A textbook on the essential lessons of the Kobe earthquake and the one that struck the Northridge section of Los Angeles on the same date a year earlier would read something like this:
Earthquakes are unpredictable. They almost invariably strike not only at times but at places nobody expects, and no one quake is exactly like any other.
Designing or, worse, retrofitting buildings to withstand the tremors is extremely expensive. Nonetheless, recent efforts have been in some ways a heartening success--and in other ways a shocking failure.
Some of the most potentially effective precautions are relatively cheap and easy, and people could take many on their own. But they don't, or won't.
If those maxims sound obvious, contradictory, or both, well, the facts about earthquakes are too. Upheavals of the earth have stunned humans and ravaged their works since prehistoric times: some scholars believe the ancient Minoan civilization and the biblical cities of Sodom and Gomorrah were wiped out by quakes. Nonetheless, the tremors have never ceased to spring surprises on those who study them and try to cope with them.
Twenty years ago, that seemed to be changing. Breakthroughs in scientific understanding of plate tectonics--the incessant shifting of continent-size hunks of the earth's crust--spurred hope that major upheavals could be predicted. In Japan polls showed that 50% or more of the public thought they could be. Tokyo even established an Earthquake Assessment Committee of six eminent seismologists to advise the Prime Minister when he ought to issue a public earthquake warning. But in 17 years no such warning has ever been issued, and many experts think the $100 million a year Japan devotes to trying to predict earthquakes could be better spent on something else, like improving coordination among disaster-relief agencies to handle a crisis when it occurs.
Earthquakes, it turns out, have a lot in common with tornadoes: they are capricious beasts ruled by what physicists refer to as nonlinear dynamics, which means precise forecasting of when and where they will occur is impossible. In theory, major earthquakes should be preceded by smaller shocks. They are, but the earliest foreshocks may be so weak as to be hard to distinguish from background seismic ``noise.'' And for every small tremor that is followed by a big quake, others may not be followed by anything much.
In Japan, at least, scientists may also have been looking and listening in the wrong places. Japanese seismologists understandably have positioned underground sensors to pick up rumblings along the notorious faults that run under the Pacific off Japan; they are believed to be the source of the devastating 1923 temblor that killed 143,000 people in Tokyo and Yokohama. American scientists have kept a close watch on the San Andreas fault that runs for 650 miles through California from north of San Francisco nearly to the Mexican border. But the Kobe and Northridge quakes occurred not along these major inter-plate faults--cracks where continent-size plates grind against one another--but on intra-plate faults that spiderweb a single giant plate.