Paul Frields stickster at gmail.com
Sat Nov 22 17:16:55 UTC 2008

Hi fellow Fedora Marketeers,

As we move forward in marketing Fedora, I wanted to say a few words
about statistical presentation. There may be a tendency among people
not familiar with the Fedora Project and its objectives to confuse
some of our statistics.  People often like to position Linux
distributions as being in competition, which I believe somewhat
obscures the commonalities of free software communities and especially
Fedora's goal of advancing free software.  We in the Fedora Project
have a very comfortable position as a leader in technical innovation
and community contribution to free and open source software.
Unfortunately, I frequently encounter misconceptions about our
statistics, and the amount of value we place on numbers of users or
machines running our software.

We continue to be completely open and transparent about the ways we
gather statistics and the ways we present them.  We don't document
these statistics for purposes of competition, but because we believe
our community and our sponsors are invested and interested in knowing
some of the end results of the work they do in Fedora.  We also use
these statistics to help us construct and refine additional
community-building strategies and initatives.

In particular, there are statistics available which show the number of
unique IP addresses that have checked in for updates for each of our
distributions from Fedora Core 6 up through Fedora 9 and current
Rawhide (and soon, Fedora 10).  Although totalling those numbers is
interesting, it is not meant to indicate a measure of users, only a
total number of connections to repositories.  We know that each of our
releases tends to be installed on machines located at 3 to 4 million
unique IP addresses.  Some of those connections may represent a
duplicated IP address from one release to the next.  However, that IP
address could mean:

* one machine that has been upgraded to a newer release
* two or more machines owned by the same person behind a NAT/router/firewall
* two or more machines owned by different people behind a NAT/router/firewall
* two or more completely separate sites where the IP address has been
re-used (cable/DSL pool)

Obviously this makes determining the total install base of Fedora
across all releases somewhat difficult.  We understand completely that
IP address counting is not a scientifically valid way of determining a
total number of users.  That's why we don't claim a number of users
from these counts; we only present them as what they are, sums of
unique IP addresses.

Anyone who's ever heard me speak to this issue knows it's never been
my intention, nor interesting at all to me, to debate over user
statistics.  I am extremely satisfied that we have a geometrically (in
some cases exponentially) growing number of account holders,
contributors, and Ambasssadors involved in Fedora, all of which
numbers we can openly and transparently document.  This is far more
compelling for the community, I think, than simply throwing large
round numbers about, especially when those numbers aren't supported by
completely open, transparent, and documented recording and reporting

Please keep this in mind as you provide feedback and information to
people about Fedora.  Our leadership position, I believe, is based on
the total contributions our community makes to the entire free and
open source software ecosystem, through our continuing, unwavering
policy of upstream collaboration, and our continual efforts to lower
barriers to contribution across the entire project.

Thanks to all of you for being part of Fedora, and I hope all of you
are looking forward, as I am, to the release of Fedora 10 on Tuesday!


More information about the Fedora-marketing-list mailing list