The way Behe approached ID, through irreducible complexity, was arguably not science.
I think this was demonstrated on the witness stand at Dover when he was asked about the background research he had conducted. The attorney interrogating him was able to produce a huge stack of articles (50 or more if I remember correctly) that he was not familiar with. He had to admit, in essence, that he came to his conclusions without benefit of scientific research and in contrast to existing scientific knowledge.
That's pretty much the same as creation "science," from which ID is evolved.
True.
Because Behe set out to prove his hypothesis correct rather than attempting to objectively test it by means of prediction and verification (or as would be the case in this instance refutation).
However I think that the irreducible complexity hypothesis is a valid hypothesis in scientific terms. Just. It's derivation is admittedly dodgy (although ID of some sort has arguably been the prevalent hypothesis for the majority of human history) and the attempts of it's proponents to verify/refute it are undeniably woeful.....
But I think it remains a valid, testable (albeit failed) hypothesis in scientific terms.