The Lindy Effect

“The longer something has been in use, the more likely it is to continue being used.”

The Claim

For non-perishable things — ideas, technologies, books, institutions — life expectancy is proportional to current age. A book in print for 50 years is likely to be in print for another 50. A technology used for 100 years is likely to last another 100. A language spoken for 1000 years is likely to last another 1000.

This is the opposite of the intuition from biology, where older organisms are closer to death. For non-perishable phenomena, every additional year of survival is evidence of robustness, and the expected remaining life grows with age.

Why It Holds

The underlying mechanism is survivorship-based filtering:

1. Most ideas, technologies, books, fashions are fragile. They disappear within years.
2. The ones that survive to age N have passed many selection tests — changing tastes, changing markets, disruptive innovations, cultural shifts.
3. The longer something has survived, the more selection pressure it has withstood and the more robust it has demonstrably proven itself.
4. Future selection pressures come from the same distribution; the already-tested item is likely to survive the next one too.

Formally: if you know only the current age of a non-perishable item, the minimum-information (maximum-entropy) estimate of its total lifetime is twice its current age. Expected remaining life = current age. This is the Copernican principle applied to time.

In This Wiki

• A Bayesian prior. Given no other information, age is evidence of durability. This connects to bayes-theorem as the “update on survival” posture: every year something remains in use increases the posterior probability of its remaining in use.
• Contrasts with Amara’s Law. Amara says we overestimate tech impact in the short term and underestimate it in the long term. Lindy provides the filter: the tech that survives the short-term hype cycle and remains useful a decade later is the one worth betting on long-term.
• Munger’s style. “If it’s been true for a long time, it’s probably still true.” Munger’s bias toward durable businesses (Coca-Cola, See’s Candies) is Lindy-style reasoning: a company that has compounded for 50 years is a better bet than one that has compounded for 5.
• fallibilism compatible. Deutsch’s fallibilism says all knowledge is conjectural, but better conjectures survive more criticism. A long-lived idea has by definition survived more criticism. This is fallibilism’s test of provisional acceptance. Lindy is the heuristic for “which conjectures to take seriously.”
• Explains the preference for first-principles-thinking on old truths. The basic laws of thermodynamics, Bayes’s Theorem (1763), classical mechanics — all Lindy-validated. The newer the framework, the less evidence of robustness. This isn’t conservatism; it’s a calibrated prior.
• Applied to Munger’s mental models. The ~100 core mental models Munger recommends drawing from are disproportionately from old disciplines — Euclidean geometry, compound interest, evolution, classical economics. They have survived the longest and are most likely to continue paying rent.
• Relates to compound-interest. Both are about patient compounding — Lindy tells you which ideas/institutions/relationships will survive long enough for compound interest to work. You can only compound in a system that still exists.

Software-Engineering Applications

• Programming languages. C (1972), SQL (1974), Unix (1969), and Lisp (1958) are deeply Lindy. They are more likely to be relevant in 2050 than any language that launched in 2020. Bet on tools and skills with long tails, not the technology of the month.
• Protocols. TCP/IP, HTTP, SMTP, DNS — all Lindy. Whatever replaces them will have to survive a comparable selection process.
• Engineering practices. Code review, version control, testing — all old, all Lindy. Continuous delivery is young; its durable form is still being selected.
• Decisions about where to invest learning time. Fundamentals (data structures, operating systems, databases) are Lindy; framework-of-the-year is not.

Limits

1. Only applies to non-perishable phenomena. A specific human life is perishable (biological clock); the idea of “human” is Lindy. The principle is for ideas, books, institutions, technologies, laws of nature.
2. Fails in the face of disruptive shifts. Lindy would have predicted that horse-drawn transport had decades left in 1910. It didn’t. Technological discontinuities can collapse Lindy-validated systems in years.
3. Lindy is a prior, not a likelihood. Evidence of imminent decline overrides the age-based prior. A long-lived bookstore chain still goes bankrupt when the underlying economics fail.
4. Survivorship bias risk. Observing that “old things are durable” can ignore the vast cemetery of things that didn’t survive. The probability that something from 500 years ago is still around is high; the probability that any specific thing from 500 years ago is still around is low.

The Taleb Reframing

Nassim Taleb’s popularisation of Lindy (in Antifragile and elsewhere) sharpens it: time is the ultimate anti-fragility filter. Anti-fragile systems get stronger under stress, which is what long-surviving systems have been doing for their entire existence. The old is old because it has been improving under pressure.

This is a specific reading — most of what Lindy-validates is robust rather than anti-fragile — but Taleb’s framing is the reason the concept went mainstream in the 2010s.

Sources

• source—laws-of-software-engineering — in the Decisions cluster.
• Benoit Mandelbrot (1984) — formal mathematical treatment.
• Nassim Nicholas Taleb, Antifragile (2012) and Skin in the Game (2018) — popularisation.
• Named after Lindy’s Deli (New York), where actors observed in the 1960s that the career expectancy of comedians was proportional to their career so far.