Build Your Own Weather Channel


Theoretical computer scientists spend a lot of time thinking about whether computations are feasible or not. Personally, I'm much more interested in how seemingly innocuous bits of technology can suddenly make entire classes of problems feasible, where before they could only be solved with heroic effort.

For example, I can lift 100 pounds, even 200 if I strain, but not 300 (I know, I've tried), and certainly not 500. Somewhere in there, a quantitative difference becomes a qualitative one, and the task goes from feasible to infeasible. Conversely, though, if you start with 500 pounds of bricks, and keep taking bricks off the pile one at a time, I'll eventually be able to lift what's lift.

Which brings me to this website, the "Experimental National Digital Forecast Database XML Web Service". Go back more than a century, and no-one had ever had this kind of data: not Casear, not Isaac Newton, not Napoleon or the tsars or the emperors of China, no-one. Sixty years ago, during the Second World War, a handful of military commanders could occasionally see a vague (and often incorrect) outline; twenty years ago, you could get month-old data on tape for several hundred bucks a throw. Today, you can get it all, in real-time, on your desktop. Never mind building your own weather channel—you can integrate tomorrow's weather into your little league baseball team's game planner. "All" we've done is move information around at ever-increasing rates, but that means that millions of people can now do what only a handful could before.

Which doesn't make up for having a CSC407 exam from 7:00-10:00 p.m. only a day after the end of classes, of course, but I hope it helps a little…