I guess "wiki for a research paper" would have been a better name, In any case, I first thought it referred to a paper wiki, which would have been a truly new idea, sort of in the tradition of "Hipster PDA."
The claim is that jumping "blindly into a wiki, and ... creating pages left and right ... building a page for every idea or piece of information, with few internal links," will not bring much benefit." Structure is supposed to be the answer.
I am not sure the proposed structure (and especially the limits on which level may link to which other level) ultimately makes sense. Creating pages without worrying about how they ultimately fit in and breaking down information into simpler parts seems to me to make for the strength of wikis. Of course, you then have to think about it, re-factor, establish links, etc. The index card method also required re-ordering and thinking before, or, perhaps better: while writing. Indeed breaking down the information according to the method "one fact, one card" made for part of its strength. Same thing with wikis.
But judge for yourself. It's an interesting idea, which would (or wouldn't) work with a personal desktop wiki just as much (or little) as with a hosted wiki.