First: How much research have you done into the physical process of getting a book printed? Meaning, who do you have to hire for printing and bindin? What materials do you have to provide to them (PDFs of each page?) Will it be hardcover or paperback? How do you choose the paper stock, etc? How much does it cost? From what I understand, people generally print books like this at "vanity publishers" who print extremely low runs with no real expectation of selling books on shelves.

Second, how many pages, what sort of content (all screenshots vs. screenshots mixed with text) how many levels per author, how many pages devoted to each level, etc.

My suggestion on the second question is to limit it to one level per author, and go in-depth on each level: 2-4 pages per level, with lots of shots and diagrams, and small paragraphs explaining the cool features (including notable gameplay such as a boss battle etc.) I'm thinking of some architecture books I have that do the same treatment with buildings. 

thanks for starting this thread up. I was going to do so when I woke up this morning(afternoon)

as for the physical process of printing - I was planning on tackling this the same way Hourences did using lulu.com. I assemble PDF files for the content and the cover.This limits some options definitely but saves from a extremely large up front investment. As for the paper stock, I think chocies are limited between b&w, and colour. Colour being alot heavier and suitable for photographs. Anyone who has Hourences book, I wouldn't mind hearing about the quality of the photos printed and the paper weight.

The final cost of each unit will be highly dependent on how many designers I can get involved. I'm unaware of any "vanity publishers" in my area, but if You know of any I'd like to take a look.

number of pages: again highly dependent on how many people I can get involved. As for content I personally want to keep the copy text very minimal, similar to most architecture/interior design type books. 2 page spreads are my current goal per map author, featuring at least 2 maps. But I do like your idea of going in depth this could work for some of the larger map projects definitely. *cough rubicon2?* :D 

bear: maps don't necessarily need to be complete/released/whatever.

generic: if there is a huge amount of involvment interest, then yes some cuts will have to be made. At this point however I'd like to have everyone I possibly can involved.