Coders At Work is a series of interviews with programmers, people with successful backgrounds and some name recognition (to the extent people recognize programmers). Each interview is roughly 40 pages, and attempts to ask a standard set of questions in an attempt to document diversity and/or consensus. Most of the coders interviewed are older, both for the obvious reason that publicly acknowledgement of success and experience take time, and to cater to a celebrity history market.
The only guy who's close to my age is Brad Fitzpatrick, who shares a bit of culture with me; writing TI Basic games on long road trips and losing them due to battery failure. His stories of rolling in advertising banner ad money is one I've heard before. Probably the most interesting part of his entrepreneurship is how open source allows him to sell a website/company and then build new sites using the GPL'd assets he wrote when building the site up. Judging by his anecdote about selling FreeVote on the cheap just to be done with it, I wonder if something similar happened with LiveJournal.
The author has a bit of an obsession with Knuth and his volume of books, The Art Of Computer Programming. As best I can tell, Knuth doesn't invent anything but instead compiles published research into the above book. Since we keep graduating new PhDs but still have just the one Knuth, the series is unfinished and unfinishable. Annoyingly, many algorithms are named after him that he merely popularized, rather than invented. I suppose it's a worthy task to cut the jargon out of conference papers, as they can be really quite excruciating. The author's overweighting of Knuth comes in the form of asking every interviewee whether they've read Knuth's books and done any literate programming, leading up to a finale interview with Knuth himself.
Despite the author's inclinations, there are some good interviews in there. The Erlang author is interviewed, and after reading it I think I need to invest some time with Erlang. It's too bad I'm currently experimenting with Python/Django. It might be neat to do something web based with Erlang, but I'm wary of anything that decides it's easier to write a new httpd than implement an Apache module.
Some observations about the group interviewed: lots of compiler and language people, many had early access to research university computers, or later in computing history, programming jobs out of high school. (I don't even know how you find that kind of work as a kid). Most of the people I recognized are mainly famous for their non-coding activities; I'd wager more people have looked at TAOCP than have used TeX. Overall, I'd say these "coders" slant academic.
Coders at Work is a pretty good read. You can easily read just the interviews that interest you and not miss anything for it. If you work in the field, consider picking it up it; makes for good night reading material.