I'll address these, assuming they refer to the same argument: namely, that a system not at 100% is worth nothing.
First: If the GMD (someday) has a 99% accuracy, do you think that out of 100 missiles, one actually WILL get through? That's a statistical fallacy. There is a 99% chance of stopping all 100 missiles. Besides, how do you know that 100+ missile will be fired? Maybe it will be a salvo of five missiles. Maybe it'll be one missile seized by a Fart East Terrorist group. (Al-Qaida operates there, don't they?) There are two problems here: one, you're playing with statistics, two, no one definitely knows how an attack will happen.
Also, there's no guarantee that if we were hit by one (or more) nuclear weapon, that we would fire nukes back. The fact that there is a chance of failure doesn't mean to not try.
Also, reducing the disastrous IS useful. Look at government's environmental policies. Can we eliminate greenhouse gas emission all at once? Is reducing the threat beneficial in missile defense? Indeed. Whether or not the launching of nukes against us creates retaliation from us in kind, the need for protection, ANY protection, is very important. Say North Korea launches nukes, then we obhliterate it, either with nuclear weapons or conventionally, and then the worst is over. Say we stop 35 out of 36 nukes. Say we lose 1 million people instead of 40 million. Isn't that better?