Antikondo: Orderliness is not the same as efficiency

Often, if you ask a human to optimize something, they'll make it orderly: straight lines, simple layouts and clean divisions, but when nature (or evolutionary algorithms) optimizes things, it produces redundancy, gradients, tangles, and complexity — ironically, robots produce systems that look like nature designed them, while humans produce systems that look like robots designed them.

In an essay called The Efficiency-Destroying Magic of Tidying Up, Uber product manager Florent Crivello lays out a thesis about the value of complexity, a subject that will be familiar to readers who enjoyed Tim Harford's 2016 book "Messy: The Power of Disorder to Transform Lives."

Crivello suggests a three-part test that "messy" systems should be subjected to before anyone attempts to tidy them:

1. How much information is contained in the system's current state? What constraints are expressing themselves through it?


2. How old is the system? How malleable is it? How strong are the forces put on it?

3. Finally: who is complaining about the chaos?

Finally: who is complaining about the chaos? If outsiders complain, but people living inside the system seem happy with it, it probably means that the chaos is serving them right, and that it's just foreign eyes who are unable to perceive its underlying order.

This is a special case of Chesterton's Fence, which states you should never take down a fence before knowing why it was put up. Here, I propose Scott's Law: never put order in a system before you understand the structure underneath its chaos.

The Efficiency-Destroying Magic of Tidying Up [Florent Crivello]

(via Four Short Links)