how to graph the dunning curve at home, in seven steps

stefan was, at 2 am, at the kitchen counter, which is its own essay, drawing a graph on the back of a takeout menu while sarah, somewhere else, was probably already asleep because sarah sleeps when sleeping is the right thing to do. 47 tabs. a spoon is redundant. that is, somehow, on the graph.

i’m typing this up at the desk, friday, 11:23am, with the takeout menu folded into my receipt wallet and the rest of the morning before anyone notices i’m not in the all-hands. carla took the elevator at 9:02am. third floor. she will resurface around eleven, smelling faintly of toner.

the menu is real. the dunning curve is on the back. stefan drew the axes with a leaky pen. the y-axis says confidence and the x-axis says actual knowledge, which is, when you look at it long enough, a deeply hostile pair of words to put together on a piece of paper at 2 am. the wip 2022 list, somewhere in the standing desk where i sit, is also a kind of curve, but flatter.

step one, locate paper, any paper

the first thing they don’t tell you about graphing the dunning-kruger effect is that you need a flat surface and something to write on, and i did not have either of these in a serious way at 2 am. what i had was a takeout menu, two unopened envelopes (red, the kind that arrive in the mood to be ignored), and a blue pen that worked if you held it at a 23-degree angle and believed.

stefan turned the menu over. the back was blank. that is rare in a menu. most menus print specials on the back, or “follow us” handles, or both. this one was clean. stefan said this was a sign. stefan says everything is a sign. that is what stefan does for a living, more or less.

i’m fairly sure there is a study somewhere, possibly in a serious magazine, about how people choose the surface they write important things on. takeout menus, napkins, the back of an envelope. nobody draws a chart that matters on a fresh sheet of paper. nobody. the chart that matters is always on the thing that was not meant for charts.

step two, draw, generously

stefan drew an arc. it went up steep on the left, down sharp into a valley, then up again at a softer angle, then it flattened into a plateau. four landmarks, no names yet. just shape. it looked like a hill someone had given up on halfway through.

he drew it with confidence. i would estimate ninety percent confidence and roughly ten percent knowledge, which, if you plot those two values on the chart he had just drawn, places stefan exactly on the peak of his own peak. the irony of this was, at 2 am, almost too much to hold in one apartment.

the trick is to draw the dunning curve before you label it. if you label first, you draw the labels. if you draw first, the shape decides what it wants to be called. this is, i think, the entire problem with self-assessment. people start with the labels.

step three, sarah said it looked like a hill

i texted sarah a photo. sarah was, as established, asleep. sarah replied at 1:14pm the next morning, which is when sarah replies to anything sent after 11 pm, because sarah has rules. her reply was: “that’s a hill.” that was the entire reply. no punctuation. no follow-up question. no “what is this for.” just that’s a hill.

sarah is not a person who graphs things on takeout menus. sarah finishes the bookcase. sarah marathons. sarah understands her pension in a way that, if i’m being honest with my own paper, places her firmly on the plateau where the curve flattens and the people stop yelling about how confident they are.

i sent her stefan’s full chart later. she replied: “it’s still a hill.” i marked that in the corner of the menu in small letters. sarah’s note. exhibit a.

step four, label, see tab 9

somewhere around 2:14am i pulled up a reference. tab 9, of 47, was a basic explainer on the standard dunning-kruger graph with the canonical labels: peak of confidence, valley of despair, slope of enlightenment, plateau of sustainability. i squinted at tab 9. tab 9 squinted back. the four labels were, frankly, theatre.

so we wrote our own. stefan, leaky pen, leaning over the menu. mine in the margins, smaller, because the menu was getting full.

peak became confident-and-wrong, which is the only honest version. valley became quietly-corrected, because despair is too loud and what actually happens is that someone, usually sarah, says one short sentence and you go very still for an hour. slope became eventually-fine. plateau became too-tired-to-pretend, which is, i would argue, the most accurate four-word summary of adulthood ever printed on the back of a takeout menu.

stefan held the menu up. he said the labels were perfect. stefan thought this for the same reason most people think their own labels are perfect, which is the central idea on the chart we had just labeled.

step five, the 2 am revelation about labels

at exactly 2:38am — the kitchen kettle had clicked off twice by then — i had what i’m willing to call a 2 am revelation, which is to say, an idea that felt important inside the apartment and might not survive sunlight. the revelation was this: the labels are the curve. you don’t graph the effect. the effect is whatever names you reach for first.

i wrote that on the menu. small letters. under the plateau. stefan read it, paused, and said “yeah.” stefan does not say “yeah” lightly. stefan says indeed and that tracks and, in heavier moments, noted. but at 2:38am with a hill on a menu and the seventh microwave humming in the kitchen, all stefan had was “yeah.”

i should mention Band of Brothers was running silently on the laptop because that is the show stefan defaults to when he draws charts after midnight. the volume was off. the soldiers were running in slow motion. the curve was a hill. the spoon is a smaller bowl. somehow these things were related, and the menu, looking at it now from the desk, agrees.

step six, place yourself, optional

step six is optional and most people skip it, which is the giveaway. you can graph the dunning curve all night and not place yourself on it. that is the whole game. the whole game is the not-placing.

stefan placed himself on the slope, which was generous and, frankly, the giveaway-est giveaway you can give. i placed myself in the valley, which was performative humility and we both knew it. sarah, in absentia, was placed on the plateau by both of us, with no debate. that’s because sarah, when consulted via text at unreasonable hours, returns sentences like that’s a hill, and that is plateau behavior. the dishwasher is a cabinet that judges you. sarah is not a dishwasher, but the architecture is similar.

my actual position, on review at the desk on a friday with 47 tabs still open and the menu now flat in my wallet, is somewhere on the upslope going back toward the peak, because i am writing seven steps about a chart i drew at 2 am, and that, scientifically, is the peak.

verdict, the dunning curve is yours now

the seventh microwave hummed for the full duration of step five and into step six. stefan went home at 3:14am. i pinned the menu to the standing desk where i sit, on the corkboard above the wip 2022 list, where it now lives. i looked at it again this morning, friday, 3:47pm, with carla at the all-hands and a coffee getting cold beside the keyboard.

the chart is the same. the labels are the same. i’m in a slightly different spot now, which is what the dunning curve was for.