2009/9/22 Mike Godwin <mnemonic(a)gmail.com>om>:
My own personal view is that, in an ideal world,
we'd post two or more
metrics for every project (article numbers, number of editors, and perhaps
other metrics like, perhaps, external links). That would create a design
problem given our current home page, but probably not an unsolvable one.
The idea here is that, with multiple metrics, we can hypothesize more
clearly about trends -- e.g., when the article number rate of increase
declines, but numbers of editors and external links increases, we may be
able to make some more reasonable guesses about what's happening on that
project.
Obviously, Erik Zachte's work in this are is extremely (I'm inclined to say
uniquely) valuable -- I'm wondering how we can better integrate his research
into how the projects initially represent themselves to users upon entry.
I don't know if we necessarily need multiple metrics on the home page,
but we certainly should be considering multiple metrics. To move from
just considering article counts to just considering participants to
population ratios would be a very bad idea. Do we have an expert
statistician around that can do some regression testing, or similar,
and work out what the real relationships are between these various
metrics? For examples, what kind of correlation is there actually
between number of participants and article creation rates? Does that
correlation vary for different sized Wikipedias (and for other
projects)? Etc. etc.