Re: User Estimates

[glen e. p. ropella]

Holly Arrow writes:
 > 
 > Averaging these numbers seems odd -- why average the number of
 > unique posters PER WEEK (?) with the number of people on swarm-support?  

Well, if you take the number of unique posters per week for a sample
of time, you get an idea of active users.  I did it "per week" because
the longer the interval over which you wait, the more unique posters
you'll get, possibly up to the number of people on the list.  I wanted
to narrow the active users down to a minimum.  A week was arbitrary.
I could've used 1 day, but that would not be enough to contain most
threads, which can span 2 days.

Then if you take the number of people subscribed to the list, you get
an estimate of the maximum number of active users, presuming that most
active users are members of the list.  Now we have a minimum and a
maximum number of users.... take their mean and you have an admittedly
lousy estimate of the number of users.

 > I think you need to define what you mean as the user community (who
 > counts as a "user"?), and partition it into different levels of membership
 > status represented by swarm-support, downloading 1.0, and
 > going to SwarmFest.  

Yep.  But, I don't have a definition.  Hence, I'm just batting things
about until something converges. [grin]

 > Thinking of core and periphery makes more sense, with fuzzy boundaries
 > between core & periphery and between periphery and non-community.

This sounds like a good way to think about it.  I assume that the
core are the people who post to swarm-support on a regular basis.
I do that because, even if there are alot of users out there using
Swarm but not posting to the list, their impact on the user community
is either non-existent, not measurable, or invisible (where invisible
means long-term vs. short-term).

The periphery is then defined by other things like downloads, web
hits, membership on the lists, etc.  And it might make sense to
further subdivide that periphery into "moderate" and "ancillary."

 > I agree with using some multiplier of SwarmFest attendance to count people
 > who are equally involved but couldn't afford the time/money or had a
 > scheduling conflict.  

Also, since it's easier to attend a user meeting than it is
to actually install Swarm or program in Swarm, it depends 
on whether you're measuring "interest" in Swarm or "usage"
of Swarm as to whether or not using SwarmFest attendees is
a good thing to do.

 > That's my 2 cents as a social psychologist and self-declared peripheral
 > member hoping to move toward the core when I have time this summer ...

Thanks!

glen