[Osdc-edu-authors] POSSE SA lead-in article on opensource.com

Karsten Wade kwade at redhat.com
Thu Sep 30 16:21:33 UTC 2010


On Wed, Sep 29, 2010 at 07:07:57PM -0400, Matthew Jadud wrote:
> On Wed, Sep 29, 2010 at 18:11, Karsten Wade <kwade at redhat.com> wrote:
> > but I welcome any feedback.  In particular, I propose the idea of
> > tracking Fedora download mirror data in South Africa as a way to see
> > the growth of contributors following an event such as POSSE.
> 
> I think this is a bad statistic. You have absolutely no way to say
> that A even correlates with B, let alone that there is a causal
> relation. For example, if a new, murderous warlord rises up who takes
> the name "Ubuntu," it may drive adoption of Fedora. So, the fact that
> a POSSE is run means almost nothing in terms of downloads. You can do
> it if you want, but the statistics will never mean anything.
> 
> Also, you can't even track institutional IP address ranges to get this
> data. For example, the ways I'm integrating open communities into my
> classroom don't really touch Fedora downloads in *any* significant
> way. Yet, I'm involving students directly in open processes as best I
> can. The mindshare impact can only be measured on long timescales...
> like years.

Yeah, that was pretty much my back-of-the-mind niggling feeling as I
worked on this.  It is too speculative for opensource.com at this
time.  I need to take that content section out and make a separate
post on my personal blog.  That's good because I already have you and
Jef Spaleta asking and poking at these sort of points, so getting them
in my personal blog will create a node for hashing this out.  Just for
expediency, I may take some or all of your points above (attributed,
if you don't mind) and include them as discussion points in that blog
post.

Meanwhile, I'm editing the article to remove the bad stats pathway and
just make it more speculative.  It *is* true that Africa appears to
have the fewest contributors per continent, but that data isn't
adjusted for population density, and I haven't drilled down to South
Africa in particular (looking at a global heat map doesn't provide
enough resolution.)

> Or, you need a way to actually correlate an increase in traffic with
> your outreach efforts, which is hard to do... I mean, what is the mean
> and deviation over the last 5 years (or however long you have data)
> for the region? That tells you how much things have to change before
> you can even claim it is "not normal" based on past behavior... (I
> think. I'm not a great statistician.)

Agreed on this one, for sure -- I'm presuming in advance that the data
might show such-and-so, but it would need more correlated data, such
as you describe, to be worked in/filtered by.

> ===
> Everything past this point is really just a "marker" for further
> conversation that really doesn't apply to this particular article,
> given the timeframe. Perhaps it should become a blog post that could
> be responded to later? However/wherever you think it is best to kick
> off some more conversation on this topic.
> ==

I'm going to take this gift and reply separately to those points, just
in the interest of time (I have to actually go do those edits I
referenced above.)

- Karsten
-- 
name:  Karsten 'quaid' Wade, Sr. Community Gardener
team:                Red Hat Community Architecture 
uri:               http://TheOpenSourceWay.org/wiki
gpg:                                       AD0E0C41
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/osdc-edu-authors/attachments/20100930/ab675831/attachment.sig>


More information about the Osdc-edu-authors mailing list