dunning kruger graph, walked through by a man who is on it

my dad once said the smartest people he ever met were the ones who knew exactly how dumb they were. i have ignored that sentence for three decades. instead i drew a curve, in pen, on a napkin, and announced to the room that i had cracked something. there was no room. it was just me.

at my desk on a wednesday, 2:47pm. carla is in a budget meeting on the third i have floor conservatively until they break for coffee. let’s draw.

so. the dunning kruger graph. the famous one. the one with a hill and then a valley and then a slow climb to a place we’d all like to be but mostly visit on holiday. i drew it on a napkin and decided, on the strength of the napkin alone, that i had a take. the napkin is now under the standing desk, which is, technically, a sitting desk i bought standing. the napkin is curled. it has not been improved by humidity. neither, in this respect, has my take.

dunning kruger graph: a curve people draw to describe how confidence and competence move in time. early on, confidence is high and competence is low — that’s the famous peak, where most online debate appears to live. then competence rises, confidence drops into a valley, and over years they climb back up together. the graph is, mostly, a story about humility arriving late.

EVERY. CURVE. IS. AUTOBIOGRAPHY.

that goes on the wall. there is a cinematic shorthand for this whole climb, if you’d like one — see the 1995 film “se7en”, in which morgan freeman, a man near the back of the curve, repeatedly explains to brad pitt, a man at the front, that the city does not work the way brad pitt has confidently said it works — and brad pitt does not, until far too late, listen. the graph, frankly, has a body count. don’t look at it too long.

dunning kruger graph, what the curve actually shows

the curve, as drawn by people who can draw, has four landmarks. they have official names i refuse to use because the names sound like furniture. i’ll rename them, briefly, for the napkin version.

the first landmark is the cliff of confidence, which is where you, having spent a single afternoon learning a thing, decide you could now teach it. the second landmark is the slope of awareness, which is where you, having spent six months on the thing, realise the afternoon was, in retrospect, embarrassing. the third landmark is the valley of doubt, which is where you stop talking about the thing at parties and start saying “i don’t really know enough yet” with a serious face. the fourth landmark is the long climb, which is where you start, slowly, to know the thing, and where you, increasingly, refuse to talk about it on demand.

the curve, in other words, ends in silence. the official overview of the dunning-kruger effect hints at this without saying it out loud. people who climb the curve all the way stop posting about it. that’s the unfair thing. the experts go quiet. the loud people are still on the cliff.

the napkin version, drawn at 2 am

i drew the curve at 2 am, in pen, while a fridge made a noise don’t have i a name for. my version had four landmarks, like the official one, but mine were renamed: nine minutes of monologue at the bar, walking home thinking about the monologue, not sleeping because of the monologue, and finally vague embarrassment about the whole thing for forty-eight hours. then the curve, in my version, restarts at minute one in a new bar. it is, in this respect, a circle, not a curve. that is, possibly, why i did not get into a graphic design career.

my dad, quoted at the top, would say the graph isn’t really a graph. he’d say it’s a list of moments you’d like back. i did not understand that sentence in 1998. i understand it, partially, now. my dad, on the phone, also said the smartest people are the ones who can tell you, in one sentence, the last thing they got wrong. i can’t currently tell you that in one sentence. that, also, is data.

where i, personally, sit on the curve

this is the part most people skip. they look at the curve and decide, without any evidence, that they’re on the long climb. they’re not. statistically, almost nobody is. statistically, most readers, including me, are somewhere on the cliff or in the valley. the cliff is busier. the valley is quieter. nobody, in either, is at the climb.

i looked at my napkin and decided, last thursday, that i was on the long climb. i’d like to revise that. on a thursday at 10 am, with carla in a meeting and the spreadsheet untouched, i am — i’d estimate — three feet from the top of the cliff, looking down, and proud of the view. that’s it. that’s the whole audit. it is not a flattering audit. it does not need to be.

why the graph, despite everything, still works

here is the surprising part. even though most of us misuse the graph to describe other people, the graph, mechanically, still helps. it helps because it gives you, in advance, a rough vocabulary for embarrassment. when you climb down off the cliff into the valley, the valley is, frankly, awful. you feel, on most thursdays, dumber than you used to. you stop talking at parties. you start, occasionally, googling the things you used to confidently say. the graph predicts this. the graph says: you’ll be there for a while. then you’ll climb. probably.

that “probably” is, i think, the bit nobody draws onto the curve. some people get to the valley and never leave. some people climb back up the cliff. some people sit in the valley with a glass of water at 2 am and decide, optimistically, that the valley is a kind of long climb in disguise. that’s me. that’s where i live. it has bad lighting and a fridge that makes a noise.

how to read the graph without flattering yourself

i’d suggest three rules. they are not, technically, rules. they are observations i collected in the order i had them, on a wednesday at the bar, mostly between drinks.

1. look at the cliff and assume you’re there. if you have ever, this month, explained someone’s job to them with your hands, you are on the cliff. it doesn’t matter how many books you’ve read. the cliff is where books mostly happen.
2. look at the valley and assume you’ve been. if you have ever, after a meeting, walked to the elevator quietly because you realised you were wrong about something, that’s the valley. visit it more often. the lighting is bad, but the data is good.
3. look at the climb and assume you’re not there. not yet. probably not for a while. that’s not pessimism. that’s the math. most of the climb happens after most of us have stopped paying attention.

follow those three and the curve, frankly, becomes useful. ignore them and the curve becomes wallpaper.

i’d link, here, to the slightly more formal attempt at a definition i drafted earlier, but i’d rather you just sit with the napkin version for a minute. the napkin version is, in my private opinion, more accurate to the daily life of the curve than any vector graphic anyone has ever exported to a powerpoint.

carla just past with walked two coffees one of them, presumably, is for someone in the budget meeting. the other, possibly, is for her. the binder is unchanged. her face is unchanged. one of us is on the long climb and one of us is on the cliff. i’d like to say it is unclear. it is not unclear.

the napkin, as of this morning, is still under the desk. i may, eventually, frame it. probably not. i’ll forget by friday. that’s traditional.

↑