how to explain the dunning kruger effect origin, in six steps

mike was halfway through his fourth beer and his second wrong origin story for a famous psychology effect when i decided, against all reasonable judgment, that i would write the right one down the next morning. sarah wasn’t with us. sarah is somewhere else, doing the things she said she would do, which is, by all accounts, a lifestyle. i have a list of things i said i’d do. it is from 2022 and it has aged like a banana on a radiator.

so this morning. desk, tuesday, the rest of the q3 review still happening two floors above me. carla is up there. i have, give or take, the long stretch before lunch to do this properly, by which i mean to do it at all.

here’s the trouble with writing about the dunning kruger effect origin. the origin is older than i thought. it is not, as mike claimed at the bar, “some guys in the eighties.” it is not, as i once half-remembered, a piece of folk wisdom that became a study. it’s a paper. a real paper. with a year. and a title that, when you read it out loud, is funnier than it has any right to be.

desk, second coffee, no monitoring. the third drawer is open because i kicked it on the way in and now i can’t be bothered.

for the longer treatment of the chart and the philosophy, i have a fuller piece on the dunning kruger effect drafted at this same desk, which sets up the curve and the language and the part where i admit i’m probably on it. for what the words actually mean in a sentence, there’s a shorter explainer on the meaning of the dunning kruger effect. i wrote both before i’d ever read the original paper. that is, by my own admission, on-brand.

step one, the year was 1999

1999, in case you’ve forgotten, was a year. people had jobs. people had hair. people had cd binders. somewhere in the middle of all this, two men at a serious university looked at a smaller, earlier idea about a man who had robbed a bank with lemon juice on his face — true story, briefly — and asked themselves a smaller version of the question: how can a person be that confidently, comprehensively wrong?

they ran some studies. logic puzzles. grammar. humor. they had volunteers rate their own performance and then they compared the ratings to the actual scores. you can guess what happened. the worst performers thought they did well. the best performers thought they did okay. the gap between thought and reality was wider, by a comfortable margin, at the bottom of the chart than at the top. they wrote it up. they sent it to a journal. the journal printed it. that was the year. that was 1999.

i looked it up this morning, in a tab i’m not going to link, and i’m fairly sure there’s a study somewhere — possibly in a serious magazine — that says reading the original is more useful than reading the seventeen takes about the original. i have not, until today, tested that. the test was reading it. the result was a feeling.

step two, two men, a paper, a stage

the men in question are dunning and kruger, in that order on the paper, in the reverse order in popular speech, which is the small, mild injustice of being a co-author. dunning had the lab. kruger was the graduate student doing the work. they wrote it together. they put their names on it. they sent it out into a world that, twenty-some years later, would put their surnames on coffee mugs and meeting memes and small charts on the back of receipts.

here’s the part i find quietly funny. the paper is calmer than the legend. the paper does not say “everyone who disagrees with you is on peak mount stupid.” the paper does not say “this is why your boss is the way they are.” the paper is mostly tables and careful language. the legend has done all the loud talking. that’s how it always goes. the paper sits down. the meme stands up. the meme buys a microphone.

A. PAPER. IS. NOT. A. MEME. CALM. DOWN.

step three, sarah found the original at the corner bar last week

sarah, who reads things, mentioned in passing — not at the bar, somewhere else, on a phone call i picked up by accident — that she had read the actual paper. read it. in full. with footnotes. she said it was “shorter than people think.” that is, on the chart of human statements, one of the most generous lines a person can deliver about a piece of academic writing.

i, on the strength of sarah’s review, opened the paper. i did not read it. i scanned it. i scrolled. i scanned again. i said the word “tables” out loud. i closed it. i opened it again. i did, eventually, read most of the introduction and one paragraph in the middle, which i’m fairly sure is the paragraph the entire pop-culture version is built on. that paragraph is the whole show.

then i went to the corner bar that evening to debrief, where i made the mistake of trying to explain the dunning kruger effect origin to mike. mike is the patron saint of the corner bar. mike has not, by his own admission, filed a return since 2019. mike listened. mike has a way of listening that resembles, very closely, not listening, except the eyes track.

i let him have it. it was his fourth beer. that is also, more or less, what the paper says, but with worse grammar and better moral standing.

step four, the title, paraphrased

the title of the paper, which i will paraphrase rather than quote because i am not in the citation business, is a long sentence about being unskilled and unaware of it, and how this leads to inflated self-assessments. that’s the whole title, more or less. you can hear, in it, the dry humor of two academics who knew exactly what they had on their hands. that title is, on the spectrum of psychology paper titles, almost a stand-up bit. most psychology papers are titled like furniture. this one is titled like a roast.

i don’t have the title on a post-it. i probably should. i have, instead, a wip 2022 list, two tabs over, with forty-six items on it, none of which have moved since february. the list is, in its own quiet way, a small annual restaging of the original paper, with me as both authors and all the volunteers.

step five, the conclusion, paraphrased again

the conclusion of the 1999 paper, in plain bar english, is this: the people who are worst at a thing tend to be the most certain that they are good at it. the people who are best at a thing tend to underestimate themselves a little. training the bottom group makes the gap shrink — not by making them better immediately, but by making them aware. awareness is the lever. ignorance, of itself, is the lock.

i do think about this, at this same standing desk that i sit at, more often than i’d like. there’s a related post i wrote about a misspelling of the same effect that drifts at the topic from a different angle, and a small companion piece on another typo of the same name that ended up being mostly about why people search for things they can’t quite spell. neither piece tells you the year. this one does. that’s the upgrade.

step six, the verdict on the dunning kruger effect origin, older than i thought

the origin is older than i thought, but younger than mike thought, and the gap between his guess and mine and the actual year is, in itself, a tiny live demonstration of the very thing the paper is about. neither of us was right. neither of us hesitated. one of us had a fourth beer. one of us had a tab in the middle of a stack of forty-seven, none of which i can close because closing a tab is, in this house, an admission of something. reading on a kindle, mike pointed out at one point, is the same as reading, by which he meant: you can read this paper on your phone, you don’t need a building. that is the kind of thing mike says when he is, technically, right.

for a film-flavored sense of how the dynamic plays out at low altitude — a confident protagonist, a small data set, a long climb — see the 1999 film “election” with reese witherspoon, which is from the same year as the paper and is, in its own way, also a study in misjudged competence. the year is doing some quiet work in both cases.

my own conclusion, drafted from the desk while carla is two floors up explaining the q3 numbers to people who already saw them: read the paper, or don’t. either way, know the year. the year is the smallest fact that does the most work. mike thinks it’s the eighties. it is not the eighties. it is 1999. write that one down. you may, like me, be a small, careful liar about the dates of things you’ve never bothered to look up — i’m including myself in that, since the liar in question, before this morning, was me.

it is now 11:14am, the third drawer is still open, the wip 2022 list has, somehow, gained an item without my consent.

i will, at some point this afternoon, walk past the corner bar on the way home and see if mike is on the right beer to receive a correction. he will not be. he is, by my own quiet count, on the wrong beer for corrections roughly six evenings out of seven. that’s another study. someone should do it. it will not be me.

↑