2) even if we have Lee Kuan Yew managing Sichuan for the past 40 years, this disaster cannot be avoided (but damage should certainly be reduced, though the amount is unclear)
3) the art of earthquake prediction is still in its infancy. Even the 'warning' of Densmore et al. (pdf of the original article here) only managed to highlight that Longmen Mountain is a high risk area.
What should the government do? issue warning? enforce evacuation?
The best solution is perhaps to rely on the market. Let the insurance (to be precise, re-insurance) companies take the risk. The governement may require the builders and owners to pay for insurance premium. The insurance companies will then choose to fund which team of scientist, for how much, and what to act upon receiving a warning. Before we find a reliable (i.e. repeatable) earthquake prediction methodology/algorithm, we will have to rely on fuzzy logic and statistics/probability. Re-insurance companies will compete with each other, and only the fittest (i.e. those which could avoid loss of lives/properties at less cost) will have the competitive advantage.
The insurance companies are perhaps the