Letters to the editor

Deploying Linux on the Desktop

 
Deploying Linux on the Desktop
------------------------------
 
This is a discussion to try explain some of the deficiencies
in Linux desktops that make it difficult to deploy to mass end
users. It's intended for sysadmins who are trying to install
Linux for users familiar only with Windows, and for developers
who are designing user interfaces.
 
I recently had a dinner conversation with Jon "Maddog" Hall,
who insisted that it was feasible to deploy Linux in a typical
end user environment as a replacement to Windows. I think he
presents the opinion of most Linux enthusiasts: most such
people believe that performing such a deployment is feasible
and that it is really managers who are stalling the migration
process only out of dogma.
 
Now there is only one group of individuals that can say for
certain whether such a deployment is possible. This group are
the end user supporters that have actually tried to do such
deployments. We should look at their experiences for
understanding. On the other hand, thought experiments that use
software feature counts as input are theoretical and
inconclusive.
 
Small businesses (10-50 employees) are the target market here.
This is the bulk of most economies. Most such discussions
ought to clarify that the large cubical farms of Dilbert are
actually a rare minority and not worth consideration at this
time.
 
Here is a list of problems and experiences:
 
1. Manager: "I tried to buy a cinema ticket on this web site
with my credit card and it didn't work. My other Internet
purchases work fine, what's wrong with this site?" I
discovered that the problem was some incompatible JavaScript
that only IE could handle. I found that he could use Konqueror
on that site. He started using Konqueror, but certain sites
only worked with Mozilla, and the time fiddling trying to get
things working was starting to escalate. This irritated the
manager who then decided that Linux was not for him, and that
he would rather continue with Windows even if he had to reboot
twice a day, etc.
 
2. How do I save to my CD-Writer? On Windows, I believe this
is merely a drag and drop operation --- all writing is handled
transparently. On Linux we have several comprehensive
interfaces with an enormous range of options. Even if I could
get an average user to remember the sequence of steps needed
to create a working CDROM, they would be convinced that
something was wrong with "Linux" because of the large number
of steps necessary.
 
3. Word document formats. Word documents load perfectly under
OpenOffice. But that's theory talking again. In practice, if
someone emails me a complex 50 page document, a secretary
cannot load it with OpenOffice, make a few changes, save it
again in Word format, and email back and expect all formatting
to be preserved. Most users are impressed by OpenOffice's
ability to handle Word documents... until the day comes when
they have to spend hours fiddling with the paragraph spacing,
margins, and page breaks --- all to get a once perfect
document looking the way it already looked under Word.
 
4. "A:" Drives. It took the average Windows user weeks to get
used the the abstraction that an "A:" represents a floppy
drive and a "C:" represented an internal drive. From file
managers in Windows, it is universal and trivial to save a
file to floppy. Many secretaries (who should be using SMB or
email) still insist on exchanging files by floppy disk. On
Linux, there are those extra few clicks that frustrate such
users and make them think there is something wrong with Linux.
Even the fact that "A:" might be called a different thing
under Linux makes them think that there is something wrong
with Linux.
 
5. Excel spread sheets / PowerPoint presentations / Drawing
Programs. There are many file formats that are problematic.
Moreover, the Linux equivalents of such programs are never
easier to use than the Windows ones. Also, in each case there
is some additional complexity when trying to include one kind
of document inside another kind. The user can easily be shown
how to do it, but they will not remember it when they have to
do it again next week. This will frustrate them.
 
6. Scanning: see point 2. Windows has several idiot proof
single click scanning programs. Even if you totally botch it,
you still get a 30 Meg A4 TIFF file that you can email to your
granddaughter.
 
7. Hardware. Every office has at least one piece of hardware
that you are going to need to replace. Finding replacements is
time-consuming. Even to establish what software within that
Linux distribution is responsible for the deficiency, is a
massive problem for any small business.
 
On Linux, *almost* everything works fine. On Linux almost
*nothing* takes less clicks or is easier than Windows.
 
Users just want things to be the same as they are used to. Any
change is difficult for them to master and wastes precious
office hours.
 
Unless Linux dsitributions can come up with a desktop that is
click-for-click the same to operate as Windows, there is no
chance of migration.
 
Here are some axioms for developers and people who create
Linux distributions:
 
0. Any computer experience that is not *even* *easier* to use
than Windows will not be able to compete. All other reasons to
switch to Linux are precluded by other solutions that the vast
Windows development community is constantly inventing.
 
1. An extra configuration parameter is a poor excuse for not
thinking about what the default SHOULD be. Example:
application fonts being too small at 1280x1024. There should
be no setting to change the application font --- the most
readable font should be the default.
 
2. If a user EVER has to type any command at a shell prompt,
then the operating system is broken. Example: when do you have
to supply DOS commands under Windows or MACs?
 
3. If it takes one more click than on Windows, then you might
as well not have that feature because Windows users are not
going to remember the steps to carry out.
 
4. Desktops are confusing to users. The more lights and
buttons, the more difficult it is to remember what to click
on. 90% of buttons under KDE/Gnome will never get selected. If
you have more options than Windows, then users will get
confused.
 
5. End users of Windows are FAR less intelligent than you
might expect. It's EXTREMELY difficult for them to remember
even how to select bold fonts under a word processor. You will
probably have to train a person for two hours just to show
them how to do headings, bold, and italic under Word. A week
later they will have forgotten unless they are pressured with
constant repetition.
 
6. Most people are not interested in playing MP3 files.
 
7. Almost no people anywhere are interested in authoring their
own raster images.
 
8. A GUI that does not factor in the intelligence of the
end-user is useless. The intelligence of the developer is
vastly superior to that of the end-user. This gap has been
underestimated by all software vendors except Microsoft.
 
-paul
 
 
Postscript: I suspect that people are going to be offended by
my insistence that end-users are near-retarded. To be offended
by an undeniable fact is stupidity in of itself. Developers
tend to assume a tremendous amount --- just to be able to
understand the concept of abstracting a problem into a
sequence of steps requires tremendous genius (by comparison to
the average person). Such genius should be normal. In fact it
is rare --- there are very few software developers compared to
clerks, salesmen, and cleaners. The intelligence to view a
computer screen (that is essentially inoperative) and surmise
that its pictures are in fact control switches, requires a
leap of faith which an unintelligent person is going to
require much time and training to make. The average desktop
interface has thousands of lines and glyphs. Making sense of
them is extremely complex for most people. DO NOT assume that
because all-the-people-you-know are able to fiddle to get
something to work that the majority fit that same category. If
you really want to understand the end-user then go take a
training course that introduces first time users to MSWord.
 
 
 

Comments (19 posted)

Good alternative names for Mozilla/Firebird

There is a famous Russian folk story about a Firebird, which was made
into a beautiful piece of music by composer Igor Stravinsky. The
Russian word for firebird is "zhar-ptitka", or "heat-bird". I suggest
the Mozilla team solve the hard feelings they have caused with the
Firebird database project by choosing another name based on the Russian
word for Firebird.
 
If that is not satisfactory, "ere" from irc.mozilla.org suggested the
Finnish word for firebird: "Tulilintu".
 
Given the strong folktale associations most people have of the word
"firebird", it would be compellingly appropriate to revive the name in
one of it's original languages.
 
Please show us that you aren't barbarians; please show us that you have
some culture. For the goodwill of the Free Software community and the
world at large, please change the "Firebird" projects name to one that
will be both more distinctive, and aesthetically pleasing.
 
Cheers!
 
Jonathan
 
--
                     Geek House Productions, Ltd.
 
  Providing Unix & Internet Contracting and Consulting,
  QA Testing, Technical Documentation, Systems Design & Implementation,
  General Programming, E-commerce, Web & Mail Services since 1998
 
Phone: 604-435-1205
Email: djw@reactor-core.org
Webpage: http://reactor-core.org
Address: 2459 E 41st Ave, Vancouver, BC V5R2W2

Comments (none posted)

LWN: Searching for software or having an itch...

In a letter to LWN you write:
>I just had a need to make a few campaign signs for my effort to get
>elected to Town Meeting locally. I'm a really lousy artist, so I had the
>idea of printing out the content of my signs on letter paper using very
>large type, and either gluing the paper printout onto my poster-board
>signs, or cutting them out in order to make stencils.
>[....]
>We couldn't find anything on Google, searching on things like 'Linux
>Large Fonts' gave lots of advice on changing font size on the video
>display, but no programs.
 
The traditional program for doing this under unix is: banner.
 
On my Debian Linux system it is in /usr/games/banner, from the package
bsdmainutils.
 
banner produces large letters in ASCII-art form, ie by drawing them with
ASCII letters. It dates back to the days when line printers were king;
the earliest copyright date in it is 1980.
 
>I'm not sure what the answer is, but it seems to me like the Open Source
>world needs a better CENTRAL catalog of available software
 
http://freshmeat.net/
 
It's not perfect, but it's the closest thing there is to a central catalog,
and generally I've found it useful when I've got a "I need a program to do
this" type query. However sometimes browsing through an appropriate
category is more useful than word-based searching.
 
In addition to that I use Debian's package repository:
 
apt-cache search ....
 
on a debian system; or visit:
 
http://www.debian.org/distrib/packages
 
and use the search forms there. Debian is useful for this because
they have a lot of free and open source software packaged.
 
Sourceforge: http://sourceforge.net/ can also be useful, but only for
the fraction of open source software hosted on SourceForge (maybe 30%-40%
at most).
 
Google is useful for some things, but you need to be very particular with
your search terms some of the time to narrow the results down to the right
set of things, otherwise as you found you end up with "lots of similar but
not the same thing" hits.
 
Ewen

Comments (3 posted)

Page editor: Jonathan Corbet