Four knobs. No villain among them. Turn any one of them a little and the readout moves a lot — that disproportion is the entire point, and it's the reason organizations get slow without anyone doing anything wrong.
People is the brutal one. Every person who must act is another multiplication, and multiplication of numbers below one is a one-way trip. Ten people who each show up 95% of the time give you a project that lands 60% of the time.
Org distance is the sneaky one. It doesn't change how likely any pair is to clash — that stays under two percent, because everyone is a professional. It changes how expensive it is to fix a clash. When your nearest shared boss is four levels up, escalating is a political act, so the friction just sits there and compounds.
Priorities is the one you'd never think to blame. Nobody is refusing to help. It's that "help at the moment it's needed" and "help eventually" are the same thing to the arithmetic, and different things to a person with eight other commitments.
Uncertainty doesn't touch the odds at all — it sets the price. Every step might hide a boulder costing ten times what the step would have. That's the gap between your estimate and your quarter.
These knobs feed back into each other. When the odds drop, everyone rationally invests a bit less, which drops the odds, which... The panel shows you a single moment. The real thing is a spiral, and it only turns one way.
Every idea, and most of the numbers, come from Alex Komoroske's Coordination Headwind: How Organizations Are Like Slime Molds. The friction exponent of 1.3 is his, and he is the first to say it's made up. The model is directionally true and precisely wrong. Read the original — it's 350 slides of emoji and it will change how you see your job.