Nassim Taleb's 2012 book names a property he argues has no word in English: things that gain from disorder.

The three categories

Fragile things are harmed by volatility — a parcel marked fragile wants to be left alone. Robust things resist it and are unchanged. Antifragile things improve from it, within limits: muscle under load, an immune system exposed to pathogens, a business that gains share when competitors fail in a downturn.

Taleb's claim is that the third category is common and unnamed, and that mistaking antifragile systems for robust ones leads to protecting them from the stress they need. Removing all volatility from a system produces something that looks stable and has lost the capacity to adapt, so the first large shock does disproportionate damage.

Asymmetry

The practical content is about the shape of exposures rather than forecasting. An antifragile position has limited downside and open-ended upside, so surprises help on average regardless of direction.

His barbell puts the large majority of resources in the safest position available and a small portion in speculative bets, avoiding the moderate middle where a serious loss is possible and a large gain is not. Applied to a career, that is a stable income base plus small experiments with real upside, in place of a single moderately risky commitment. See optionality and asymmetry.

Via negativa

He argues that knowledge of what harms is more reliable than knowledge of what helps, so removal beats addition. Subtracting a fragility is a more dependable improvement than adding an optimization, because the first survives being wrong about the model and the second does not. This is inversion applied to system design.

The style

The book is discursive and combative, and the argument is interrupted by attacks on named opponents. Readers who find that tiresome can take the barbell and via negativa chapters and leave the rest.