But not necessarily fatal before the organism reaches breeding age -- and nature has a lot of individuals with which to experiment.
So you're suggesting that a species can evolve from A to C by way of an intermediate stage B that has less survivability than A, but leaves it possible with luck to reach stage C before it dies out?
"B" has a survival advantage over "A" (after all, it was a small change in the DNA that got from "A" to "B"). "C" never entered into it.
Your statement that clotting too much or too little is "not necessarily fatal before the organism reaches breeding age" is considerably weaker than "has a survival advantage over" and does not imply it. It is by no means clear that too much clotting has a survival advantage over too little, or vice versa.
The clotting cascade started out relatively simple and weak (like the type found in invertebrates). A little stronger would be a survival advantage for the weak cascade. As the cascade gets progressively "stronger" the occasional individual will not have as strong a cascade as the population in general (think hemophiliacs). This is not necessarily fatal.
I don't see where that's a problem, since he was making two separate points, in response to two separate issues.
That's not what he said, and in any case you're oversimplyfing the issue.
First, it's hardly a "choice" between "too much or two little". Instead, the path could have been from "way too little" to "too little" to "a little too little" to "just right", for example. In that case every step would be an improvement over the last, without ever having to contrast "too little" against "too much".
Second, even if the pathway did oscillate around "just right" before homing in on it, "too little" and "too much" are hardly binary conditions. They exist on a continuum, and something along the lines of: 50% too little -> 30% too much -> 20% too little -> 10% too much -> just right would again be an example of continuous improvements.
Finally, the biggest thing that you're missing is that the clotting system can, and almost certainly did, evolve over periods where the requirements it had to meet were changing. What is "not good enough" today may have been close to "just right" back when the modern vertebrate clotting system was in an earlier stage. Thus it's a mistake to try to ponder whether the system(s) which were precursor(s) the modern vertebrate clotting sytem might be workable *in modern vertebrates*, since its precursors would have actually arisen in amphibians, and before that fish, and before that notochords, and before that invertebrates.
In short it's a fallacy to consider only whether a more primitive form of the clotting system would have been workable in modern vertebrates, since that's *not* where the earlier forms of the clotting systems evolved.
Come to think of it, this is yet *another* mistake that Behe makes, both in his examination of the vertebrate clotting system, *and* in all of his other "IC" analysis -- he never takes into account the fact that evolutionary precursors might have been "built" in *very* different conditions (i.e., in very different ancestral species). Behe only examines whether removing a component of the system would "break" it *in the modern species in which it now resides*. That quite simply is not a valid test.