it twice. that is also, i suspect, evidence of the thing the sentence describes. there is, by the way, a cinematic shorthand for the same dynamic — namely will ferrell’s 2004 film “anchorman” — but that version is louder than mine, and shorter, and has a band.
the version i’ve quoted above is a paraphrase of the original 1999 finding, which i did not read. i read about it in roughly four places, none of which i can now find with certainty. they all said roughly the same thing. they all had the same chart. i’m fairly sure i could find the chart again if i wanted to, in the literature i’m fairly sure exists, but i don’t want to. the point of the paraphrase is that the paraphrase is what the term has become. that’s where the conversation lives now. at the bar. in the office. on a wednesday. in places like this.
the personal definition, mine, longer
my own definition, drafted by a man who is, for the duration of this post, exhibit a, runs longer. the dunning kruger effect, in my version, is what happens when a person without enough information to be cautious is also, by the design of the situation, the person most likely to volunteer first. that’s the engine. that’s the loop. you don’t know enough to be careful, so you go. people who know more sit there listening, because they remember the last time they were sure and were wrong. you, who haven’t had that yet, take the floor.
stefan, for example, is the most confident wine drinker i have ever sat across from. he holds the glass up to the light. he says things like “this is a good year” and “you can taste the soil”. stefan has never visited the soil. stefan studied accounting. on tuesdays he is an expert about wine; on wednesdays, cycling; on thursdays, cryptocurrency. stefan is not lying. stefan is on what people, online, call peak mount stupid, with a glass of red and a strong opinion. he is also the most likable person at any dinner. these two facts are not unrelated.
i am no better than stefan. i am worse, possibly, because i write blog posts about him on company time.
the chart, briefly
there is a chart that everyone draws when they explain this. confidence on one axis, knowledge on the other. it goes up too fast at the start, crashes, and climbs back up slowly without ever returning to the early peak. the early peak is the dangerous one. the climb back is the honest one. most people you’ve ever worked with are on the peak. most people you’ve ever trusted are, in retrospect, on the climb. the difference shows up in how they answer questions. the peak says “easy”. the climb says “depends”.
i tried, once, to draw this chart on a napkin at a wedding. i over-explained. the napkin tore. that’s a different story. the chart, however, was not bad.
now, let me say this clearly.
the chart is, in the end, a confession. nobody draws this chart about themselves. they draw it about other people, in their head, while a colleague presents. the act of drawing it is itself a peak move. you are, in that moment, the most confident person in the room about who is overconfident. there is no exit from this hall of mirrors that doesn’t involve, at some point, shutting up and listening for six full seconds. i’m fairly sure of that. i did not look it up.
i rest my case.
why i don’t have it, a thorough defense
i’d like to use this section to defend myself, as a matter of public record, against the suggestion that i, the author of this post, am the very example i keep describing. i have several arguments.
first: i have, on this machine, in front of me, forty-seven open tabs. i have not closed them. each tab represents a thing i thought i needed to know, opened on a tuesday, abandoned by thursday. a man with the dunning kruger effect, i submit, would have closed the tabs the moment he felt he understood the topic. i still have them open. that’s evidence of doubt. that’s evidence of, possibly, intelligence. i’d like the record to reflect.
second: there is, on a pinned tab, the wip 2022 list, which i opened, in good faith, four years ago. the list has forty-six items. it has been, in functional terms, the same list since february of that year. a man on peak mount stupid would have closed the list and called the project finished. i have left the list open as a kind of monument, which is, i’d argue, a sign of self-awareness. i’d argue.
third: my hot take, which i have published in places i can no longer recall, is that showers over 4 minutes are theatre. i hold this take with full conviction. i would defend it in court. and yet — and this is the key move — i am willing to say, here, in writing, that i may be wrong. you cannot say that about everyone with a take. you cannot, certainly, say it about stefan. stefan would die for the soil he has not visited. i, in contrast, am willing, on the right kind of wednesday, to consider the possibility that water is not, fundamentally, theatre. that’s growth. that’s data.
fourth: chatgpt, which i use to triage my emails like a triage nurse who is also bad at triage, regularly tells me i’m wrong. i listen. sometimes. that, also, is evidence.
four arguments. case rested. i am not, on the balance, the example. the post-it on the desk says otherwise. the post-it is, at this point, decoration.
verdict, write it down
let me put it in one line — and you can post-it it to the screen of the most confident person in your office, when they step away.
the dunning kruger effect, defined cleanly, is the gap between what a person knows and what they think they know, when the person is not equipped to measure either side. that’s the whole sentence. you can write it on a post-it. i would suggest you do — and that you put it on the screen of the most confident person you know, when they are away from their desk. they will not understand the point. that, too, is evidence.
i’d like to add, in closing, one note. the cure for this is not more confidence — the world has plenty. the cure is the habit of saying “i’d want to look at the data” before saying anything else. small habit. unfashionable. it works. for a longer treatment of the same instinct, i covered a working method for becoming smarter, drafted on company time — same desk, different morning. the central move there is the central move here. slow down. let the chart move.
case closed. the manual is, somewhere, in the literature. the manual is, technically, optional.
carla wandered past the desk. screen swapped. no eye contact, no comment. clean exit.
that’s the definition. that’s the take. that’s ninety minutes that, on a careful reading, were not technically mine to spend.
yours stupidly,
idiot again
volunteer commentator on, definition-of-dunning-kruger department
P.S. stefan, last i heard, is now confident about beekeeping. stefan does not own bees. stefan is, in his way, the entire point of this post.
