It's not so obvious to me that it's a straw man argument (maybe my logic training is falling out of practice) so much as Guilt By Association. (∃x∈S:φ(x))→ (∀x∈S:φ(x))

The fallacy can be stated thus: "If there exists a project which has adopted a dynamic language and has failed, then all projects which adopt dynamic languages will fail."

Of course, this is not to say that the adoption of Python wasn't a symptom of this project's failure, or even a contributing factor, but it would take a lot more argumentative foundation to get away from the fallacy. However, even with the flawed opening paragraph, I'm not convinced that the author didn't salvage the argument by focusing more on the selection of dynamic (read: languages without good refactoring tools) languages as symptoms of a project's inability to focus on what's really important.

After all, there is no real silver bullet (the reference to Brooks' Law at the top of the article is unmistakable). One of the most important statements in the entire paper is: "The ease of employing meta-object protocols add an additional burden to understandability by making behaviors less traceable." If this argument could have been linked less to "dynamic languages," and more to "some of the people that use dynamic languages," then I would have been more convinced.

If there's anything we've learned from Fred Brooks and those after him, it's that, "It's the people, stupid."