Even small genetic changes of this size can be counterproductive. Example: “If even one amino acid in the sequence is changed, that can potentially change the protein’s ability to function. For example,sickle cell anemia is caused by a change in only one nucleotide in the DNA sequence that causes just one amino acid in one of the hemoglobin polypeptide molecules to be different. Because of this, the whole red blood cell ends up being deformed and unable to carry oxygen properly.”

Of course any single mutation can be highly deleterious. But that is taken into account in the Behe article and in any experiment involving rates of mutation and time frames. Most point mutations will change a single amino acid in a protein sequence and knockout experiments tell us that most mutations are fairly neutral, that you could replace a sizable portion of the amino acids with another amino acid, one at a time, without significantly impairing the protein’s function. But there are certain key areas in the sequence – and a binding site is typically one of them – where mutations will tend to be harmful.
All of that is taken into account when doing this type of time frame experiment. Only a small percentage of the mutations will be beneficial or neutral. For a change to be preserved, it need only not be fatal. If a point mutation is neutral, as it will be even in a key binding site a small percentage of the time, it can be preserved and passed on to await the next point mutation. That is exactly what Behe’s computer simulation found, that even without any beneficial preservation of intermediates and even with wildly unrealistic assumptions to skew the experiment against it, the results show that an irreducibly complex system can develop in a relatively short period of time.

I always find myself wondering what they consider irreducibly complex now, what they would have considered irreducibly complex 10 years ago, 20 years ago, etc.

This is a very good point, one I make often in this dispute. It’s especially appropriate to ask Michael Behe because there are so many complex biochemical systems (those requiring the interaction of multiple parts for function) for which he accepts an evolutionary explanation. But at some point, a scientists researching that particular system came up with a plausible explanation, hypothesized the mutations that may have taken place, performed an experiment (usually a knockout experiment) that validated his hypothesis and published a compelling explanatino for the developmental pathway for that particular system. And at any time up to that point, Behe could and would have made the very same argument regarding that system as he does today about the flagellum or the blood clotting cascade. And the argument would, by his own admission today, have been wrong. So why is it any more compelling when applied to systems whose development is not currently understood and explained thoroughly? Not because there is an objective difference between the two systems, but because this truly is a god of the gaps argument.

http://biology.clc.uc.edu/courses/bio104/protein.htm