focus! (was Re: Focusing on innovation re: mono, python et al)

[Havoc Pennington <hp redhat com>]

Iain * wrote:
> Really?
> depends on your context...
> For some people a terminal and text editor are completely worthless,
> but take away photo management....
>
> Once again, who are we targetting with the desktop. Apple know who
> they're targetting, which is probably why text editor and terminal are
> not high on the list of features.

Yes! I was hoping your thread about this would catch fire instead of the 
 one about mono, because answering the "what is gnome anyhow?" question 
would make the mono-type debates much simpler.

If GNOME can't figure out a way to answer that question, its only option 
is to be a platform provider for Elisa, Maemo, SLED, Fedora, Ubuntu, 
Palm, Firefox, WINE, etc. etc. Those are all more focused, more 
target-audience-decided-upon solutions that in many cases use GNOME 
technology but diverge to a small or large extent from the GNOME desktop 
release because guess what, actually shipping something useful requires 
more focused, specific thinking.

There's nothing wrong with the platform provider path, and it's probably 
inevitable by inertia and industry dynamic, but if taking that path it'd 
be interesting to do it consciously and optimize GNOME as a platform 
provider - with the providers of all those more focused solutions as the 
primary customers. And this _also_ helps answer the Mono debate - the 
question would become how to best serve the specific solutions and the 
teams building them.

To me there are two hard parts to answering the target audience / what 
is GNOME question:
 1) how does "GNOME" decide anything? it's a big swarm of people
 2) which audience or focus to choose?

Here's one way one might approach it.

: Step 1. Collect underpants.

j/k

: Step 1. Redefine GNOME as in the original charter; provide an open 
source computing platform to the general public. Do this on the 
foundation level and get wide buy-in. Hammer the message consistently 
through the web site and other communications. The goal is to fight off 
the "GNOME = desktop environment" legacy.

Note, "platform" in the charter I think has to be understood as 
"environment" or "solution" not as "APIs" - might be worth officially 
rewording in that way. In fact, I think it has to include both software 
bits AND finding some way to work with "content" and "online services" 
if there's a serious interest in offering open source alternatives to 
today's proprietary software companies.

So, let's assume "platform" includes all that stuff for purposes of 
redefining GNOME in this way.

: Step 2. Kill the single "desktop" release and replace it with 
target-audience-specific/solution-to-problem-specific more focused 
releases. For example, while they may not be interested, Maemo and Elisa 
would be candidates. The current "desktop release" should become one 
thing among peers; or it's even worth considering splitting it up to be 
multiple peers.

Don't call the desktop release "desktop" either because it's too vague. 
More specific examples might be an "enterprise unix/linux GUI" release, 
or "tech-oriented consumer/hobbyist" release or "tech workstation 
release" or "high-powered MS Office user in an office release" or 
"computer lab / thin client release" or whatever people feel is the 
right focus.

The word "desktop" is like a cancer. Its problems include:
 - it's vague as hell - includes a zillion target audiences and apps
 - it accepts an existing category definition (essentially, "what
   windows and mac are") thus precluding meaningful innovation
 - it excludes "content" and "online services" -
   key elements of all the new stuff going on in the tech
   industry today

The huge debate here is how to split things up; the important thing to 
remember is that there can be lots of code sharing (where it makes 
sense) between related offerings. So e.g. almost everything could use 
GTK, but only some offerings might want the GNOME panel.

i.e., doing the split by _codebase_ is wrong; the split is by _target 
audience_ and _focus_; some splits might be worthwhile _just to change 
the default config options_ even.

The technology can be made to support such things, and in fact it should 
be made to do so.

Also of course, the split depends on having volunteers to "own" each 
release.

The counterintuitive and hard-to-accept reality is that trying to be 
universal just leads to being vague and useless. The right approach is 
to try to be specific (and useful), then factor out common elements 
between multiple specific solutions, resulting in a platform. aka "top down"

This is happening de facto _anyway_! Look at all the different things 
people are building on GNOME tech. It's just that GNOME is not 
acknowledging it, and not taking credit for it. GNOME still sits here 
claiming to be a "desktop" and that's a very limited view.


A couple other notes:

 - something like http://live.gnome.org/Personas is not helpful, because
   it's way too broad. Need to pick only some of the people there, and
   then (even harder) pick only some of the activities listed there that
   the people are doing. Or better, find new activities they aren't
   doing but would like to do, and offer something for that.

 - look just within the "linux distribution" space for huge diversity.
    - Federico's deployment study revealed that most current users
      are the "government/edu computer lab, often with thin client"
      variety and the main needs are e.g. management/admin tools
    - I would say GNOME upstream isn't primarily focused on this, but
      on the Fedora/Ubuntu "tech enthusiast consumer" kind of people
      offering a kind of "play with technology" platform
    - SLED (and Ximian, Bluecurve / Red Hat 8, etc. before that along
      the same lines) has its main features oriented in some other
      direction still, which I won't try to summarize for fear of
      getting it wrong

   For any of these focuses, GNOME could be much better. e.g. the
   age-old "remove crack" vs. "don't dumb it down" debate is essentially
   a target audience confusion among some of the above.

   BUT more importantly - there are so many audiences and solutions
   beyond these, and we see companies pursuing them with GNOME tech,
   while GNOME itself is stuck in its Linux distribution history.

 - again, there's really nothing wrong with the platform provider angle;
   but it could be done better and more strongly by focusing on it
   deliberately. For example: change my "step 1" above to revise the
   charter so instead of "general public" the goal is "provide a
   platform for organizations building focused solutions [insert some
   kind of qualifier on kind of solution - GUI? consumer? I don't know]"

 - look how broken even the subject line of this thread is without
   having any real focus - is innovation valuable at all? (for
   "unix replacement" customers the answer is mostly _no_).
   in what direction should the innovation move? who does it benefit?
   does innovation outweigh increased memory usage? (for tech
   enthusiasts probably yes, for thin clients and Maemo probably no,
   etc.)

 - Once "what is gnome?" gets hashed out it's essential to drive clear
   communication about the solution throughout the project: the charter,
   the web site, the mailing list names and divisions, the release team
   roles, reach out to certain external projects... everywhere the
   concept "desktop" is used to define the project, it needs to be
   remapped to a set of distinctions around focused benefits for
   specific people.

Anyway, while I'm advocating a thread about a perhaps a much larger can 
of worms than this whole mono thing, it's also a much more _useful_ can 
of worms (and might help solve stuff like the mono debate as a side 
effect). It also sheds enormous light on some past conflicts (e.g. the 
whole OpenOffice/Firefox vs. native GNOME stuff debate).

Havoc