I've said it before, and I will say it again: I regularly read theses. Maybe I am insane, but I find them rather enticing when they're really, really good. OK, only one thesis every 100 years can be as monumental as Shannon's thesis, but that does not mean other all thesis suck. I often find myself reading theses by former MIT and Caltech grad students. I like Math, CS and Engineering theses. Needless to say, I don't read bullshit theses on women's studies or other pseudo-intellectual junk.

Conference papers are limited to 6 pages or so. Authors need to cite the right papers to ensure they won't offend anyone. They need to revise prior work. They need to stick to the core of the idea and refer to 1000s of other papers. Theses have less editorial and political constraints, so they're much more enjoyable to read IMHO. And, fortunately, theses don't have that hideous double column format.

If I want to learn a new field, I don't read papers. I read theses.