Tsunami Caught Japan's Nuclear Industry Off Guard - NYTimes.com

Design is nothing more than siting, configuring, proportioning a system in such a way that it will perform according to needs in a safe manner. So everything that can counter effective and safe performance must be identified and designed against. Excellent engineering means, among other things, not missing nor underestimating potentially threatening outcomes. There is a useful discussion of assumptions made on tsunami threats to the failed nuclear plant in Japan in

Tsunami Caught Japan's Nuclear Industry Off Guard - NYTimes.com.

After any failure of this magnitude, it is common for statements to be made that everything was predictable. We saw this last year with the Gulf of Mexico accident. The reality of the matter though is that, at the design stage, engineers may always miss or underestimate the likelihood of a limit state (undesired system state caused by some event) in connection with some project somewhere, even in important projects like a nuclear power plant. This may happen because it was really difficult to foresee it, because of pressures of economical or political nature (which shouldn't belong in the world of engineering but nonetheless are there, and some may be susceptible to it) or just plain incompetence. I would say that, in connection with this type of project, a sort of "wishful thinking" state of mind, where the "unthinkable" is associated too low a probability of happening, is always the first suspect as a cause of failure.