the second hundred · metaphor 111
Of two ideas, books, or customs — the shiny new one and the one that has already survived a century — which should you bet will last longer? For things that don't age like bodies, survival predicts more survival: the old is the safer bet.
Walk into a bookshop. The novel that has stayed in print for two hundred years will likely be read in another two hundred; this season's sensation will likely be pulped by spring. The fork, the wheel, the handshake, the proverb — "it's been around forever" is usually said as a shrug, but it is quietly a forecast, and a good one. The tool, the recipe, the institution that has "been around forever" has, by lasting, told you something about how long it will keep going.
For a body, age is a countdown: every year lived is one closer to the wall, and the time expected to remain only shrinks. For a non-perishable — an idea, a craft, a custom, a technology — age runs the other way. Each decade survived is evidence the thing belongs to the durable kind, so its expected future lengthens the longer it has already lasted. The Lindy effect is the name for that reversal.
Expected remaining life vs. age already survived
Two worlds, overlaid. Drag anywhere on the chart (or use ← →) to move the age you are asking about.
↑ expected years still ahead · → age already survived · drag / arrow keys
Age as evidence of robustness
The Lindy effect is a single, checkable claim: for non-perishable things, expected remaining life increases with current age. The engine underneath is the hazard rate — the chance of dying in the next instant, given you have made it this far. For a body that hazard rises: parts wear, and each year makes the next one deadlier. For a fad, most die young, so merely surviving thins the field and lowers your hazard — the survivors are the sturdy ones, and you are now among them.
A power-law lifetime makes this exact. Its hazard is α ⁄ t — literally falling as one-over-age. Integrate that out and the expected life ahead is m(t) = t ⁄ (α−1): proportional to how long the thing has already lasted. At α = 2 that is the folk rule of thumb — expect it to last as long again as it already has. A century-old book: another century, on the house. This is why heavy tails, and only heavy tails, do this: a fat tail means the truly long-lived cases are common enough that being old is strong evidence you are one of them.
What to try
Drag the age marker. At young ages the slate curve sits higher — a newborn body has a whole life ahead, while a week-old idea has no track record and most week-old ideas are gone by month's end. Push the marker right and the bronze curve overtakes it: past the crossover, the century-old custom out-promises the middle-aged man. That switch — from a rising curve to a falling one being the better bet — is the whole story in one gesture.
Now work the α dial. Pull α toward 1 and the bronze line steepens into super-Lindy: a heavier tail, where the old strongly predicts the ancient and the crossover slides young. Raise α and the rise flattens toward the dashed memoryless line, where age is no evidence at all. What the dial never does is bend bronze downward — no tail exponent turns a non-perishable into a body. Below the chart, the cohort makes survivorship visible: count how few of ten thousand ideas reach old age, and how long each rare survivor still expects to run.
The mapping
The move is a filter for durability. Choosing what to read, learn, build on, or trust, prefer what has already survived. A theorem proved two thousand years ago will outlive this year's framework; the structure of a classical grammar will outlast this season's slang; the recipe kept for five generations will outlast the viral one. When you pick a skill worth learning, weight the ones that have paid off for centuries over the ones minted last quarter — the old skill is the safer bet precisely because it is old.
"It has survived this long" is not nostalgia. For the right class it is a live, quantitative argument about the future: this thing has repeatedly passed tests you cannot even enumerate, and each pass is evidence of the next. Technologies, institutions, tools, traditions — where the lifetime is heavy-tailed, endurance is collateral. The metaphor hands you a reusable move: before betting on how long something will last, ask which curve it is on, then let its age speak.
The honest model
Everything above holds for exactly one kind of thing: the heavy-tailed, non-perishable class. The panels draw theorems, not opinions — a falling hazard forces expected remaining life to rise, a rising hazard forces it to fall, a constant hazard pins it flat. But which curve a given thing sits on is an empirical question the math cannot answer for you, and getting it wrong inverts every conclusion.
Two honesties come attached. First, Lindy is a statement about a category, never a promise to any member: the specific old book in front of you can still burn tomorrow. Second, you are always reading the future off the survivors — and the survivors are all you see. That is both the source of the effect and its trap, and it is where the metaphor tears.
Mathematics ↔ life
| Mathematics | Life |
|---|---|
| current age t | How long it has already lasted — the length of the track record you can see. |
| power-law lifetime | A non-perishable: an idea, a custom, a tool. Its lifetime is heavy-tailed, so the long-lived cases are not rare. |
| expected remaining life m(t) | How much longer to bet on — the expected future still ahead, given survival to now. |
| rising with age | Lindy: the old lasts. Each year survived lengthens the expected future; endurance predicts endurance. |
| bell-curve lifetime | A perishable: a body. Lifetimes cluster around a typical span, so survival past it is genuinely running out. |
| falling with age | Mortality: every year survived is one subtracted. Age is a countdown, not a credential. |
Where the metaphor tears
Lindy holds only for the heavy-tailed, non-perishable class — and never for perishables. Applied to a person, a fruit, a machine part, a mortal body of any kind, it is simply false: there, every year survived shortens the expected future, not lengthens it. Mistaking a perishable for a non-perishable is the entire error. Before invoking "it has lasted, so it will last," check which curve you are actually on.
The rule reads the future off the survivors — but the survivors are all that reach you. The thousands of contemporaries of the old book that died leave no trace, so the class looks more durable than any single bet is safe. Lindy describes the expected future of the surviving population; it does not promise that your particular relic is one of the sturdy ones rather than a fluke about to fall. The survivorship that creates the effect also flatters it.
The math assumes the world that let a thing survive keeps holding. It does not always. A technology Lindy-safe for centuries can die in a decade when its substrate shifts; a custom robust for generations collapses when the conditions that fed it vanish. Heavy tails describe steady regimes — and a regime change is precisely the event the tail never priced. Age is armor against ordinary risk, not against the ground moving.