The Lindy Effect — What Survival Tells Us About Duration
The Lindy Effect proposes that the longer something has survived, the longer it is expected to remain. This article examines the mathematics, the evidence, and the limits.
A comedian performs at a New York deli in the 1960s. The club is called Lindy’s. The observation is simple: a book that has been in print for forty years will likely be in print for another forty. An idea that has survived for two thousand years will likely survive for another two thousand. The older it is, the longer it will last.
This observation has a name: the Lindy Effect. For non-perishable things — ideas, technologies, books — age is not just a record of survival. It is a predictor of future survival. The effect does not apply to perishable things. A loaf of bread does not become fresher as it ages. A human being does not gain a longer life expectancy by reaching old age. But for things that do not decay on their own, the relationship is reversed. Survival itself becomes evidence of durability.
The origin
The term comes from a comedy club at Lindy’s Deli on Broadway in New York City. Comedians observed that older routines tended to be better — because the bad ones had already died out. The observation was informal. It was not derived from data.
Nassim Nicholas Taleb formalized the observation in his 2012 book Antifragile: Things That Gain from Disorder. He drew the distinction between perishable and non-perishable things and gave the effect its name. The term is a proper noun — the name of the deli where the observation originated.
Taleb’s formulation is straightforward. For a non-perishable item, every additional year of life increases the expected remaining lifespan. A technology that is ten years old has a longer expected remaining lifespan than a technology that is one year old. The key constraint is that the item must be non-perishable. A book does not rot on its shelf. An idea does not decay in the absence of refutation. The Lindy Effect applies to things whose primary threat is oblivion, not deterioration.
The mathematics
The Lindy Effect is a mathematical consequence of a specific class of probability distributions.
Consider a random variable T representing the total lifetime of an item. The survival function S(t) = Pr(T > t) gives the probability that the item survives beyond time t. For the Lindy Effect to hold, the survival function must follow a power law:
S(t) = (c / t)^ε
where c is a constant and ε is the shape parameter. This is the same functional form as the Pareto distribution.
The expected remaining lifetime, given that the item has already survived to time t, is:
E[T - t | T > t] = p · t
where p = 1/ε is the Lindy proportion. The remaining life expectancy is proportional to the current age. If p = 1 (which corresponds to ε = 1), then the expected remaining life equals the current age. A book that is forty years old is expected to be in print for another forty years.
The mathematical derivation is due to Itai Eliazar, who formalized the relationship in his 2017 paper “The Lindy Effect,” published in Physica A: Statistical Mechanics and Applications. Eliazar showed that the Lindy Effect holds if and only if the lifetime distribution follows a power law with shape parameter ε between 0 and 1. When ε >= 1, the expected remaining lifetime is finite. When ε < 1, the expected remaining lifetime is infinite. The condition ε between 0 and 1 means that the Lindy Effect is a property of heavy-tailed distributions.
What the evidence shows
The Lindy Effect has been observed in several domains.
Books. Eliazar analyzed the lifetimes of books published in the United States and found that the survival distribution follows a power law with a shape parameter close to 1. A book that has been in print for twenty years is roughly twice as likely to survive another twenty years as a book that has been in print for ten years. The effect is strongest for books that have already survived the initial period of low demand.
Technologies. The Internet Protocol (IP), first specified in RFC 791 in 1981, has survived multiple generations of networking technology. It is not the most efficient protocol. It is not the most modern. It has survived because it has been embedded in the infrastructure of the internet, and every year of survival has increased the confidence that it will continue to be used.
Ideas. Mathematical theorems are among the strongest cases. Euclid’s proof that there are infinitely many prime numbers, published around 300 BC, has survived every attempt at refutation. An idea that has been debated for centuries and not displaced is likely to remain part of the discourse.
Why the Lindy Effect works
The Lindy Effect is not a law of nature. It is a selection effect.
Non-perishable things are subjected to a continuous filter. Every year, some books fall out of print. Some technologies are replaced. Some ideas are disproven or abandoned. The things that survive are the ones that have proven their durability. Survival is evidence of fit — fit with the environment, fit with human needs, fit with the constraints of the medium.
The filter is not intentional. It is the cumulative result of millions of small decisions. Readers choose which books to buy. Engineers choose which protocols to implement. Each decision is local and short-sighted, but the aggregate over time produces a pattern: the things that survive longest are the ones that are most durable.
This is the same mechanism that drives natural selection. The Lindy Effect is a cultural and intellectual analog. Ideas and technologies are filtered by use over time. The ones that survive are the ones that are most useful. The mathematical reason the pattern holds is the heavy tail of the survival distribution. Most books fail within a few years of publication. A small fraction survive for decades. The heavy tail means that the oldest items are disproportionately likely to be very long-lived.
The limits of the Lindy Effect
The Lindy Effect is not a universal predictor. It has several important limitations.
The distribution must be power-law. The Lindy Effect holds only if the lifetime distribution follows a power law with shape parameter ε between 0 and 1. Consumer product lifetimes often follow a different distribution — one where the hazard rate increases with age. A phone does not become more reliable the longer it is used.
External shocks can override the pattern. A technology that has survived for centuries can be displaced by a single innovation. The telegraph survived for decades before the telephone made it obsolete. The Lindy Effect is a statistical tendency, not a guarantee.
The effect can create inertia. A technology may survive not because it is the best, but because it is embedded in existing infrastructure. A book may survive because it is assigned in schools, not because it is the best book. Survival is evidence of durability, but durability is not the same as quality.
What the Lindy Effect reveals about verification
The Lindy Effect is relevant to verification. A claim that has survived centuries of scrutiny is more likely to be robust than a claim that has survived only a few years of testing. This is not an appeal to tradition. An appeal to tradition argues that a claim is true because it has been believed for a long time. The Lindy Effect argues that a claim is more likely to be robust because it has been tested for a long time and has not been disproven.
The distinction matters. Religious doctrines have survived for millennia, but through institutional enforcement, not empirical testing. The Lindy Effect applies to things subjected to genuine selection — ideas that compete in a market of thought, technologies that compete in a market of use.
The Lindy Effect also has implications for confidence decay. Concept drift, discussed in a previous article in this journal, describes how verified claims expire when the underlying distribution changes. The Lindy Effect offers a related insight: claims that have survived long periods of testing are less likely to expire, because they have already been tested against multiple past distributions.
Practical implications
The Lindy Effect suggests a practical heuristic: when choosing what to read, study, or build, give weight to the options that have already survived. A book that has been in print for fifty years has passed a filter that most books have not. A programming language that has been in use for decades is not the newest, but it has proven its usefulness across multiple generations of hardware and software.
This is not an argument for conservatism. The Lindy Effect does not say the oldest option is the best. It says the oldest option carries information that newer options do not: the evidence of survival. Ignoring that evidence is a failure to use available data.