[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Debian and the User Friendlies

Bill Leach <bleach@bellsouth.net> wrote:
> There is fundamentally no reason why debian has to have a large
> market share.

Yes and no.  And yes, and no.


(1) Our audience provides a valuable service:  testing, evaluating, etc.

(2) We don't sell anything (at least, not as a group, not for money).

(3) Personally, I wouldn't be doing this if I didn't think it had value
for someone (and yeah, one of the people it has value to is me).

(4) People that don't know about debian aren't going to use debian.
[This is in some ways completely independent technical quality, by
the way.  It's mostly a reflection of what people are writing about us
(or not).]

Since we're not doing sales, we can't measure our value in terms of
sales figures.  It doesn't make sense, and I think it would be a waste
of time to get involved in tradeoffs to drive up our sales figures.
[I hope no one here disagrees with me on this.]

But we are trying to put together systems which are useful to people.
And, being unapproachable isn't going to help us on that.

And there are tradeoffs involved -- if nothing else, there's the old
80/20 rule.  [Or perhaps I should say rules:  80 percent of the time
is spent in 20 percent of the code, and 80 percent of the gains can be
realized with 20 percent of the work.]

Right now, our biggest hot-spot is X configuration.  We have lots of
other things which are important to deal with, but X configuration is
the most glaring thing likely to trip up a new user.

It's very hard to incrementally approach a good X configuration -- you
have to learn by experimentation on a system where the documentation
says that you can fry your hardware.  You have to solve a set of linear
equations which are highly quantized and which are not associated with any
visibly linear sort of behavior.  Too often, critical information is not
available, leaving trial and error (or maybe very clever web searches)
as the only way of dealing with the problem.  The answers you give are
frequently ignored when you're forced to start over.  If you discover
you've made things worse, it's not very easy to get back to what you
had before.  Basically, we're requiring our users to be skilled in black
magic, or to stumble on some well-behaved local properties where they
can get "good enough" results with simple guessing or approximating.

At least, that's the way it was last time I configured an X server.


To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Reply to: