> A proposal: Stumblebum Software Development -- keep the cost of
> software R&D low enough so that it can be paid for my a support
> organization.

I came to a similar conclusion a couple of weeks ago while discussing
"Free Software" with a friend.  Many new products of low quality are
generated, somewhere.  Those that are truly popular will generate
some amount of consulting/bug-fix work, which will improve them
(making them more acceptable) and set the ball rolling.

One problem with this is products that contain a "fatal flaw".  An
example of this is an "easy-to-use" product that depends upon use of
emacs.  Implicit in this product is the idea that emacs itself is
"easy-to-use".  Some people believe this to be the case -- certainly,
I've invested enough time in learning emacs over the years that it
would not hinder me.  On other hand, some people (some hackers, some
non-hackers) don't use emacs, hate emacs, and refuse to learn emacs.
These people are, of course, potential customers, so it is wise (from
a business point of view) to treat them politely.

This may not be a problem -- if the FS market is generally people who
are programmers, more or less (i.e., people who use gcc, gdb, emacs)
then that's ok.  On the other hand, there are people who are definitely
hackers who have chosen (over time) to not use a free product of high
quality in favor of a non-free product of lower quality.  I'm one
example -- over the years I've hacked both TeX and LaTeX, built both
from scratch, installed them, tinkered with dvi-to-foo software, written
style files and macro packages.  Nowadays, I prefer to use MacWrite or
Claris Works to write things, even though they cost money, don't format
mathematics as nicely, don't auto-hyphenate, etc, etc.

David Chase
Sun