The Lindy effect
This is a named pattern with documented cases and a sound logic behind it, not an effect anyone has measured. Treat it as a lens that keeps proving useful rather than as a law.
In June 1964 the writer Albert Goldman published a piece in The New Republic called Lindy's Law. It described a belief held by the comedians and agents who gathered at Lindy's delicatessen in New York, where they conducted what Goldman called post-mortems on recent show business action. Their rule concerned how long a television comedian would last. They held that a performer had a fixed stock of material, so heavy exposure burned through it. Take a weekly show and your chances of surviving the first season were slight. Confine yourself to specials and guest appearances and you might last for decades.
Goldman was not endorsing this. He described its holders as bald-headed, cigar-chomping know-it-alls, and the article is closer to satire than to analysis. There is also a textual oddity worth noting, since the sentence states that life expectancy is proportional to exposure while the examples describe the inverse, which is why later writers quote it with the word inversely inserted.
Benoit Mandelbrot took the name and turned the claim around. In The Fractal Geometry of Nature in 1982 he argued that for output bounded by the life of the producer, future life expectancy is proportional to the past. The more appearances a comedian had already made, the more he could be expected to make. This is the version that survived, and it carries the deli's name while contradicting the deli's belief.
Nassim Taleb extended it in Antifragile in 2012, drawing the distinction that made it travel beyond entertainment. Perishable things have a natural upper bound on their lifespan, so every additional day shortens what remains. Nonperishable things, meaning ideas, books, technologies and practices, have no such bound, so every additional day of survival lengthens what can be expected. By Taleb's own account he proposed that boundary to Mandelbrot, who agreed that the nonperishable case would be power-law distributed while the original comedian story worked as a metaphor.
The mathematics underneath is a decreasing hazard rate. If the instantaneous risk of something ending falls as it ages, then conditional remaining life increases with age, which is exactly the Lindy claim. Power-law and Pareto distributions have this property. So does a population with hidden variation in quality, where the weak examples are removed early and the survivors are disproportionately the durable ones.
That last mechanism is the reason the effect is contested. The winnowing that produces it is survivorship, which means the effect can be described as survivorship bias functioning as a filter rather than as an error. Critics have argued the claim is close to true by definition once you have assumed the right distribution, and that direct empirical validation is largely absent, since testing it properly would require sampling things at a moment in time and watching which survive.
The scope is also unsettled. Toby Ord, working at Oxford, has pointed out that the effect has no agreed precise definition, and that Taleb's perishable and nonperishable boundary does not hold cleanly: species with high early mortality are perishable and still show increasing expected remaining lifespan across most of their range. So the direction of the idea is widely accepted and the edges of it are not.
The practical use is about durability rather than quality, and keeping those apart is what makes it usable.
QWERTY is around 150 years old. The mechanical problem its layout was arranged around stopped existing decades ago. It survived because millions of people had already learned it, and by then replacing it cost more than living with it. So age told you correctly that QWERTY would persist, and told you nothing about whether it was any good. Anyone using the Lindy effect as an argument for quality has misunderstood what survived.
What it is genuinely good for is the question of what you are building on top of. Every process you design rests on things you did not build, and each of those is a bet that the thing underneath will outlast the thing above it. Your finance process rests on double-entry bookkeeping, five hundred years old, and on whichever accounting product you signed up for last year. Only one of those is certain to still be there when you need it. The other can be acquired, discontinued or repriced, and the process you built around it comes apart when it goes. The same question sits under the artificial intelligence tool somebody is building a workflow on this quarter, which is at most a couple of years old, while the spreadsheet underneath it is forty. That difference is not proof that either one is good. It is the cheapest evidence available about which will still be standing when the workflow depends on it.
It also reorders who has to argue. The default in most organisations is that the existing practice defends itself while the new tool is assumed sound until it fails. The old practice has already produced evidence of durability. The new one has that still to demonstrate, and asking it to is not conservatism, it is reading the evidence that exists.
Where the whole thing breaks is genuine regime change. The old way dies when the conditions it was adapted to stop holding, and no amount of prior survival protects it. Film photography had a century of persistence behind it. So the honest version of the heuristic is that age predicts persistence under conditions that resemble the past, and it is silent about conditions that do not.
Two related entries sit close to this one. Survivorship bias is the failure mode of studying only what came back, and the Lindy effect is the same selection working as a filter rather than as an error, which is why the two are easy to confuse and worth holding apart. And Taleb ties durability to exposure: things that have lasted have been tested by people who bore the cost of being wrong, which is skin in the game.
Lindy says the longer something has lasted, the longer it can be expected to last. The replication crisis says a third of psychology's most famous findings failed when re-run, and Hawthorne, the prison experiment and the ten thousand hour rule each lasted decades before anyone checked. They look like the same rule and they are not. Lindy measures persistence, which is evidence about durability and silent about truth. Replication measures whether a finding survives being tested again. A finding can persist for eighty years in textbooks because nobody went back to the data, which is QWERTY, not evidence. The version of Lindy that holds for knowledge is the replication entry's own rule: trust what has been re-tested for fifty years, not what has merely been repeated for fifty years.
Lindy says what has survived a long time is a good bet to survive longer. The Matthew effect says recognition flows to whoever already has it, and the gap that opens has little to do with how large the original difference was. So the thing that lasted may have lasted because it was already known, not because it was better. The Lindy entry concedes this itself: survivorship is both its mechanism and the strongest objection to it. What separates them is exposure. Something that persisted because people kept choosing it while bearing the cost of being wrong has been tested. Something that persisted because it was already the famous one has only been repeated. Age tells you which pile a thing is in only after you ask who paid to keep it there.
Source: Goldman, Lindy's Law, The New Republic, 13 June 1964, pages 34 to 35. The inversion is Mandelbrot, The Fractal Geometry of Nature, 1982, page 342. The extension to nonperishable things is Taleb, Antifragile, 2012, pages 317 to 318, developed further in Skin in the Game, 2018. On the unsettled definition and the counterexamples to the perishable boundary: Ord, The Lindy Effect, 2023.
Nassim Nicholas Taleb, 2012
Where the Lindy effect is set out in full, inside the larger argument about things that gain from disorder. It argues with you, which is the point.
Draw your own card. It does not take long, and it rewards taking your time.