dunning and kruger — what they didn’t say in 1999

mike at the corner bar summarized two psychologists in the time it takes to finish a beer. he had not read them. he had heard about them, possibly on a podcast, possibly from a guy at a bus stop. he was completely sure. a hot dog, by the way, is a sandwich, and that is also from mike.

desk, friday, 11:34am, screen angled away from the hallway. carla is on the third floor in a vendor walkthrough that is, by my count, on its second hour. that gives me the rest of the morning to put two surnames on a page in the right order, which is harder than it should be.

the phrase i’m trying to handle, with the gentleness mike already gave it, is dunning and kruger. people type both surnames into a search bar more often than you’d guess, in that order, like a question to a switchboard. they are looking for two men. they are also, secretly, looking for permission to call somebody else confidently dumb without doing the work of saying so. mike, on his stool, did the work of saying so. mike said it was monday. mike says everything is monday.

dunning and kruger: the two psychologists, david dunning and justin kruger, whose 1999 paper put a name on the chasm between confidence and competence. the bar version, delivered by mike on a friday, is shorter than the paper and ends up at roughly the same conclusion. mike has not read it.

TWO. SURNAMES. ONE. POLITE. INSULT.

that goes on the wall above the printer nobody uses. there is, naturally, a film for this exact dynamic — see the 2004 film “anchorman” with will ferrell, in which a man with a strong jaw and weak research becomes the loudest authority in his own newsroom by refusing to be corrected. that’s the visual i kept coming back to, on a stool, while mike was talking. that is, in fact, the entire short version of the longer walkthrough of the dunning-kruger effect i drafted last month, which has charts and slightly fewer beers.

dunning and kruger, the two men

so, the two men. david dunning was, and as far as i can tell still is, a psychologist with a serious chair somewhere serious. justin kruger, his then-graduate-student, was the one who, by tradition, did the lion’s share of the spreadsheet labor, which is what graduate students are statutorily required to do. they put their names on a paper in 1999, in that order — dunning first, kruger second — and the order has been frozen there for a quarter century. when people write the phrase, almost nobody flips it. that is, in its way, the most stable thing about the whole field.

mike’s version of the introduction was: “two guys, late nineties, one paper, a chart shaped like a hill.” mike was wrong about the chart, technically — it is shaped less like a hill and more like a ski slope into a small valley and then a slow climb up the other side — but mike was right about everything else, which is the thing about mike. mike runs a low error rate by saying very little. mike pays for what he says by mostly not saying it.

i, by contrast, am writing roughly sixteen hundred words about two surnames i did not personally meet. that imbalance is the whole subject of the post.

what mike thinks they got right

mike’s read on the paper, condensed past the point a peer reviewer could forgive, was this: the people who are worst at a thing are also the worst at noticing they are worst at it. mike said this between sips. mike said it like a man who had once run a small business with a payroll, although mike has not, and who therefore knew what bad work looked like from the receiving end.

mike then offered an example. mike’s example was a man two seats down, the previous monday, who had explained to a stranger, for nineteen straight minutes, how mortgages work. the man rents. the stranger paid for one of his beers. that, mike said, was the paper in motion. that was dunning. that was kruger. that was the slide deck of a man with no slides.

i said: “you know there’s a chart, right?” mike said: “i don’t need the chart.” he was correct. mike, who had not seen the chart, had still passed the test the chart was designed to administer. the chart, for the curious, is unpacked in the post i wrote about the famous chart, which is itself less serious than mike. that is also data.

what the paper actually says

the paper itself, which i will admit immediately i have not read in full and have looked at in the way a man looks at a menu in a language he half-knows, says the following in plainer english than mine: ask people to do a thing, ask them how well they think they did, and watch the bottom quartile rate themselves the most generously. the worst performers, in study after study mike will never read, were the most confident. the best performers, by contrast, hedged. they assumed everyone else had done about as well as they had. that, in a strange way, is the part that breaks me. competence is a quiet form of empathy.

i looked it up, briefly, in three places. one had a paywall. one had a broken image where the famous curve should have been. one was a podcast i listened to for eight minutes while pretending to organize the wip 2022 list, which has forty-six items and has not moved in just over three years. the podcast host, i am fairly sure, had the names in the wrong order, which is its own demonstration of the phenomenon under discussion. i did not correct him. neither, presumably, did anyone else.

for a self-test that does not require any of this — and that mike, on his stool, would respect — there is also the small test i drafted on a post-it, which has five questions and one ugly verdict. it is shorter than this section. it is also more honest.

what they did not anticipate, namely me

here is what dunning and kruger, in 1999, did not anticipate. they did not anticipate me. they did not anticipate a man at a standing desk that he, in fact, sits at, with forty-seven open tabs and a chatgpt window that screens his email and increasingly screens his own takes back to him in cleaner sentences than he came in with. they did not anticipate the recursion of a person looking up the phrase that describes him in order to perform the phrase that describes him.

i’d add, for completeness, that this is also the secret country bordered by the more general business of being a confident liar to oneself, which is the territory the liar essay covers — when a person believes their own draft hard enough that the draft becomes the room. dunning and kruger described one wing of that country. mike, on his stool, runs the customs office.

findings, the men are valid

the findings, then, after eighty minutes of writing and one cold coffee:

1. the men are valid. dunning and kruger, in that order, identified a pattern that the bar identifies every monday night without footnotes. the bar got there faster. the paper got there with a chart.
2. mike is valid. a man who has not filed a return since 2019 still managed to summarize two psychologists in fewer words than the average linkedin headline. the medium is the message. mike is the medium.
3. i am, on a careful reading, exhibit a. i wrote sixteen hundred words about a paper i skimmed, on company time. the chart predicted i would do this. i did this. the chart is, in this respect, also valid.

the seventh microwave is off the table this week. that, like the names of dunning and kruger, has been frozen in a particular order for a long time now, and i don’t expect either to move soon.

carla just walked back past the desk. she did not stop. the vendor walkthrough must have ended at noon, on schedule, which is the first scheduled thing this floor has done in weeks. i am closing the tab with the paper i did not read, and the tab with the podcast, and the tab with the search bar that started this. forty-four tabs left, by my count. the wip 2022 list is, of course, untouched.

↑