Just to chime in on the "purity" of scans, long after the fact--and I apologize in advance if I bring the argument back to the surface.
Basically, I don't mind reconstructed books, except for a couple of things. I think it's kind of a waste of effort and bandwidth to produce, upload, and download it when you can more easily just post a comment that "hey, the missing pages are probably here." Also, since I kind of blindly download everything and sort out the details later...well, that bad organization is nobody's problem but my own, but it's annoying to find out that I have two or three different versions of the same book and need to guess at what's going on.
By contrast, I also don't mind partial books. I'd like as much of it that's available, of course, but I'll also take what I can get.
As for the research issues...well, there are sources that claim more credibility than we do hawking worse information. I understand that it's nice when everything is perfect, but to believe that a (really, no offense to anybody) bunch of random scanners are producing research-quality materials is sort of naive. I'd prefer it if everything were clearly documented, but I also would never expect it.