Work that used to take a week takes a quarter, and you have quietly started to wonder whether the problem is you. It isn't, and the reason is arithmetic. These are five ways into that idea — all of them adaptations of Coordination Headwind, Alex Komoroske's 350-slide emoji flipbook. Every idea in them is his.
Each is self-contained and takes a few minutes. They make the same argument in deliberately different ways, so pick whichever sounds like your kind of thing.
Answer four questions about your actual job, and the whole thing re-runs on your numbers. Watch your own obviously-worth-doing project decay into a coin flip, and finish on a single figure: how much of your effectiveness the structure is quietly taking.
You are Amber, you have until Friday, and you make four genuinely defensible choices. Then it plays out all 81 weeks you could have had and shows you where yours landed. The best available week is 19%, and every decision you made moved it by 14 points total.
Four knobs and a readout, built like a piece of hardware. Drag any one of them and watch a project's odds collapse. Ninety seconds, almost no reading, and the gap between how far you turned the knob and how far the number moved does the whole argument for you.
Scroll through the argument with a real slime mold growing underneath it — a live Physarum simulation, thousands of agents sensing and depositing and reinforcing trails, thinning and fragmenting as the argument turns.
Two identical colonies race for the same distant goal. One goes straight at it; the other has five small, boring food sources on the way. Watch what each actually banks, and when. This is the part of the deck that tells you what to do instead.
The first four share one model, and it reproduces Komoroske's own stated figures exactly — 0.99¹⁰ = 0.90, 0.95¹⁰ = 0.60, 0.90¹⁰ = 0.35, and 45 pairwise relationships in a team of ten. A project succeeds if everyone invests properly and no pair of them seizes up, and uncertainty sets what it all costs:
P = p(load)ⁿ × (1 − 0.002·h^1.3)^(n(n−1)/2) ÷ (1+10u)(1+u/2)
The friction exponent of 1.3 is his, and he is the first to admit it's invented. The base rate is calibrated so risk per relationship never exceeds 2% — nobody in the model is behaving badly, ever — while the total across 45 pairs is still devastating. That gap between innocent parts and a terrible whole is the argument.
Where the simulations are and aren't evidence. The colonies in 04 and 05 are real Physarum models running live, not recordings. But a simulation that illustrates an argument is not the same as one that proves it, and the two are kept apart here.
In 04 the colony is atmosphere. It is not solving your org chart, and the Tokyo rail result is credited as Komoroske's cited finding rather than something happening on screen — two stages that overclaimed were cut rather than reworded.
In 05 the comparison is real and reproducible: the moonshot colony cannot detect its goal, banks a trickle by luck, and finishes with about a third of what the stepping-stone colony banks. What it does not show is that moonshots fail — given enough wandering, that colony can stumble onto the goal. That weaker, true claim is the one the page makes.