Re: The number of popcon.debian.org-submissions is falling
Petter Reinholdtsen wrote:
> The number of submissions to the Debian popularity-contest collector
> is falling, and has done so for some time now. This can be easily
> seen on <URL: http://popcon.debian.org/stat/sub-i386.png >.
> This is mostly caused by a fall in the number of Lenny installations,
> as can be seen from
> <URL: http://popcon.debian.org/stat/release-1year.png >.
> Anyone got any idea how to can get more machines to report to
> popcon.debian.org? Or can there be some other problem causing the
> fall in the number of submissions?
> Happy hacking,
(I'm not a native speaker and my writing style might feel paternalist or
lesson-teaching: that's not the intention - add a "IMHO" in front of every
... The facts ...
The measurement tools we have⁰ are now somehow showing a decrease of
Debian's direct userbase: less popcon reporters, less debian-news
As for popcon, many have explained their reasons for not enabling popcon on
their machines. Some other temptatively explained what could increase the
popcon reporters ratio within our userbase. But I chosed the word "ratio"
carefully: we can try hard to increase the ratio of Debian users with popcon
enabled (by enforcing it on the new installs, by making the reporting easier
or less verbose): this will only get us a more precise measurement of our
userbase: this doesn't make it bigger.
I think we should now face the facts: what we long feared is now probably
beginning to happen: our direct userbase is shrinking.
... Widening our view ...
But again: words. Our _direct_ userbase is shrinking. Our biggest
derivatives — Ubuntu — sees a constant growing in popcon reporters (more
than one order of magnitude both in absolute and in growth). And it has been
established that around 70% of Ubuntu packages are shipped unmodified: all
those users are using directly our packages!
Recently a bunch of Ubuntu-derived distributions decided to base their work
on Debian instead of Ubuntu (gNewSense, Linux Mint, OpenGeu, Crunchbang, …).
Those distributions now get closer to us: there will be a step less between
our packages and theirs.
All those distributions carry their own set of users and developers. Those
developers know the Debian packaging techniques (not always up to Debian's
quality, but still…); their users are using most of our packages unmodified.
I feel that many of the recent new DDs had grown their skills on children
distros before looking up (or down if that matters) to Debian; to finally
become DDs: we would be seeing a flow back of developers from children
distros to Debian?
All in all, although our direct userbase might be shrinking, it might be
good to be reminded that our wider userbase is not: more and more people are
using our packages. Sure they are not Debian users nor Debian popcon
reporters, but they still use (and sometimes enjoy) our work. And that's
Based on that, we might feel that we are only giving without getting back
(in form of bugs, quality, users, developers): that's where we have a word
to say and it's our turn to play.
... The future ...
Although we might have always thought Debian as an distribution "made for
real users", it has always had derivatives. Debian now has the most
successful ever distribution as derivative! Ubuntu is not an users sink
only: it is a chance we must handle! We certainly lost tons of users in
favor of Ubuntu, but they still use our work, our packages: for most of
Debian has no parent distribution: it is the sane root on which many others
base their work in order to achieve different goals than Debian's: wide
adoption for a commercial success¹ (Ubuntu), more ease to get non-free
software (Mint), Enlightenment-based (Opengeu), live-cd for easy boot
(Knoppix), education (Skolelinux), phone (Hackable:1), … [There are many
more examples]. Each derivative is basing itself on the rock-solid base of
Debian and enhances (in its view) one particular aspect of it in order to
achieve a more specific goal. That's free software and it's very good.
This very special position of Debian within the distributions ecosystem is a
great chance: a great part (not to say the majority) of GNU/Linux users are
direct _or_indirect_ users of Debian.
We can't anymore consider our direct users as our only users: all those
derivatives are also special Debian users and their users are indirect users
of Debian: our direct+inderct userbase is growing!
We must now teach all those derivatives how easy our processes and tools
are, how kind we are with informed bugreporters and with patch providers,
how open we are to integrate all sorts of improvements (even distribution-
specific²), how both universal and flexible Debian (as distro and as humans)
can be. At the top of all that: we must teach to all our derivatives that
contributing to Debian directly :
1) doesn't forbid to keep a difference/specificity (not painful),
2) not only benefits the derivative but a) Debian and b) all other
derivatives (good for them, good for us, saves the whales too),
Finally, I think that finding patches to get more popcon users is useless;
finding ways to attract more users is good, but the most important is
informing and teaching our derivatives on how they can work with us and why
And the new firstname.lastname@example.org is a very good step in that direction.
Cheers, and thanks for reading,
⁰ As everywhere else, measuring a phenomenon is altering it: when we inform
(and it's of course a good thing) our users that we want their data (huh!?)
for statistical purposes, we get a biased measurement (with an unmeasurable
² See e.g. dpkg-vendor hooks in debian/rules