I think there are actually two "problems" going on at the same time. On one hand, open source software is "too difficult" for most people to use. I agree. But on the other hand, most people aren't putting that much effort into learning the software.
Is software supposed to be easy to use? As much as possible, I would think. But some "hacker" software that has a tremendous initial learning curve (TeX, Emacs, even Lisp, depending on your background) offers equally tremendous rewards over similar software that is easier to learn.
The reason, perhaps, is that in learning TeX, you are learning how to prepare documents at a more abstract level than Microsoft Word. The resulting documents may be largely identical, but TeX offers the user more power and control in exchange for taking the time to learn how to operate it. This is not necessarily a bad thing, especially if your job is to prepare documents. It may be a totally awful thing if you only need to prepare an occasional document, as the benefits of TeX are not likely to be worth the effort required to learn it for you.
The cost-benefit equation just doesn't work for "hacker software" as you say it. We're busy doing econometrics (we do use GNU R, but Matlab, Stata and SAS are far superior for most work) and statistical work, and just need to get reports done. We're trying to make money in here, not join a religion.
It's only human. I was a Haskell religious nut for a while. (Then I stopped programming and changed hobbies to skateboarding. But I'm studying modal logic from Konyndyk, so I'm still in the Haskell-ish geekspace).
Microsoft Word's UI is sometimes counterintuitive, but it's simple enough that people outside the hacker community can use it.
One problem with open source I didn't point out because the main idea was to focus on de Landa is that too many people thought competing with Microsoft had much to do with compatibility. If there was a working environment for collaborative writing as usable as Microsoft Word people would eventually go for it because it's cheaper.
I know, I know. I actually wrote my undergrad thesis on compatibility issues in software competition and how Apple was attempting a "partial compatibility strategy" by adopting Unix while keeping Macintoshness. But with time I realized that wasn't really the core problem, it was just a road bump.