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

Re: MATE chosen by default instead of gnome for blind people [Was: Debian Installer Stretch Alpha 6 release]

Samuel Thibault <sthibault@debian.org> writes:

> Cyril Brulebois, on Sat 21 May 2016 23:13:17 +0200, wrote:
>>  * brltty:
>>     - Install MATE desktop by default when brltty is used in d-i.
>>  * espeakup:
>>     - Install MATE desktop by default when espeakup is used in d-i.
> This change has triggered discussion on the debian-boot IRC channel. To
> summarize, gnome people are surprised that MATE would be preferred by
> blind people and wonder what they need to improve in gnome. I answered
> that it was more a problem of general trend towards visual ways of
> using the desktop, which can't be made really usable, but that's only a
> sketchy answer, they need more precise examples.
> Could blind people here comment on this: why you don't use gnome and
> prefer MATE instead? (or the converse of course, the idea is not to
> blame gnome, we just want to select by default what is best for users,
> according to their situation.

I have to submit that I am not really qualified to do a comparison, as I
have not used MATE personally.  However, I can try to describe the
regressions since GNOME2 that still haven't been fixed from a
perspective of a *user* not actually familiar with the visual appearance
of the current desktop:

GNOME3 compared to GNOME2 feels very unpredictable.  This might be
because the way how keyboard based control is now been done has been
revised significantly.  I still haven't really figured out how to use
*any* GNOME desktop features from the keyboard except Alt+F2.  I
basically use GNOME as a launcher for firefox.  Partly because I do most
of my work still in a console, but also because the way how I am
supposed to use GNOME3 for anything else then launching a command by
hand is totally mysterious to me.  I don't know how to use the menu, if
there is one.  I vaguely remember someone showed it to me, but it
definitely didn't work like expected and was apparently so unintuitive
that I forgot again how to do that.

I don't see anything on my "desktop" as I used to do in GNOME2.  I am
somehow suspecting that is because GNOME3 has changed the whole approach
of how a desktop is supposed to "look", however, as a blind person, I
don't really know either way.

As I have been told, MATE is mostly how GNOME2 worked.
And from that perspective, I can totally understand that blind people who use
a typical graphical desktop for more then just launching a single app
would prefer MATE over GNOME3.  GNOME2 was actually very accessible in
its final versions.  The app launch menu was easily accessible, and even
desktop widgets like GNOME Weather worked perfectly.  So if you were
into that thing of customizing your desktop to your needs, you could
actually do it, even in an exploratory mood, without having to refer to
documentation that probably doesn't exist anyway.

That said, I am a little bit upset that GNOME people, after all these
years of working on GNOME3, are still surpised that blind people are not
happy with it.  I would have assumed that if devs did honest research on
that topic, they would immediately know that there is still quite a lot
of ground to cover.  Accessibility doesn't get simpler, it does get
increasingly complicated with every new and fancy technology
introduced.  And with new ideas like Clutter, or the web framework
switch we seem to see every 3 or 4 years now, a lot of work is needed to
get accessibility up to a level of real usability.
Whenever you deliberately introduce new and fancy technology, and the
early GNOME3 days very much felt like that, you create a horrible
situation for Accessibility.  If you don't take Accessibility into
account from the absolute first move you do, you will have to patch it
up after the fact, which usually leads to horrible systems.  Sorry, that
is just a fact, I have seen this happen over and over again.

While this is not really related to MATE vs GNOME3, I'd like to raise
the web framework switch as an example to show that Accessibility
doesn't really "work" yet because it is not treated as a first-class citizen
in the requirement space: I dont quite remember when it exactly was, but
the internal web rendering engine of GNOME switched from Mozilla to
WebKit one day.
When the switch happened, WebKit accessibility was not ready yet.
During that time, things like the GNOME Help were simply not accessible
I think this case should be clear enough.
If, due to some other technical needs, we decide to switch to something
but don't make Accessibility a first-class requirement, we end up
leaving the users that depend on it totally in the dark for a while.
While I can sort of sympathize with those that need to juggle human
resources, the end result of ending up with an inaccessible HELP system
is just totally unacceptable.  As long as upstream projects are willing
to do such gross moves, we really can not claim that these systems are
Accessible.  They might be Accessible by accident, for a few days.  But
for all we know, they might also stop to do so tomorrow.

What I am trying to say is, if a desktop wants to provide Accessibility
that is actually useful to users, they will have to invest more time
into it then they currently are willing to do.  This problem is not
going to go away by discussing it.  The only way we can improve
Accessibility of mainstream upstream projects is to convince them to put
more focus on the issues.  These issues are often subtle and not easy to
see if you are not doing honest testing.  And many people don't know how
to really test a screen reader, because if they still see the screen,
they will fail to notice many accessibility issues.  It just feels
different to interact with a system if you don't see all its rich and
colourful attempts to make things more clear to the user.

I realize what I have said is actually quite vague.
For a concrete list of problems, I think it would be best to
 * Go through the GNOME Bugtracker.
 * Do some real usability testing with blind users.
   Unsupervised solo experiments do often lead to very vague and emotional results.
   If we really want to have a quantified measure of where we stand, we
   should do it like academics teach usability.  However, this very probably requires
   face-to-face time and rather careful preparation of use-cases.
   And it is probably something upstream needs to do, not Debian.

> Ideally we'd just say "pick whatever, they're all good").

Of course.  On a general note, while I can understand why people are
pushing for a system that works like GNOME2, I think ultimately we will
have to convince mainstream systems to make their offerings comparable.
Choice is a good thing.  But I am not sure it will be a good idea to
move all blind people off to MATE by default, this takes the pressure
away from GNOME.


Attachment: signature.asc
Description: PGP signature

Reply to: