Amit is the author of Mac OS X Internals, a book eclipsing the 1600 page mark (according to Amazon) making it one of the longest technical books I've ever seen. He spent some talking about his experience writing:
- no editors checked on him because most technical authors give up
- eight to nine chapters were cut from the already monstrous book
- he claims to not be an author and has no interest in writing another book
- he spent months at a time writing most of which he found all-consuming
I didn't find much value in the technical content of the talk (not because of Amit but because I'm rarely in a position to need details at this level of the OS and I do have some Mach background from NeXT days) but the feedback on authorship was worth attending to confirm what I already suspected. I've been approached about book writing before but have balked because of the time commitment and lack of financial reward -- I find writing technical articles and blog posts time consuming enough.
He did have some gems I happened to note (these are not direct quotes but close):
If not too much work then it's crap and there's no need to write crap.
About the mystique of kernel
hackingdevelopment and crashing the kernel:
No one wants to write code that crashes. If you crash the kernel, reboot the machine and fix the bug.
About work in general:
No matter how cool the work is it's always boring so I'm always looking for something else to do.
Boy did I empathize with this one. In his case, and mine, this means working on open source software.
About releasing MacFUSE and the biggest surprise:
I had released this very difficult software and the biggest surprise came from people complaining about the disk icon!
An audience member offered it must have been because there were no bugs and therefore nothing else about which to complain.
Only at the end of the two hours did appear the ever-so-gentle "Google is great" mantra and it was more in the tone of "Google is very good about allowing open source development and utilizing their extensive reach to make the project successful". This can't be discredited in my opinion as marketing an open source project can be very difficult and Google's backing would certainly help.
I'm happy to have gone, I got to catch up with an old friend and tested the lights on the bike on the late ferry ride home.