While I agree with your point, the author's argument does not allow for rockstars and slackers - all programmers have been reduced to the mythical $10k/month level (i.e. man-month is the measure of all things).
Actually, the large differences in individual productivity have more to do with the slackers than with the rockstars.
Some programmers, in some environments, are negative net productivity programmers. If some programmers literally get nothing accomplished at all, then you can easily say that "some programmers are 10^3 or 10^6 or 10^100 times more productive than others"
Another factor in "rockstar" productivity is that you need to get all the bull out of the requirements gathering process. I've seen so many places where it takes six hours of time working with "customers" to figure out the requirements for something that takes 6 minutes to implement. Tasks that are absolutely trivial can stretch on for weeks if your organization can't be decisive about getting them done.
Of course, in large teams there's a lot more bull to go around.
It's one of the things the author misses, at a certain scale the requirements are such that gathering them and managing clients requires more personnel than the actual writing of the code.
And a thirty person team is not only more likely to have several interns, but also people who spend a significant amount of time training, mentoring, and managing them.
