Let's be clear; I mostly agree with you.

But the pure mathematics of the situation does imply a slight increase in discovery. Suppose, for the sake of argument, that there's a 1/10000 chance of discovering that someone's 100ml bottles contain an explosive during a single screening event. If a Real Terrorist(TM) has to make five separate trips through security in order to get enough liquid through, the chance that they'll get through successfully is 1-((9999/10000)^5) = 5/10000. Yippee! We just increased security by a factor of five!

Now, if you want to argue about whether my 1/10000 ratio is an accurate estimate, or whether the increase in detection to 5/10000 is worth all the hassle, expense, and diversion from other screening activities, that's a perfectly reasonable discussion. All I'm saying is that the pure mathematics behind the scenario does imply a provable increase in security.

Also, my argument assumes that each screening event is completely independent, and that the Real Terrorist(TM) doesn't give any indication that they're going through screening multiple times. If they blow the secret, the chance of detection should increase appropriately. Sure, with a little bit of effort, this level of deception shouldn't be hard to achieve (print multiple boarding passes, use different checkpoints, etc.). But LEOs catch criminals all the time based solely on their stupidity.