We instinctively improve things by adding — more features, more rules, more supplements. But the most robust gains usually come from removing: cutting the harmful, the unnecessary, the toxic. What you take away is more certain than what you add.
Via negativa: improvement comes more reliably from subtraction — removing what harms or doesn't work — than from addition. Knowing what to avoid is more robust than knowing what to add.
Taleb borrows the term from theology (defining God by what he is not) and applies it to decisions: we know what's bad far more reliably than what's good. A doctor is on surer ground removing a known poison from your diet than prescribing a new supplement; a writer improves a draft more reliably by cutting weak sentences than by adding clever ones. Negative knowledge — what to avoid — is more durable than positive knowledge, because it takes only one counterexample to know something harms, while 'this helps' is forever provisional.
Yet our instinct runs the other way. Faced with a problem, we add: a feature, a rule, a meeting, a layer of process, a new habit. Subtraction feels like doing nothing, so it's chronically undervalued — researchers have shown people overwhelmingly default to additive solutions even when removing something would work better. The result is bloated software, tangled bureaucracies, cluttered lives, and 'solutions' that create new problems. Often the highest-leverage move is to take something away.
Practically, ask 'what can I remove?' before 'what can I add?' Health: cut sugar, smoking, and sitting before chasing the perfect supplement. Productivity: remove distractions and commitments before adding systems. Design and writing: delete before you embellish. Decisions: eliminate the clearly bad options first, which often leaves the answer obvious (this is inversion's cousin). Subtraction is humbler, more reliable, and usually cheaper — and because what we remove is more certain than what we add, it carries less risk of backfiring.
Our default is to fix problems by adding — features, rules, habits — which bloats and backfires; via negativa points at the more reliable, lower-risk lever we habitually overlook: taking the wrong thing away.
It's not 'never add anything' or 'minimalism for its own sake.' Some problems genuinely need new things built. The point is that we're biased toward addition and undervalue removal — so subtraction deserves to be considered first, not that it's always the answer. Add when you must; but check what to cut first.
Lock this idea into memory with a 5-minute active-recall session — the science of spaced repetition, no signup.
Try this idea free →