The Lindy Effect

How can you estimate which technology might remain relevant in the future? How can you prioritize which books to read?

The Lindy Effect, from Taleb’s book “Antifragile,” can help you.

The Lindy Effect: what is it?

The Lindy effect, sometimes also called “Lindy’s Law,” relates age to life expectancy.

For people, every year of life DECREASES their remaining life expectancy. A 70-year-old is expected to live 14.4 more years, and a 71-year-old is expected to live only 13.7 more years. One year of life has reduced life expectancy by 0.7 years.

Conversely, for ideas and technology, every year of life INCREASES their life expectancy. For example, books on the NYT bestseller list remain there for only an average of 5 weeks. However, a book that reaches the 5-week threshold is expected to stay there for more than 5 weeks. The longer a book is on the NYT bestseller list, the longer it is expected to stay.

In Antifragile, building on Mandelbrot, Taleb describes the Lindy Effect as follows:

What justifies the Lindy Effect?

The older something is,

• the more conditions it must have been fit for,

• thus, the broader range of possible futures it is fit for,

• thus, the longer it is likely to survive,

(in the absence of bounds such as senescence).

Taleb also presented a statistical justification for the Lindy Effect in his books. I won’t cover it here, as I try to keep the understanding of the Lindy Effect intuitive.

Perishables and non-perishables

The reasoning above doesn't apply to people, as senility poses a natural limit to the maximum age they can reach. An 80-year-old person cannot survive another 80 years.

The Lindy Effect mostly applies to entities with no natural boundaries to life expectancy, such as technologies and ideas. For example, it applies to books, movies, and technologies like bicycles (but not necessarily to objects subject to decay, such as a bicycle).

However, the applicability of Lindy based on the criterion of perishable/non-perishable is not as clear-cut as it seems. For example, Lindy doesn't apply to adults but does apply to babies. A baby that survives its first week has a considerably longer life expectancy than a newborn. Therefore, we can say that the Lindy Effect applies to perishables, but only when they are far from natural limits such as senility. As an entity approaches its natural limits, decay dominates Lindy (more on this later).

The Hazard Rate

For non-perishables, such as objects and ideas, the main determinant of life expectancy is the hazard rate (the chances of dying/disappearing at age X).

When we observe an object’s life, we can use Lindy to estimate its life expectancy or hazard rate. For example, we can estimate a book’s life expectancy on the bestsellers’ list (its life expectancy) or its chances of dropping off next week (its hazard rate). Of course, the two are negatively correlated.

That said, we can reason the following.

The older something is,

• the more conditions it must have been fit for,

• and thus the broader range of possible futures it is fit for,

• and thus the lower its hazard rate.

Our estimate of an entity's hazard rate decreases as time passes without that entity disappearing.

The first keyword is "an entity's." A book staying for months on the NYT bestsellers' list does not mean that all books on it are less likely to drop off next week. It just means that that specific book is less likely to disappear.

The second keyword is "our estimate." The book's hazard rate does not decrease over time; its hazard rate is probably constant. Instead, it is our estimate that decreases. The longer the book survives, the more reasons we have to lower our hazard rate estimates.

The hazard rate for perishables

We previously saw that Lindy applies to perishables, but only when they are distant from natural limits, such as senility. Now that we know about the hazard rate, let’s clarify this sentence.

We can decouple the effects of Lindy and of decay into multiple hazard rates that we can aggregate together to obtain an entity’s total hazard rate. For example, a person’s total hazard rate is made of:

• The hazard rate from accidents (subject to Lindy; the more a person survives, the more we can suppose them to be cautious, and thus, the lower our estimate of their hazard rate from accidents).

• The hazard rate from illnesses and internal conditions (e.g., stroke) is a component not influenced by genetic causes (this increases linearly or exponentially with age).

• The hazard rate from illnesses and internal conditions is a component influenced by genetic causes (subject to Lindy – the more a person survives, the less likely they are to have genetic conditions).

The total hazard rate of a person is the sum of the three points above. The second one becomes dominant as one person approaches the natural limits of human longevity. Hence, it’s not that Lindy does not influence the life expectancy of perishables – it does, but it loses relevance over time.

The Lindy Effect, generalized

Lindy applies not just to time, but also to other dimensions: space, cultures, uses, conditions, etc. Here are a few examples of practical applications.

Continuing the NYT bestseller example, a book sold in one country might be successful because it’s a great book or because it discusses something very relevant to that country.

Once it’s translated and does well in another country, the odds that it’s a great book increase.

In general, the more geographically widespread something is,

• the more conditions it must have been fit for,

• thus the broader the range of conditions it is fit for,

• thus the lower the estimate of its hazard rate upon entering a new geography.

I suppose the same works across cultures, use conditions, and most dimensions. (Remember the limitation that “estimates made by the Lindy Effect are subordinate to intrinsic limits.” For example, a book read in 150 countries is not likely to be read in 150 more countries if there are only 200 countries on Earth.)

For example, bicycles are Lindier than cars. Not only are they expected to be around for longer, but they can also be used in a wider range of conditions (off-road, in the absence of fuel) and can be built/repaired by more people with less specialized tooling.

Therefore, we can often use the Lindy Effect to estimate not only life expectancy but also usefulness, relevance, and maintainability across a wider range of conditions or use cases or skills, etc. (again, a reminder: it is probabilistic, not deterministic)

Before closing this essay, I have two more remarks.

What the Lindy effect is not

The Lindy Effect estimates an entity’s hazard rate, not whether that entity is good or bad. You can’t say, “It’s Lindy, therefore it’s good.” Mosquitoes are Lindy.

Second, being Lindy doesn’t mean that something cannot disappear tomorrow. It only suggests we have reason to believe it is less likely to disappear than if it hadn’t been around for so long.

The Lindy Effect doesn’t tell you how long something will survive. It helps you estimate its hazard rate or life expectancy, both of which are probabilistic.

Lindyness, what is it?

Lindiness is the property of being Lindy, in other words, of having been around for a long time and, therefore, being expected to be around for a long time from now.

It only applies to the non-perishable (e.g., ideas, book contents, technologies, songs, etc.) and carries no moral valence.

Its use is to estimate whether an assumption will still be relevant over long time horizons.

What are some examples of Lindy?

Some examples of things that are Lindy:

• Books
• Songs
• Ideas
• Technologies
• Recipes

Some examples of things that are not Lindy:

• Food
• People
• In general, anything with a bounded life expectancy

Further readings

I first learned about the Lindy Effect in Nassim Nicholas Taleb’s Antifragile, a book I strongly recommend. In this piece, I’ve shared some thoughts on the process behind it and how we can apply it to more use cases.

Much like this essay, my book on Ergodicity simplifies a complex concept related to survival, making it practical.

Conclusions

• The Lindy Effect: For ideas and technology, every year of existence increases their life expectancy.

• The Lindy Effect also applies to perishables, but only when they are distant from their natural expiration.

• The Lindy Effect is not deterministic but probabilistic; it does not tell you how long something will survive, but helps you estimate its hazard rate or life expectancy.

• The Lindy Effect does not tell us whether something is good or bad.

• We can often use the Lindy Effect to estimate not only life expectancy but also usefulness, relevance, and maintainability across a wider range of conditions, use cases, skills, and so on.