It's "better safe than sorry".

No, its “the occurrence of a rare event is an indication of the existence of the conditions in which that event occurs in that time and place, so if you have evidence that such an event has recently occurred, you should treat the place where it did as a place where that kind of event has an elevated likelihood of occurring”, which is far more specific and actionable than “better safe than sorry”.

This is a lovely example of HN pedanticism.

The English word is pedantry, not pedanticism.

It's not pedantimums to complain that you're swapping in a completely different moral.

I think part of the idea for the metaphor is that, if you see lightning hit a tree, you shouldn't think it is so rare that it wouldn't hit the tree a second time (as some people then think "lightning never strikes the same place twice")... it is, in fact, so rare to have hit anything at all, that it having hit that tree in the first place is an indication that that tree is acting as some kind of preferential local lightning rod and it is highly likely to hit the same tree again.

https://www.theweathernetwork.com/ca/amp/news/article/can-li...

> Believe it or not, this long-held myth is far from the truth. While the odds of being struck by lightning are low, the chances of lightning striking the same place twice are high. Lightning can, and often will, hit the same spot multiple times.

Right, but this is different than if you see a tree that has clearly been hit by a bolt.

That is, seeing the strike shows that things are primed at the moment. Seeing a singed tree is not. (That make sense?)

That said, I can see how that still fits the idea.

I think other people have sort of answered it, but a more general version might be:

If you observe an extremely unlikely event, you should update your Bayesian prior about how likely that event is.

Makes sense. I was stuck with the idea that "seeing a car on the side of the road" is closer to seeing a tree that had been struck. Actually seeing it get struck would be like seeing the car actively crash.

Close in thought to each other, but still different things. Granted, I think your point still fully holds. There is a decay function for the likelihood of repeating events. And if you don't have evidence that what you are looking at is old, much safer to assume it is not and that the circumstances are still primed for it to happen again.

Repeating and periodic are not the same. Seeing a singed tree or a spun out car tells you nothing about how frequent these events are. They do tell you this place is likely to see those events, at any rate.

It's not the time, it's the unknown/unseen factor that's making it likely or unlikely. You think you are seeing lighting striking a tree. But maybe what you are actually seeing is lighting striking a lighting rod.

Lightning will not strike a TREE again because usually it destroys the tree in some way, changing it's height. But for many other things (buildings, mountains, etc), lightning will happily strike it over and over again.

Is the car in a ditch for a truly unlikely reason, or there a hidden lightning rod there? Perhaps you shouldn't assume a seemingly unlikely event won't happen again.

* * *