Ask a sculptor how to make a statue and you get the least helpful answer in art: take a block of marble and remove everything that isn’t the statue. Michelangelo swore the figure was already inside the stone — his job was only to chip away the excess. Ask an editor how to improve a paragraph and you get the same shrug: cut the words that aren’t pulling weight. Neither adds. Both subtract, and the result is stronger for it.
That instinct — that the road to a better system runs through removal rather than addition — is what Nassim Taleb calls via negativa, and it is the beating heart of applied antifragility. We’ve spent three lessons on the theory: the fragile–robust–antifragile triad, hormesis, and how over-smoothing breeds fragility. This is where it becomes a way of acting.
Before you read — take a guess
You want to make a system more robust and you can either (a) add a new speculative feature you think will help, or (b) remove a known source of harm. Which is the more reliable bet, and why?
Via negativa: defining by what you remove
The phrase is borrowed from negative theology — the medieval habit of describing the divine not by listing its attributes but by stripping away everything it is not (not finite, not mortal, not visible). You approach truth by elimination, because the negative claims are the ones you can be sure of.
Taleb repurposes this into a decision rule:
Via negativa — improve a system by removing the fragilising agent, rather than adding a speculative “improvement.”
The sculptor and the editor are doing exactly this. So is the doctor who takes you off a drug, the investor who pays down debt, the engineer who deletes a module. Subtractive knowledge — knowing what to cut — is more robust than additive knowledge, and the rest of this lesson explains why, then puts it to work.
When to use it
Reach for via negativa first whenever a system is misbehaving and you’re tempted to add something to fix it. Before bolting on a patch, a hedge, or a feature, ask the cheaper, safer question: what can I take away? It’s the default move, not the last resort.
Why subtraction beats addition
There are three reasons removal is the more robust play, and they stack.
(a) A known harm has a more reliable payoff than a speculative good. When you remove something that is demonstrably hurting you, the benefit is close to certain and the side effects are minimal — you’re subtracting a cause, not introducing a new one. When you add something, you import every unknown it carries. Medicine has a word for this: iatrogenics — harm caused by the healer. The intervention that was supposed to help does damage you didn’t foresee, because complex systems answer additions with surprises.
(b) Negative knowledge is more robust than positive knowledge. This is Karl Popper’s point in miniature: no pile of confirming observations proves a theory (“all swans are white” survives a million white swans and dies on one black one), but a single counterexample disproves it for good. Disconfirmation is decisive; confirmation is only provisional. We know what’s wrong with far more confidence than we know what’s right.
(c) So subtract with confidence, add with suspicion. Because our negative knowledge is sturdier, actions built on it inherit that sturdiness.
Here’s the contrast in numbers. Suppose you want to improve your health and weigh two moves:
| Move | Direction | Payoff certainty | Side-effect risk | Net |
|---|---|---|---|---|
| Take a new “optimising” supplement | Add a speculative good | Low (weak, contested evidence) | Unknown (interactions, contaminants) | Fragile |
| Quit smoking | Remove a known harm | Very high (overwhelming evidence) | Essentially none | Robust |
The supplement might help by a little and might hurt in a way nobody measured yet. Quitting smoking has a near-certain, large payoff and no downside tail. Same goal, opposite risk profiles — and the subtractive move wins without contest.
When to use it
Whenever the evidence for an addition is thin or the system is complex enough to surprise you. The thinner your positive knowledge, the harder you should lean on the negative.
Via negativa in practice
The rule earns its keep across wildly different domains. A few concrete cases:
Health and medicine. The largest health gains almost always come from removals: quit smoking, cut added sugar, drop ultra-processed foods and industrial seed oils, break up prolonged sitting. The oldest rule in the field — primum non nocere, “first, do no harm” — is pure via negativa. Modern practice even has a name for the subtractive art of taking patients off unnecessary medications: deprescribing. Notice the pattern: the wins are things you stop.
Finance and business. The great fragiliser is debt — it turns an ordinary bad quarter into insolvency, so cutting it is the highest-leverage subtraction there is. Then remove single points of failure, fire the one client whose departure could sink you, and kill the fragile product line that only prints money in good times. Steve Jobs put it bluntly: focus is “saying no to the hundred other good ideas.” Strategy, for him, was deciding what not to do.
Writing, design, and engineering. You improve prose by deleting, a UI by removing options, a system by cutting the module nobody trusts. And when a process is slow, you don’t sprinkle in more capacity everywhere — you find the one bottleneck (the tightest constraint) and relieve it. Adding effort anywhere else just piles up inventory in front of the jam. Removal, aimed precisely, is the whole game.
A hospital wants better patient outcomes. Which intervention is the most via-negativa move?
The barbell strategy
Now the flagship application. A barbell strategy combines extreme safety on one end with a small allocation of bounded, aggressive risk on the other — and deliberately avoids the fragile “moderate” middle. Picture a literal barbell: heavy plates at both ends, nothing in the bar between them.
Why does this work? Straight from the asymmetry-and-optionality lesson:
- The safe end guarantees survival — its downside is capped. You cannot be wiped out.
- The risky end captures convex upside — its downside is small and bounded, its upside open.
- You never bet the survival money. The two ends don’t mix, so a blow-up on the risky end can’t reach the safe end.
The trap is the middle. A “medium-risk” position feels prudent, but it quietly carries exposure to the fat tail without the protection of true safety — it survives normal times and dies in the crash, the worst possible combination.
Here’s a worked example. Compare two $100,000 portfolios across a normal year and a crash year. The barbell holds 90% in ultra-safe cash/short bonds (call it +2% in normal times, and −2% in a crash as a haircut) and 10% in high-optionality bets that can lose it all or multiply (say +150% in a good year, −100% in a crash). The medium-middle holds 100% in a diversified “moderate-risk” portfolio (+8% normal, −45% in the crash tail).
| Normal year | Crash year | |
|---|---|---|
| Barbell — 90% safe | $90,000 × 1.02 = $91,800 | $90,000 × 0.98 = $88,200 |
| Barbell — 10% risky | $10,000 × 2.50 = $25,000 | $10,000 × 0.00 = $0 |
| Barbell total | $116,800 (+16.8%) | $88,200 (−11.8%) |
| Medium-middle total | $100,000 × 1.08 = $108,000 (+8%) | $100,000 × 0.55 = $55,000 (−45%) |
In the good year the barbell’s tiny risky sliver actually out-earns the whole medium portfolio, because convex upside compounds. In the crash, the barbell loses only its risky 10% and its safe end barely flinches — you’re down 12% and fully alive. The medium-middle is down 45% and gutted, because it had tail exposure with no fortress behind it. That is the barbell’s promise: survive the crash and keep the upside, precisely because you refused the moderate middle.
Fragility tester
Via negativa on a fragile system
Pick a system by the SHAPE of its payoff-vs-stress curve: FRAGILE (concave — big shocks hurt disproportionately), ROBUST (flat — indifferent) or ANTIFRAGILE (convex — it gains from disorder, up to a dose limit). Set the volatility, then fire shocks and watch the cumulative outcome. The same storm bleeds the fragile and feeds the antifragile. Flip on via negativa to REMOVE the ruinous exposure and re-run.
The system — the shape of its response to disorder
Payoff vs stress (the curvature)
Cumulative outcome over shocks
Fragile · concave · concave — harm accelerates · volatility 65 · 0 shocks fired · mean per shock — · worst single — · cumulative +0.0. via negativa — capping the catastrophic downside turned a bleeder into something that shrugs off the storm; you added nothing, you removed the ruin.
The tester makes the point physical: the fragile system doesn’t need a clever new feature to survive the shocks. Cap its worst-case downside — remove the exposure that kills it — and the same volatility that was draining it becomes survivable. Turn the cap off and the bleeding resumes. No addition ever entered the picture.
Skin in the game
The last move is as much ethics as engineering. Skin in the game means the decision-maker must bear the downside of their own decisions — symmetry between the upside they can capture and the downside they can suffer.
It forces antifragility on a system in three ways, and each is a removal:
- It removes the hidden transfer of fragility. The classic corruption is heads-I-win-tails-you-lose: the trader pockets the bonus in good years and the taxpayer eats the blow-up. Skin in the game deletes that asymmetry — you can’t offload your volatility onto others while keeping the gains.
- It filters out reckless actors. People who must live with their own tail risk stop taking insane ones. The market of decision-makers self-cleans, because ruin removes those who court it.
- It aligns incentives. This is the principal–agent problem dissolving: the agent and the principal now share one payoff, so the agent’s private interest and the system’s health point the same way.
Notice that skin in the game is itself a via-negativa move. It doesn’t add a monitoring bureaucracy or a compliance manual on top of a broken incentive — it removes the ability to offload risk in the first place. Fix the exposure and you don’t need the paperwork.
And there’s an ethical edge we’ll sharpen in the final lesson: transferring your volatility to others while keeping the upside isn’t clever antifragility — it’s predation. An antifragile individual who achieves it by making everyone around them fragile hasn’t solved the problem; they’ve merely relocated it, and billed someone else.
Which of the following are true of skin in the game? (Select all that apply.)
Sorting the moves
Sort each move by whether it robustifies through subtraction (removing a known harm) or through addition (bolting on a speculative good).
- Pay down debt
- Add a new optimising supplement
- Bolt on another external dependency
- Remove the single point of failure
- Add more leverage to boost returns
- Quit smoking
- Add a complex hedge nobody understands
- Cut the one client who could sink you
- Delete the risky product feature
Match the tools
Match each concept to its precise definition.
Recap
Big picture
Via negativa: strength by subtraction
- Via negativa
- Why subtraction wins
- Known harm removed → certain payoff
- Addition imports side effects (iatrogenics)
- Negative knowledge is more robust (Popper)
- In practice
- Health: quit smoking/sugar; deprescribe
- Finance: cut debt, kill single points of failure
- Design: delete, relieve the bottleneck
- Barbell strategy
- Extreme safety + small bounded risk
- Avoid the fragile moderate middle
- Survive the crash AND capture upside
- Skin in the game
- Bear your own downside (symmetry)
- Removes the fragility transfer
- Offloading risk = predation
- Why subtraction wins
Takeaways
- Via negativa improves a system by removing the fragiliser, not adding a speculative fix. Subtraction is the default move.
- Removal wins because a known harm has a near-certain payoff, while additions import iatrogenic side effects — and negative knowledge is more robust than positive claims.
- The barbell — extreme safety plus a small bounded bet, no fragile middle — lets you survive the crash and capture convex upside.
- Skin in the game is itself via negativa: it removes your ability to offload risk onto others, aligning incentives and filtering out the reckless. Transferring volatility while keeping the upside is predation, not antifragility.
You now know how to robustify — by subtraction. But that leaves the sharper question: what should you subtract, and how do you find it before the shock lands? In Lesson 5, “Spotting Fragility in Advance,” we build the detective’s toolkit for identifying hidden fragility while the system still looks calm — so you can cut the right thing before the tail arrives, not after.