Attention / Interest / Desire / Action, getfedora.org, or why marketing 'the Fedora project' is a bad idea.

[Jeff Spaleta]

On 6/23/05, Tom Adelstein <adelste at yahoo.com> wrote:
> A Sun market study, Canonical Ltd., Debian's web site analysis, my
> publisher - O'Reilly, former Ximian - now Novell people, the executive
> suite of Novell, IBM, a survey of users, classified information from a
> DoD study, results of a series of articles reviewing various
> distributions, Linspire, Xandros, and sampling from Linux forums such as
> Linux Questions.

I was hoping for citations that i can get a look at with some raw data
and potentially a summary of the methodology for sampling. I'm very
interested in getting statistically verifable analysis for any usage
numbers for any distribution.  I'm especially interested in any
analysis break down of ubuntu's numbers not lumped in with as a debian
aggregate.
The last publicly available data I am aware of that even attempts a
comparison... is the webserver survey of netcraft:

http://news.netcraft.com/archives/2005/03/14/fedora_makes_rapid_progress.html

I'm more interested in the raw data than the glowing conclusions. In
fact I'm really not thrilled with how fast fedora is growing in that
study of webservers. The implication that fedora web servers are going
leaps and bounds sort of worries me that hosting companies are
choosing fedora without telling customers about the support lifetime. 
I really didnt expect fedora usage in the web server space to increase
as fast as it has.  There's a lot of room for interpretation of that
data. But sadly its the only comparable raw data with a set sampling
methodology that I am aware of.   I am of aware of the distrowatch.com
polling, but even they don't attempt to claim their poll is an
expression of real world userbase stats. At best its "popularity" and
they admit the sampling methodology can easily be manipulated by the
self-selecting participants.  Remember linuxquestions.org poll..

http://www.linuxquestions.org/questions/showthread.php?s=&threadid=272090

what was the best distro of the year?  slackware.   Not to knock
slackware or anything. But that result seems to indicated a very
different world order than you are suggesting... and you even site
sampling of linuxquestions.org as a credible source. Sorry, but forum
activity just can't be used for anything more finer grained than
"critical mass" userbase analysis.

Also unfortunately ubuntu shows up as an aggregate with other "debian"
releases so even in the very narrow scope of web servers I still don't
have publicly comparable raw numbers for unbuntu alone. We could very
well lump fedora in with rhel and get equally meaningless results
similiar to lumping all debian derivatives together.

If you have any references to publicly available raw data, I'd
absolutely love to see it.
Even getting accurate stats of downloads is really tough. Because of
mirrored distribution and secondary torrents  you have a hard time
knowing how many copies are really floating around out there. At best
you can get a nice lower-bound to know if you are getting a critical
mass of support, but anything beyond that is impossible to do
accurately with download stats.


-jef