the dunning-kruger effect, explained by someone who definitely doesn’t have it

stefan poured me a glass of his own wine last summer and i told him, with a straight face, that the finish was ‘a little flat.’ he owns the vineyard. he has owned it for eleven years. he smiled the way you smile at a dog that has learned a small trick. that was the moment i should have known.

desk, second coffee, no monitoring. carla is two floors up at the quarterly review and won’t resurface until lunch unless the agenda runs short, which it never does.

here is the trouble. a person who has the dunning kruger effect is, by the design of the thing itself, the last person who would notice they have it. that’s the joke. the only people qualified to write about the dunning kruger effect are the ones smart enough not to. the rest of us, who feel qualified, are exhibit a. i am exhibit a. i’m going to keep going.

dunning kruger effect: the gap between how good a person thinks they are at a thing and how good they actually are, where the people least equipped at the thing tend to feel the most certain about it, and the people most equipped tend to hesitate. it explains a lot about meetings, online comments, and the man at the bar with the second beer who is about to teach you about taxes. it does not, on a careful reading of my own behavior, apply to me. probably.

DUNNING. KRUGER. IS. NOT. A. LAW. FIRM.

for several years i thought the term was a high-end law firm specializing in something dignified, like estate planning. turns out it is named after two men who, a while back, ran a study about how confident people are about things they’re terrible at. there is a film tradition for this exact dynamic — see the 2004 film “anchorman” with will ferrell, in which a man with no qualifications becomes the smartest person in his own newsroom by refusing to hear anyone else. that’s the visual. that’s also, i suspect, the auditory.

what the dunning kruger effect actually is, by someone who pretended to skim it

the dunning kruger effect, in plain bar english, is what happens when not knowing something protects you from knowing you don’t know it. you need a certain amount of skill to even recognize what skill looks like. below that threshold, you can’t see the gap between what you know and what’s there. you see a flat, friendly horizon. the horizon, you assume, is the whole map.

this is why the most confident person in any meeting is, statistically, the one to worry about. that person isn’t lying. they have an incomplete map and are giving you the tour with the energy of a man who lives there. it is, in a strange way, beautiful. it is also why your project is two months behind.

i’m aware of how that sounds. someone with the dunning kruger effect would, in a post like this, sound exactly like this. there is no pose i can strike that gets me out of it. the only safe move would be to stop typing. i’m not going to.

the chart, let me walk you through it

there is a famous chart, which i have looked at on screens of varying brightness over the years. on the x-axis is your actual knowledge of a topic. on the y-axis is your confidence in your knowledge of that topic. the line, if you draw it honestly, does not go up in a straight diagonal the way self-help books would have you believe. the line spikes early, then crashes, then climbs back up slowly, and never really gets as high as the early spike, which is, frankly, a depressing piece of geometry.

the early spike is sometimes called peak mount stupid. i didn’t make that up. someone, somewhere, has given the climb that name. the language is wonderful. you can be standing on it right now and the only sign would be that you’d feel pretty good about your view.

then comes the crash. the crash, sometimes called the valley of despair, is what happens when reality finally walks past your desk holding a binder. you realize the topic is bigger than you thought. you realize the people who actually know it are quiet for a reason. you sit, briefly, in a kind of polite shame. then you climb, slowly, and never quite catch up to your earlier confidence — which is fine. earlier you was wrong. later you is sane. that’s the trade.

examples of dunning kruger in the wild, not me, others

i’ll give you three. these are real, in the relevant sense. the names are not.

example one. a man at the corner of the bar i go to, on his third beer, explained to me how the tax system works. he was confident. he was warm. he had a theory. mike, who has not, by his own admission, filed a return since 2019, listened politely and said nothing. mike, in matters of taxes, has earned the right to silence. the man with the third beer had not. afterwards, walking to the door, i did manage one exchange with mike:

that’s peak mount stupid. the bar was the slide deck. we all nodded.

example two. a colleague — not carla, another one, in another department — once told me, in the elevator, that the entire concept of project management is a “vibe”. this person has never managed a project. they were on the chart. they were standing on the peak with a clean lanyard.

example three. i once watched a video of a man explaining that audiobooks were not really reading, a conclusion he reached by listening to a podcast. that’s the third example. the structure of the example is the example. books on tape are cheating is, for the record, a take i happen to hold — but i hold it as a confession, not as a verdict.

why smart people sometimes look stupid

the other half of the chart is the impostor half. genuinely capable people often underestimate themselves. this is not modesty. modesty is a costume. when you actually understand a topic, you can see how much of it you don’t know. that’s not pessimism. that’s an honest map.

so the smart person, in the meeting, hesitates. they say things like “it depends” and “i’d want to look at the data”. meanwhile, the man on the peak says “easy. three steps. i could do it in a weekend.” which voice gets the project? you know which voice gets the project. that’s why the project is two months behind.

why people who think they’re smart usually aren’t

this is the section where my own face is most clearly in the photograph. people who feel smart, in a stable, low-grade way, are usually running on small wins in narrow domains. they were good at something at fourteen. they got a compliment in college. they once won an argument at a dinner that mattered to them. on the strength of those data points, they have built a personality. the personality is “i am the smart one”. it is portable. it goes to every meeting.

the trouble is the personality stops updating. i wrote a separate post about a working method for being smarter, drafted at this same desk, and the method was, basically, slow down and notice things — the opposite of running on a fourteen-year-old’s win. the people who think they’re smart usually aren’t, because at some point they stopped letting the chart move. they parked. i, for the record, have parked many times.

self-test, am i dunning kruger

i wrote a small test on a post-it, stuck to the corner of this standing desk that i, in fact, sit at. the post-it has come unstuck twice. it is not a real test. it is a thing you ask yourself in a tone you’d normally use on someone you were trying to be honest with.

1. have you, in the last week, said “it’s not that complicated” about something you don’t do for a living?
2. have you, in the last month, given advice on a topic to someone who actually does the topic?
3. do you have an opinion about modern art you’d defend in front of a person who works at a museum?
4. do you find that meetings would go better if everyone just listened to you?
5. is there a wip 2022 list, or a folder named “ideas”, on your machine, that you have not opened in eleven months and yet still consider evidence of your potential?

if you score three or more, you are not, statistically, the exception you think you are. i scored four. the wip 2022 list, on this very machine, is open in another tab. forty-six items, none moved since february. that’s not data. that’s a flag in the ground.

what to do if you suspect you have it, you don’t, that’s the trick

the cleanest test, offered here by a man who has failed it, is to admit out loud that you might be wrong about a specific thing on a specific day. not “in general”. on monday, about taxes, you might be wrong. saying it out loud, to a person who does not have to forgive you, is the test. you either say it or you don’t.

chatgpt, by the way, is the tool i use to draft and screen more emails than i’d care to count. when you ask the machine for help on a topic you don’t understand, the machine sounds confident. it produces clean sentences. it gives you a map. then a person who actually knows the topic looks at it and tells you, gently, that the map is to a different city. the machine doesn’t know it’s wrong. you, holding the printed map at the airport, don’t know either. that’s dunning kruger effect in three actors. you, the machine, the city. the city wins.

working method, if you want one: (a) name a thing you said this week with high confidence; (b) ask one person who does that thing; (c) listen for six seconds before responding. it’s small. it works. it has, on three occasions, made me briefly less stupid.

verdict, i rest my case

carla just walked past my desk. she did not stop. that means the audit prep is still going. either way, this stays open in this tab, and the tabs above it, which now total forty-seven on the dot.

the wip 2022 list — a calendar choice that has aged poorly — is in the third tab from the left. one of the items on it is, i kid you not, “read about dunning kruger”. i am, in this very moment, closing that item. that may be the most productive thing i’ve done this quarter. it may also be evidence of the very effect we just discussed. i’ll leave the ambiguity on the table.

↑