dunning kruger effect explained, i looked into it
dunning kruger effect explained, i looked into it
mike at the corner bar was not present, which is rare and which i felt in my chest. the 47 tabs were present. they are always present. i was, at the standing desk, comparing two explanations of a famous psychology effect, one simple and one complex. books on tape are cheating. unrelated. but true.
it is 9:08am on a tuesday. carla is upstairs in a training session that has snacks, which is how you can tell it will run long. i have, give or take, the rest of the morning before someone wants something from me, and i have decided to spend it doing what passes for research in my life, which is opening more tabs.
the prompt for all this is the office_third_floor breakroom whiteboard, where someone wrote the chart i can’t unsee — that little curve that goes up, then crashes, then climbs slowly back. underneath, in dry-erase, somebody had added: “are we here?” with an arrow pointing at the peak. the arrow was correct. the arrow is always correct.
dunning kruger effect explained, simple vs complex side by side
i pulled up the main writeup on the effect itself, which is the long one, the one i’d send a stranger if a stranger asked, and then i pulled up the academic-flavored version, the one with footnotes and small graphs, and i put them side by side and made a table. i love a table. a table is a way to pretend you are a serious person. ChatGPT screens my email and i screen my own confusion with tables. fair trade.
| stage | simple version | complex version |
|---|---|---|
| 1. early days | you know a little. you feel like a king. | low task competence correlates with inflated self-assessment, owing to a missing meta-skill. |
| 2. the cliff | you learn more. you feel like a fraud. | competence improves; confidence corrects sharply downward as feedback enters. |
| 3. the crawl | you learn a lot. you feel like a person. | calibration tightens; confidence climbs back, but slower and with caveats. |
| 4. the plateau | you know enough. you stop talking at parties. | expert calibration; reduced verbal display, increased domain humility. |
both columns are saying the same thing. one says it in eight words and one says it in eighty. the simple one fits in a text message. the complex one fits in a paper from the late nineties that the two researchers wrote and that everyone has either read or pretended to read. i am, as you might guess, in the second group.
mike’s explanation, between the two
mike is, normally, the man you ask about this kind of thing. mike at the corner bar has, for years, run a kind of unlicensed evening seminar on human behavior, the curriculum of which is whatever happened that day at the office park across the street. mike has not filed his taxes since 2019. mike believes this disqualifies him from no opinion. on this point i defer to mike entirely.
mike was not at the corner last night, which is the cold open i opened with, and the reason i’m doing the explanation myself this morning instead of borrowing it from him. i texted him. he sent back the word “doctor” and then a thumbs up, which i read as routine, not alarming, and which i still find vaguely alarming. anyway, mike.
here is mike’s version of the effect, reconstructed from the last six times i asked him about something similar over a beer. mike says: the brain has two settings, “i got this” and “oh god what.” most people live in the first one, which is why parking lots exist. the trick is to spend a little time, on purpose, in the second one. that is the entire intervention. mike calls it “the oh god what tax.” mike taxes himself this way often. he has not, as established, paid any other tax since 2019.
mike’s version is between the simple and the complex. it is the bar version. it is the one you can repeat to a friend without sounding like you read a book. i prefer it slightly to both other versions, but i would not put it in a chart.
why simple is enough most days
most days, the simple version does the work. you are at a meeting. someone says something with great confidence. you note that they have, by the look of it, just learned the term they are using. you keep this to yourself. you nod. you go back to your standing desk where you sit. you do not need a footnote for this. you need pattern recognition and a little compassion. the simple version covers it.
i think about “The Office” here, the conference room scenes specifically, where someone is explaining something to someone else and a third person in the back is doing the face. that face is the simple version of the dunning kruger effect explained. it lives in the face. you don’t need the chart. the face is the chart.
also, and i say this gently, the simple version is what you can give to a friend who is doing the thing right now. you cannot, in a kind tone, hand a friend a chart. you can hand them a sentence. a sentence with one comma in it. that is, frankly, the maximum amount of correction a friendship can absorb in one sitting.
why complex is enough most other days
but. and this is where i surprise myself, halfway through writing this and on what i am told is my third coffee. some days, the complex version is the one you need. specifically: the days when you, personally, are the one in the early stage. the days when you are the one with the eight-word opinion. on those days the simple version is too short. on those days you need the long one with the footnotes, because the long one slows you down enough to think.
this is the trick. the simple version is for diagnosing other people. the complex version is for diagnosing yourself. you need the friction of more sentences, more caveats, more “see also.” you need to be bored back into humility. it is, in its way, a self-administered oh god what tax, to use mike’s term, which i am now using because i find it, on reflection, accurate.
here’s another thing nobody talks about.
the chart everyone shows you, with the peak and the valley and the slow climb, is itself a small instance of the thing it describes. we look at it for ten seconds, we feel we understand it, we close the tab. that is the peak. some of us then read the actual paper, or try to, and feel suddenly that we don’t understand it at all. that is the valley. some smaller number of us go back to the chart, six months later, and find it more or less correct, but boring, and put it away. that is the climb. very few of us live on the plateau. the plateau is mostly empty. there are folding chairs.
i rest my case.
findings, depends on the day
i was not going to do the table. then i did the table. now the table exists, and i cannot, in good conscience, throw it away, because i spent forty minutes on it and that is, in my system, the same as it being important. this is the dunning kruger effect explained on the level of one tuesday morning, with one writer and forty-seven open tabs and a coworker upstairs who has, at this point, definitely been served snacks.
the cluster has two adjacent investigations on this exact effect, one on the underlying theory and one on the wording itself, both useful, neither of which i am borrowing from here. the third yoga mat is still under the couch, unrelated, but i mention it because i can see, from where i sit, the receipt for it on the wip 2022 list. the receipt has outlasted the intention. so it goes. i am also fairly sure there is a second pop culture reference that explains all this faster than i can — see “Anchorman”, where the protagonist discovers, mid-broadcast, the limits of his own preparation. the discovery is the chart.
findings, then. the simple version is a tool for looking outward, calmly. the complex version is a tool for looking inward, slowly. mike’s version is a tool for the bar. you need all three. you also need a fourth tool, which is shutting up, and which the literature i am fairly sure exists somewhere does not cover, because it cannot be charted.
carla is back from training. she is carrying a small plate. she did not look at me. that is, on balance, the result i wanted.
yours stupidly,
idiot again
writer of one table on a thursday, between the standing desk and the breakroom whiteboard
p.s. mike texted again. just the word “fine.” i am choosing to read it as a verdict on the wine, which he was not present for either. the chart i can’t unsee is now, technically, two charts.







