Linux on the desktop
[ Main ] [ Home ] [ Work ] [ Code ] [ Rants ] [ Readings ] [ Links ]

Boy, how often did I discuss this issue with my friend Sam? We talked about it over a cup of coffee, when checking the computer books and magazines over at Barnes & Nobles or Borders, when... we had a few minutes to talk to each other. The problem is, of course, that we always discussed it in a very informal way, without any structure, so I finally decided to sit down and clarify my ideas on the topic.

Date: Mon, 4 Mar 2002 21:00:04 -0600 (CST)
From: Jesus Ortega
To: Sam Moreno
Subject: Linux on the desktop


Let's see then if we can have a serious, deep and unbiased debate about the
Linux on the desktop issue.  I prefer to do this via email because that way
we have more time to think about what we say, and can even take the time to
reply the following day if we want to.  As usual, please reply _below_ each
paragraph or section that you want to discuss in order to make it easier to
follow the thread.


First of all, I hope we both agree that the current market share of Linux
on the desktop is nearly non-existent or at least as small that it doesn't
display on the radar of most companies.  In any case, there should be little
doubt that its current market share is well belong that of the MacOS which
is already quite low itself (around 5-6% of the desktop market).  Now, if
we talk about the server market that's a whole different picture but that's
not our topic.


Second, when some analists mention that the major Linux vendors have given
up on the desktop market... well, like it or not, they seem to be accurate.
Several Red Hat top executives (including Young) have recently acknowledged
that much.  As for other vendors such as Caldera and SuSE, they have not
said so in an explicit way but the fact is that most of their recent product
releases (Caldera released Volution and also the Messenger server for MS
Outlook environments, which is a clear hint when it comes to their position
on the desktop market, while SuSE has also released a series of products to
act as a firewall, email server, groupware server... again to be set up in 
a MS Outlook environment) are obviously geared towards the server market.  
Perhaps Mandrake is the only major vendor that still centers most of its
efforts on the desktop.  Now, all this is _not_ to say that these vendors
will never sell for the desktop market but simply that they have, at least
for the time being, given up that market, and seem to believe that Linux's
strength is on the server side.


So, once we clarified the picture about the _current_ status of Linux on
the desktop (and, hopefully, there was no major disagreement so far), let's
see what we can discuss about the _potential_ for Linux on the desktop.

Let's start with the obvious.  Is the Linux desktop up there yet technically
speaking?  Can we expect to simply switch a Windows or Apple desktop with a 
Linux desktop and have the users up and running in a matter of days?  Well,
I submit that yes and no or, in other words, relatively.  I'm basing this
not on a hunch, but rather on my own personal experience at home (Leslie's
computer broke and she had to run a Linux desktop with KDE for a couple of
weeks) as well as the experience of a few people at work who were relatively
new to Linux (Carolina, Rene, Mario, Jeff...).  

3.1. The home desktop

What do I think the Linux desktop is still missing then?  Where does it need
to improve in order to catch up with Apple or Microsoft?  Let me mention a
few issues here:

o The menus (both in KDE and Gnome, and also in most distributions I have
  installed and run so far) are still quite raw and inconsistent.  There are
  several instances where you see an icon in the menu but the application 
  is not installed.  Let's not even talk about a default installation of
  WindowMaker or Enlightenment... and yes, I know what I'm talking about 
  because I ran all these within the last few months (SuSE, Red Hat, Debian,
  Gnome, KDE, Enlightenment, WindowMaker).  I've never come across this
  behavior in the Windows or MacOS environments unless I went through the
  trouble of removing the executables myself, and I _definitely_ never
  encountered this behavior on those other environments in their default
  installation.  Caldera is perhaps the only exception to this problem.

o Installing applications in Linux is still a major pain in the butt, and
  even an otherwise experienced UNIX user like Carolina struggles when 
  trying to install a given application.  Windows used to have its own
  "DLL hell" and Linux seems to have its own "dependencies hell" (I believe
  Miguel de Icaza already pointed this out in one of the documents that
  from time to time he publishes on this very same topic).  Sure, you also
  have some nice GUI installation utilities in Linux but they still don't
  solve the dependencies problem and are not nearly as nice as the old
  Installation Wizards in Windows.  I know, Windows compiles most things
  statically and Linux doesn't, which means that Windows wastes a lot of
  disk space... but the reality is that a regular home desktop user doesn't
  give a damn about it.  All they want is to be able to install an application
  quickly and easily.  That's all.  Linux is still not there.  

o The browser experience.  More and more often, the Web surfer comes across
  web sites that _only_ support MSIE and are either not viewable at all
  under any other browser or are severely distorted.  I understand this is
  not something that can be blamed on Linux but still the regular home
  desktop user doesn't care much about who is to blame.  All he knows is
  that web site X cannot be viewed in Linux.  Period.  

o Related to the browser experience goes the plugin experience.  It still
  is a pain to run most plugins under any Linux browser and it's not easy
  to configure them quite often.  When it comes to Windows or MacOS, on 
  the other hand, they usually come bundled with the plugins and you can
  choose to install them or not from the Installation Wizard.  Sure, you
  can install the Crossover plugin but you still need to install and
  configure Wine and, most importantly, _pay_ for it.  That defeats the
  whole argument about Linux being a cheaper alternative to Windows.  No
  matter what, there are several plugins that are not easy to run under
  Linux if at all: Quicktime, Windows Mediaplayer formats, the MP+ format...

o When it comes to multimedia support, Linux is still lagging behind too.
  I'm not so much referring to multimedia applications but rather to the
  configuration itself.  Windows and MacOS are usually "self-configured"
  or pretty much so.  Linux, on the other hand, is not easy to configure 
  for sound, video, etc.  Easier than it used to be?  Sure.  Easier than
  Windows or MacOS?  Not really.  Let's be honest.

o There is a clear lack of games for the Linux platform, and this is so
  obvious that even well known Linux hackers have no problem acknowledging
  that they _have_ to run Windows at home for gaming.  The same applies to
  educational software for kids.  In this respect, the MacOS is not as good
  as Windows either but it still is years ahead of Linux in spite of it.

o Administration utilities.  These are still inmature, needlessly complex
  and inconsistent.  To top it all, they are different for each major Linux
  distribution.  There is no point in comparing them to the ones for the
  Windows OS or MacOS.

3.2 The business desktop

Quite a few of the points I mentioned above would also apply to the 
busines desktop.  Still, perhaps the most important point here would be
the old mantra: "applications, applications and applications".  There is
still no equivalent of MS Outlook, Clarify, Primus, Lotus Notes, Peachtree,
MS Money or Quicken for Linux.  As for the office suites, Star Office was
getting there (especially in its 6.0 release), but Sun appears to be ready
to change its licensing policies now.  

Other than that, please apply many of the same points I mentioned in the
previous paragraph. 

3.3. The technical desktop

I'm referring to techies and developers.  This is perhaps the only market
(or should I say "sub-market"?) where the Linux desktop not only has 
potential, but it has already arrived.  It offers the reliability, power,
stability and flexibility that neither Windows nor MacOS offer as of now.
When it comes to Web development, Linux is perhaps the best tool that there
is in the market right now... at least as long as the code you are writing
is also going to be deployed on Linux or UNIX systems.  Of course, it's a 
whole different story when we talk about deploying web software on a 
Windows server.  In that case, you'd better run a Windows desktop too.  
Finally, when it comes to applications development, the very same story
applies here too.  If you are going to deploy on Linux or UNIX then run a 
Linux desktop, but if you plan to deploy on Windows or MacOS then run one
of the those OSes.  

The scientific and research market, I hear?  That should be Linux's market
within a few years, and that's precisely the _only_ reason (I think) why
SGI didn't bail out of the Linux community.  They can see this one coming.
The same applies to the high-end visualization market.


So, is this to say that the Linux desktop has no future at all?  Well, keep 
in mind that I've been talking so far about the _current_ situation.  That
is, I believe, the current situation and I'm not going to lie to myself
about it.  Well, what about the future then?  Well, you see the problem 
about the future is that Linux will only have a chance against Microsoft
(a "de facto" monopoly, lest not forget) if there is also a major shift of
paradigm perhaps caused by a disruptive technology.  Is it possible to see
any of these coming already?  Not sure, but let's try.  

o The "money" issue.  Since Microsoft appears to be getting quite cocky
  about the license for their software, is there any chance that businesses
  may turn their backs on the Redmond giant and adopt Linux due to the 
  possible savings?  I doubt it.  This could be _a_ factor, but never the
  main one, especially among big companies and perhaps even among the
  middle sized ones.  Sure, these companies are forced by Microsoft to 
  upgrade their OS every 2-3 years, but the reality is that they already
  do their own hardware upgrade every 2-3 years anyways... and guess what
  OS already comes pre-loaded in those boxes?  Sure, the latest one from
  Microsoft.  So, what's the deal?  Will they truly save so much money by
  switching to Linux?  I doubt it.  You also need to take into consideration
  other factors like training, applications, etc.  As for small businesses,
  that's a different issue but they normally avoid any risks in fields that
  they don't know much about.  Also, don't forget that they don't tend to
  have anyone who is Linux-savvy on board... or even technically savvy at
  all for that matter!

o The "network-centric approach".  Accepted wisdom has it that since the
  Internet and networking technologies were born out of UNIX and there is
  no doubt that both UNIX and Linux are network-centric, then the expansion
  of the Internet will also benefit the adoption of Linux.  Well, not
  necessarily.  Microsoft may be semi-monopolistic and it may also incur
  in a lot of unfair business practices, but we have to admit that Bill
  Gates knows what he is doing.  He did see the Internet threat coming 
  (perhaps late, but he did see it), and had the wits to change the course
  that the whole company was taking to adapt to it.  Now, with the .Net
  gambit, Microsoft is arguably again riding the wave.  

o The embedded market.  There was some hope that Linux may take over in 
  this market and beat Microsoft at least here.  Well, it's not so clear
  anymore.  There are clear signs that Linux will instead take over all 
  the market currently held by UNIX-like companies such as Wind River or
  VxWorks or QNX.  Microsoft, on the other hand, is also encroaching into
  the market starting with the PDAs, gaming boxes and TV set-top boxes.
  So, it looks more and more like that market will be shared among the 
  two sides more than anything else.

So, is there any hope?  Like for SGI, I think there is still a possibility
although that doesn't mean I'm highly optimistic it can be accomplished.  If
anything, I think it's far easier in the case of Linux... although I'm not
saying this with a happy face.

Here is a possible outline of what could be a successful Linux strategy
in this field, and keep in mind that I _am_ in favor of concentrating _all_
our efforts in these markets first of all and forget about the general home
desktop market completely at least in the first phase:

o Market and sell Linux high-end workstations for the scientific, research
  and visualization markets.  I mean not only IA-32 systems but also IA-64
  when they are finally out, and I also mean _aggresively_ market and sell
  these workstations.  SGI can be a big player in this sense, and it should
  be a big player in the IA-64 market.  I also think we should have not 
  completely pulled out of the IA-32 market, but rather limit ourselves to
  selling high-end performance Linux workstations.  Any of our traditional
  customers who turned to SGI looking for a Linux solution was bound to be
  disappointed if a 230 is all we had to offer.  Let's face it.

o Pay special attention to the Government and foreign markets.  Here the
  traditional opensource and "free as in speech" arguments should be quite
  strong to the point of being able to carry the day.  The Government _has_
  the know-how and the means to fix a bug in the code, and it doesn't have
  to depend on a commercial vendor's plans to do so.  On the other hand,
  foreign governments have the added interest of not depending too much on
  US-based Microsoft.  Linux vendors should play this card shamelessly.

o Increase the efforts on the browser too.  If the Linux community is
  not able to come up with something that competes against MSIE the whole
  Web slips away.  That's one of the worst possible developments.  Hopefully,
  the UNIX vendors such as Sun or SGI will see that this is in their best
  interest too.  Add here what I wrote about earlier about the plugins, etc.

o Develop an easy to install and configure server _and_ desktop replacement
  for the MS Outlook + Exchange juggernaut geared towards the SMB market.
  The "money issue" may make a difference in this market, especially if they
  don't lose any functionality and the offer also includes nice administration
  and support deals.  

o Keep fighting for the embedded market.  There is a good chance the future
  lies more with things like set-top boxes, network aware phones, home
  communications appliances and the like.  

From these few strongholds, it may be possible to expand to other markets
and even to the home desktop market eventually.  However, as you can see,
I don't even think we should pay any attention to that market in the near
future.  The strategy would go through these other battlefields I laid out

Let me know what you think about it.


Linux XFS-->