[JumpinJack_Cash]

The obvious answer to that is being able to reliably detect asteroids on a collision course, and compute where is it gonna fall in order to proceed the emergency evacuation.

But first you have to be able to detect it, second you have to do it with reasonable time to save as many people as possible

[jacquesm]

Again, these people aren't stupid. You will have a very large amount of uncertainty based on a series of observations which essentially project a circle of probability on the planet that shrinks as more information from newer observations comes in and then just before impact that probability will either drop to zero because it is deemed to be a near miss (sorry, George) or it will then become a certainty. By that time it will be too late to evacuate. I can dig up the report from one of these committees if you want (I should be able to find it), they make for very interesting reading, it is a really nice example of science at work, even if the result is a negative. Anything more advanced makes for great special effects in movies and science fiction but isn't going to work. You'd have a very short time to move the entire population of a circle with a radius of a few hundred kilometers to outside of that circle starting from the center. The longer you waited the fewer people you'd have to move but the bigger the chance you'd be too late.

[vkou]

Any links to these reports?

[jacquesm]

Found it:

https://www.boulder.swri.edu/~bottke/Reprints/National_Acade...

Note that Arecibo is now gone.