Counting Linux
Desktops
(This
article appeared earlier this week, in similar form, on my weblog, http://amywohl.weblogger.com)
Since
I spoke at the Linux Desktop Summit in San Diego on April 22nd, I
have been tossing around the question of how to get a better count
on Linux desktops with a German journalist and fellow speaker I met
there, Bernd Kretschmer.
The
problem is simple to describe. The research firms who offer the
operating systems statistics computer industry companies base their
development and marketing plans on are seriously undercounting Linux
desktops. Not because of any bias on their part, but because the
technique that serves them very well in the Windows or Macintosh
desktop market doesn't work at all in the Linux desktop market.
Most
research houses count desktop operating systems by counting the
number of new PC's shipped with that operating system installed.
That works great for Windows (almost all PC's are shipped that way)
and even better for Mac (every Mac ships with a Mac OS).
Linux
is another story. Most Linux desktops started out in life with
another operating system personality. Probably Windows. Someone
decided to change that by downloading a copy of Linux (or getting a
copy from a magazine, a book (I have half a dozen books right now
which include copies of Linux distributions), or a friend. All of
that is completely legal since Linux is Open Source code and its
copying is unrestricted. (In
fact, it’s worse than that, since some Linux counting algorithms
assume that many low-end systems shipped with Linux are then
stripped of Linux and reloaded with a pirate copy of Windows; the
algorithm may compensate for this by discounting or undercounting
these systems.)
We
need a system that takes into account all of those downloaded copies
(for all of the different Linux distributions) and understands with
some accuracy, on average, how many times a copy will be used.
-
It
might be less than one -- maybe people accumulate lots of
different distributions, checking them out, but ultimately only
using the one they like best.
-
It
might be more than one -- perhaps Linux users are sufficiently
enthusiastic that they generally share their skills and their
software.
And
that's not all. We have other kinds of counting that we need to do:
-
There
are Linux desktops that are loaded on non-PC's such as thin
clients from companies like NeoWare. We need to add those to the
count.
-
There
are Linux desktops that are created on non-Linux environments
via Linux Terminal Server (or its workalikes). We need to know
how many copies of these useful multi-user tools are installed
and, on average, how many users are served by each.
So
this is what we'd like to do.
We'll
put up a table with categories and suggested rules.
We'll
go to the likely sources and ask them for their best estimates
(Linux distributions, major Open Source web sites, Linux notables,
etc.). We're asking you to play the Open Source game.
-
Look
at our categories and suggest changes, refinements or additional
categories.
-
Look
at our estimates and how they were arrived at (we'll provide as
much information as possible; transparency is, after all, an
Open Source specialty) and tell us if you can make a better
estimate.
-
When
the process has run for a month or two we'll publish the
results, keeping them up on the web for further updating.
If
you think you can help us formulate the categories, send me a note.
We expect to have something for you to look at in a week or two.
(back to top)
