Sir Martin Rees thinks we should be planning more for low-probability, high-consequence events

[bran]

I think he talks about it a little here. What do you think?

I'm not sure that it's worth the effort, or even possible, to plan for every eventuality.

like a zombie invasion?

Well, he's a scientist so I don't think he was considering things quite a low-probability as that. Although if govts planned properly for a high-mortality global pandemic then the same measures could probably be used against zombies.

I think it depends what he means by planning really. At work (in IT) we have to think through what could happen, then decide what steps are reasonable to take about it - in some cases, we don't actually do anything specifically about that risk, just decide that our planning for other more likely things would be acceptable if it happened.

eg we plan for a small fire in the server room, which might happen, by making sure we take regular backups and store them in another building. The same precautions would also be useful in the event of a plane crashing on the building, a small earthquake, a localised plague of zombies, etc.

But that doesn't mean we include "localised zombie attack" in our documentation....

He doesn't really specify, but I guess he is referring to end-of-the-world-as-we-know-it type events, like a big asteroid or Yellowstone park erupting. For a cheerful man he's fairly downbeat about humanity's chances of seeing out this century. He wrote a book about it.

I suppose to ensure the survival of humanity there ought to be different types of shelters in widespread parts of the world. There would have to be a collection of useful animals near enough to the shelter that they could be brought inside at short notice. Each shelter would have to be close enough to people to ensure that a large enough group of people are saved to prevent inbreeding, but also secure/remote enough to avoid being over-run by more people than can be accommodated. There would have to be a large seed store and enough food to feed everyone until they can start farming again. It would be sort of the equivalent of your data backup, as it would probably work equally well for many different events.

All of that probably wouldn't be too expensive, but it would only save a tiny proportion of humanity and tbh I'm not convinced that humanity is particularly important to save (although individual lives are important). Would it be a representative sample that would be saved, or just a mixture of scientists and high-ranking civil servants and politicians?

If he means that we should be thinking of ways to deal with the event itself, then that is hugely complicated and costly as each event is unique. Fending off an asteroid is entirely different to 'managing' a super-volcano. We could end up pouring enormous resources into some things only to have humanity ended by an unforseen and unplanned-for event.