[FM Discuss] scribn the bazaar

[Andy Oram]

I have a few thoughts about Cathedral versus Bazaar manual production,
or whether you should let people write what they're passionate about
instead of assigning topics.

Letting people do what they want is great if you don't have a list of
must-do topics. But what if there's some information that nobody
happens to write about, and that you need in the book because it's the
basis for understanding other things, or because readers want it and
the organization sponsoring the project is committed to providing it?

Many of the books done by FLOSS Manuals are about easy-to-describe
technologies. If the project leader notices that a key piece of
information is missing, somebody can probably be found who knows the
information and can take a couple hours to fill it in.

Furthermore, volunteers don't owe anything to anybody. Deadlines are
also rare. FLOSS Manuals is in an enviable position, because you can
pretty much release whatever you have and say, "If you find that
something is missing, go ahead and add it."

The situation is different with conventional publishers and advanced
topics, which are the basis of most of the O'Reilly books I edit, for
instance. I sometimes find it extremely hard--particularly because I
have deadlines--to find someone to cover a topic that absolutely needs
to be in the book.

But there's another model sort of halfway between the two Adam
described. You find smart people you trust and just let them write
about any topic they want.

This is the familiar model for conference proceedings. They contain
whatever the leaders of the field happen to be working on at the time.
Some of the books produced that way at O'Reilly include Open Sources
(which should interest people on this list) and Beautiful Code.

Even in those anthologies, you want variety. Conferences achieve it by
simply choosing the best representative of some line of thought and
rejecting the other proposals. For Beautiful Code, we worked hard to
get diversity in every way we could (national, gender, technology,
programming language, etc.). We were mostly successful, but not
entirely.

Andy