So the system is useless, as far as actually defending against missiles. It is, however, very useful for other purposes:
The system has been in development since 1946, and so far it has achieved a grand total of nothing. You wouldn't know it if you read the press releases published by the Pentagon's missile defence agency: the word "success" features more often than any other noun. It is true that the programme has managed to hit two out of the five missiles fired over the past five years during tests of its main component, the ground-based midcourse missile defence (GMD) system. But, sadly, these tests bear no relation to anything resembling a real nuclear strike.
All the trials run so far - successful or otherwise - have been rigged. The target, its type, trajectory and destination, are known before the test begins. Only one enemy missile is used, as the system doesn't have a hope in hell of knocking down two or more. If decoy missiles are deployed, they bear no resemblance to the target and they are identified as decoys in advance. In order to try to enhance the appearance of success, recent flight tests have become even less realistic: the agency has now stopped using decoys altogether when testing its GMD system.
This points to one of the intractable weaknesses of missile defence: it is hard to see how the interceptors could ever outwit enemy attempts to confuse them. As Philip Coyle - formerly a senior official at the Pentagon with responsibility for missile defence - points out, there are endless means by which another state could fool the system. For every real missile it launched, it could dispatch a host of dummies with the same radar and infra-red signatures. Even balloons or bits of metal foil would render anything resembling the current system inoperable. You can reduce a missile's susceptibility to laser penetration by 90% by painting it white. This sophisticated avoidance technology, available from your local hardware shop, makes another multibillion component of the programme obsolete. Or you could simply forget about ballistic missiles and attack using cruise missiles, against which the system is useless.
Makes a fair amount of sense actually. Thanks to Frustrated Mess in this babble thread for the link (also being discussed in this Kitco thread).
So why commit endless billions to a programme that is bound to fail? I'll give you a clue: the answer is in the question. It persists because it doesn't work.
US politics, because of the failure by both Republicans and Democrats to deal with the problems of campaign finance, is rotten from head to toe. But under Bush, the corruption has acquired Nigerian qualities. Federal government is a vast corporate welfare programme, rewarding the industries that give millions of dollars in political donations with contracts worth billions. Missile defence is the biggest pork barrel of all, the magic pudding that won't run out, however much you eat. The funds channelled to defence, aerospace and other manufacturing and service companies will never run dry because the system will never work.