How many lives would have been saved had people been warned about the giant tsunami off Sumatra? (Image from Thinkquest.com.)

Not all of them, obviously.

This was a once-in-a-lifetime event, a 9.0 quake (and remember the Richter scale is logarathmic) close to land that shifted the seabed 10 feet, displacing megatons of water in a wave that spread outward at 500 kph.

Banda Aceh was too close to the epicenter to use a warning, even if it came as the quake happened. (The quake hit the town, too, so there actually was a warning of a sort.) But the resort town of Phuket in Thailand could have gotten maybe 15 minutes warning, long enough for most swimmers to reach high ground, and Sri Lanka could have gotten even more warning. (It's what happened there that seems most tragic, since they barely felt the quake and could have have had over an hour's warning of the wave.)

The key to giving warning isn't having a seismometer. You can build a good seismometer for $34 that will could have detected the Sumatra quake from your living room.

The key is detecting a mega-event on the sea and broadcasting a warning across a wide enough frequency band to be followed by everyone who gets it.

It's a system integration problem, a communications problem, a service and maintenance problem, and while there's plenty of money to fund such solutions right now what about 20 years from now, when this quake is a distant memory?

It turns out that was the problem in this case. Indonesia had a system for detecting tsunami for many years, based on buoys, but abandoned it in 2000 when cash ran out.

So what we need is a worldwide endowment that can be invested to yield the money needed for maintaining a system beyond the point where we care anymore. And you need trustworthy people to manage both the system and the money.

Preventing the next short-term disaster, in other words, will take a very long-term vision indeed.