Fedora 8 most popular (then 7?)

[Jeff Spaleta]

On Fri, Oct 31, 2008 at 7:49 PM,  <orion at cora.nwra.com> wrote:
> Is that telling us something?


I think you have to look at the unique IP mirror manager stats as well
to get an idea of whether the smolt opt-in stats are valid as a metric
for overall adoption.  Do those stats support the same conclusion you
are trying to draw from the smolt data?  Can you reconcile what they
are saying?  I wish we had the maps up, they are snapshot of the
MirrorManager logs from the last week, so they give a different
picture versus the total unique ip count since release.

The sampling statistics for smolt's voluntary opt-in mechanism are not
necessarily statistically valid to extrapolate from a population
sampling theory point of view, since smolt users may be an
unrepresentative sample of the wider userbase.  Are certain types of
users more likely to use smolt than others? We've no statistically
valid survey of active workloads or usage scenarios.  Are people less
interested in optting into smolt now than they were a year ago? We've
no idea.  Is smolt usage itself trending downward with each release?

And likewise with the mirror manager unique ips.  That is clearly
going to be missing some people who reconfigure their systems to use a
static mirror.  Do we have any idea how many people do that? No. Do we
know if people are less interested now in using MirrorManager than a
year ago? No idea.  Is MirrorManager usage itself trending downward
with each release?

It's also interesting to look at this from the global map point of
view as a distribution of Fedora installs globally, instead of trying
to get a single number.  I think trending the global distribution of
clients matters more in terms of figuring out where to concentrate new
community building efforts.

Let me be very clear... no one in this project has stepped up and made
the case for an adoption increase goal for each release.  If you are
looking for a measurable uptick in adoption then you need to step up
and lead a user adoption effort.  Looking at the metrics for total
adoption goal numbers and parsing out small relative percentage
changes in adoption from release to release when we've no organized
effort to drive those numbers up is absolutely pointless.  Until the
goal of driving user counts up with each release becomes personally
important to someone, someone who will lead an effort, it will not be
important.

Let me be even clearer.  I personally do not care about seeing
adoption driven significantly higher for its own sake with each
release.  I believe in the concept of  the "right" users for
sustainable growth, I do not need to see user numbers driven upward
just for the sake of pointing to it patting ourselves on the back. I
only care about sustainable growth, where users become contributors
and help sustain new efforts.  Which is why I care about the map
densities as a tool to see where in the world new community effort can
be incubated.

-jef