a builder's codex
codex · operators · Nassim Nicholas Taleb · ins_taleb-iatrogenics

Iatrogenics — when the intervention causes more harm than the disease — most "fixes" in complex systems are net-negative

By Nassim Nicholas Taleb · Risk theorist; former options trader; author Fooled by Randomness, Black Swan, Antifragile, Skin in the Game · 2012-11-27 · book · Antifragile — Iatrogenics

Tier A · TL;DR
Iatrogenics — when the intervention causes more harm than the disease — most "fixes" in complex systems are net-negative

Claim

Iatrogenics is harm caused by the healer — intervention that causes more damage than doing nothing. In medicine, in policy, in product, and in management, intervention is often net-negative because complex systems have side effects, cascading failures, and hidden dependencies that the intervener does not see. The default move in poorly-understood systems should be inaction, not aggressive correction.

Mechanism

A complex system has many interacting components, most of which the intervener does not understand. Any change introduces second-order effects (changes in other components reacting to the changed one) and third-order effects (interactions between the second-order changes). The intervener typically optimises for the first-order effect they can see, while the second- and third-order effects often produce harm that exceeds the benefit. Iatrogenics is the structural reason that "wait and observe" beats "do something" in many complex situations — the something usually does more than intended, and the unintended part is usually worse than the intended part.

Conditions

Holds when:

Fails when:

Evidence

"Iatrogenics is Taleb's term for harm caused by the healer: intervention that causes more damage than doing nothing."

— see raw/expert-content/experts/nassim-taleb.md line 18.

Signals

Counter-evidence

Iatrogenic caution can become an excuse for organisational inertia. Companies that wait too long to intervene on real problems suffer different failure modes (declining metrics, accumulated tech debt, eroded customer trust). The discipline is matching intervention size to confidence: small interventions are often safe even under uncertainty; large interventions need correspondingly large confidence in the diagnosis.

Cross-references

Open the interactive view → View original source → Markdown source →