In 2005, a control room for the A and C subway lines in NYC caught fire. "No larger than a kitchen," the room held 600 relays, switches and circuits that keep track of trains and keep everything running. Officials originally thought it would take three to five years to get the lines back to normal capacity. (Thankfully it didn't.) The epic repair time was because the fixed-block signaling system dates back to 1904 and only two companies in the world were able to repair it, one in Pittsburgh and the other in Paris. This is technology's trailing edge, according to Peter Sandborn in IEEE Spectrum: the huge, crippling problem of obsolescence.

Three percent of all the electronic components in the world become obsolete every month. When you imagine all the shit coming out of China, it's pretty staggering. The problem is actually worse for the military, which spends about $10 billion a year on keeping up obsolete electronics parts. Ironically it's because they've switched to using off-the-shelf consumer electronics for 90 percent of their components—with a much shorter service life, four years at best—rather than "military-spec" gear, which was designed to hang around for a decade or more.

IEEE Spectrum lists a couple of egregious examples: The B-2 Spirit, one of Jesus' favorite planes, started flying in 1989, and by 1996, lots of its electronic components were obsolete. And in the Navy's new sonar system, 70 percent of the parts were obsolete when they started installing it.

Finding the parts isn't just difficult, it's expensive as hell, so the cost of maintaining obsolete but very necessary wares basically keeps you from upgrading. In the NYC subway case, instead of moving to a new, modern computerized system that would probably be cheaper in the long run, the Metropolitan Transit Authority has had to focus its limited budget on maintaining the frail, antediluvian network, trapping New Yorkers into an transit system light years behind, say, Japan's. (There have been stories in the recent past about the subway's upgrades, but they have mainly been superficial.)

Not all of you depend on the subway, or fly B-2 bombers, so here's a closer to home example: Windows vs. OS X. The latter is lighter, faster and springier, because it dumped all of the Classic OS's code. A fresh start, with a transition eased by the Classic emulation scheme. Windows Vista, on the other hand, is burdened by 20 years of legacy code, code that it could be argued is essentially obsolete. So we pay the price with a bloated operating system that struggles under its own massive girth. Dumping all that dead weight for Windows 7 and starting fresh—while painful—would be the best thing Microsoft could do. But it's not that easy, or they'd have done it, obviously. Maybe. You got any better examples of painful obsolescence? [IEEE Spectrum, NYT]