Carlos De la Guardia

The Lindy effect is not a universal law. It doesn't apply to technologies, scientific theories, or turkeys, all of which can die sudden deaths after long periods of growth and success. What (obese) people do depends on their ideas.

Take a population of people using 2019 calendars. In one year, will there be more people who are still using it, or who have switched to a 2020 calendar?

You’re comparing apple to oranges. A good comparison would be: given a population with a 2018 calendar in 2018, in 2019 are they more likely to have a 2019 calendar or to not have one?

(Lindy is about past behavior being an expression of an underlying quality which predicts future behavior)

A healthy-eating person eats junk one day. Will they do so the next day? It depends on why they ate healthy food before and on why they ate junk one day. A cheat day, or cheat month, could *increase* the number of healthy meals one eats in a lifetime.

“I shouldn’t eat this, but...” (devours delicious donut) In cases like this, there’s a conflict between one’s ideas, and no fixed algorithm can tell you which (if any) is right. Having failed to pursue some target behavior, I consider why, and may revise or reject the goal.

Another example: I’ve been overweight for 3 years in my youth. On the 4th year I got myself straight and am now fit. On 3 out of 4 years in which I was obese, I gained weight. 75%. You can see why “given obese, weight likely to increase” even if at the end I changed my behavior

That heuristic may be good for betting on the behavior of others, but it doesn’t follow that one must blindly adhere to prior goals, and never ‘skip a day’. In general, one’s ideas for how one’s future self should behave are riddled with errors that will need correcting.