This debate took place around the time of the Dover trial, and Miller included a supposed refutation of Michael Behe's argument vis a vis the irreducible complexity of the bacterial flagellum. Miller showed an exhibit which demonstrated that you can remove a whole bunch of parts of the flagellum, i.e. strip it all the way down to the base, and what remains looks like the Type III Secretory System. He then quoted Behe who said that if you remove a part from an irreducibly complex system, it is by definition non-functional. Behe must be wrong, Miller assured us, because he removed a whole bunch of parts from the flagellum and what remained closely resembles something that's known to be functional.
It was a clever trick. I could take away all the parts of a mouse trap until all that's left is the base, then use the base to prop up an uneven leg on a table so it doesn't wiggle, and then claim that the mouse trap isn't irreducibly complex either, because that base has function all by itself. When Nelson pointed out that Miller misrepresented Behe's argument, Miller rejoined that he did not, because he quoted Behe directly. Surely Miller knows that one of the most common ways of misrepresenting someone is by quoting the person directly, while leaving out clarifying material that the author has offered. That's what Miller did.
Behe has pointed out about 10,000 times that you can use the various parts of a mouse trap to do other things, but when you take away one of the parts it's no longer a functioning mouse trap. The same applies here. Behe isolated a specific function, motility, and pointed out that if you take away one of the parts of the flagellum you loose that function.
Once Miller was forced to address Behe's complete argument rather than the denuded straw man he came to the debate prepared to tackle, the power of his refutation was thoroughly diminished, and could be likened to a balloon that had just been introduced to a pin, to wit (Miller's response):
"Now the type III system doesn't have the function of flegellar motility, but intelligent design people use this idea of 'irreducible complexity' to explain why these machines couldn't evolve. If you say, 'Well, this system only does protein secretions, and that system only does surface recognition, and this system only does signal transduction,' you know what you're doing? You're giving away the store, because you're explaining these systems could evolve, because first we evolve this part, then we evolve another part, then we evolve a third part, and the whole function comes from the totality of parts."
"...first we evolve this part, then we evolve another part, then we evolve a third part, and the whole function comes from the totality of parts."