E.P. wrote:

"It would probably cost several billions, that is to say 1,000's of 1,000,000's, to build a lunar observatory. But for the last 5,000 years, about every 1,000 years or so a small asteroid or comet has hit the Earth and destroyed most of a continent. The value of a continent, which is to say its cost, assuming you could buy one, is 1,000,000's of 1,000,000's. "

Nick wrote:
"If we only have 5 continents now, how many were there 5,000 years ago?

What will we do in 5,000 more years when they are all gone?

How could one man be so right and the rest of the world so wrong?

Rest easy McFly, EP has miscalculated by several orders of magnitude. There IS a danger from impacts, but the solution is not to build a tin umbrella to shelter under, but instead to diversify away from the Earth.

EP does not agree with this, but it's a free world... "

Hmm, I don't think Ed has miscalulated by several orders, but nor do I think he is accurate. Let's see, first off the north American continent is certainly easily worth trillions of dollars in real estate improvement value- in California alone it's in the trillions. So you could say "on average" each continents worth trillions and be "close enough".
An continent getting hit every 1000 years isn't very accurate- but it's more problematic in that it is misleading. For instance, if you simply divided the land area into 5 equal areas you would almost have 12 regions of oceans which are the same size in area. So for every rock hitting a continent 2 1/2 are hitting the ocean. But at the same time if the rock is large enough, a one hitting an ocean could affect more than one continent- and damaging higher property value along shorelines (and higher population density-and not to mention crippling ocean transportation infrastructure).
So, what I think is interesting and where Ed might be pretty close is the potential dollar loss due to impactors- trillions of dollars per thousand year worldwide is, I believe, a fairly conservative number. I would find interesting if we could get a more accurate number- I would guess it's closer to 10's of billion per year- on average- worldwide- and the US would probably share a much higher share of this cost [for a number of reasons].
-gb