nLab entry, check this and also this
Conrad: Cohomological descent (hypercover.pdf)
http://ncatlab.org/nlab/show/descent+for+simplicial+presheaves
http://ncatlab.org/nlab/show/hypercompletion
http://ncatlab.org/nlab/show/Carlos+Simpson
Descent: See Borceaux vol2 chapter for a categorical viewpoint.
Deglise uses something for rigid cohomology
Guillen and Navarro Aznar uses something called a “descent category” in their “extension” paper.
Friedlander, Walker: Semitopological K-theory of real varieties. Proves Nisnevich descent! Review says that the corresponding fact for alg K-th was proved by Thomason.
Giraud: Technique de descente, 1964.
R. Street, Descent, Oberwolfach preprint, mentioned by Toby Bartels at nLab
rectified infinity-stack at nLab
A must-read for descent: Toen: Thm de RR pour les champs de DM. File Toen web publ rrchamp.pdf. Discusses K-theory and RR for DM stacks, but also various notions of descent, including basic stuff on simplicial presheaves, homological descent, and etale descent.
For the notion of descent data, see Toen course in cours folder under Toen, chapter 5. Also in Vistoli maybe.
Possibly interesting: Descent theory for derived categories. Alexei D Elagin 2009 Russ. Math. Surv. 64 748-749
Email from Peter, July 09:
I stick to the slogan: descent means glueing is respected… what ever that means. So it is some sheaf condition. Somehow descent for a functor (aka this functor being a (higher) stack) means, after Toen and many others, that F(X) = holim(cech simpl. object of any (hyper)covering of X). See the notes, hopefully… For seeing how Galois descent is an instance of that I found the relevant chapter in Waterhouse illuminating - I don’t claim I understood it but it seems to put one on the right track.
Brown rep.: After Neeman’s article (“Brown rep. and Grothendieck duality via Bousfield’s techniques”) Brown rep. holds for compactly generated triangulated categories which SH(Z) should be; I think connected schemes are generators for sPre(Sm/k) - every presheaf is a colim of Homs, i.e. of schemes (and non-connected schemes are coproducts of connected sch.s), every simplicial set is a colim of its simplices, now one would have to combine these, maybe one needs additionally the constant sheaves with value as generators (?). Then hence they would be generating for SH(Z) (localizing only makes them a “bigger” subset of all objects). Intuition says that these are compact in SH(Z) but I know no proof.
One can shift the difficulty thus: Neeman proves in his book that Brown rep. holds for “well-generated” categories (see Krause for a definition) and Rosicky says in the introduction to his article that he shows that combinatorial model cats give well-generated triangulated cats. Now here the Brown rep. theorem is probably more difficult but it is easy to see that SH(Z) falls under its scope. Roughly: Combinatorial = accessible i.e. everything is filtered colim of presentable objects, i.e. objects s.t. Hom out of them commutes with filtered colims. This is clearly the case for sPre(C), C any cat: It is a functor category, namely and in such everythin is filtered colim of representables…
A bibliography:
nLab page on Descent