It was a difficult issue.
We finally decided to filtered out all annonymous users, consistently, in all our studies (and also in the results I will present in my Ph.D. thesis).
NAT and proxies prevent us to trace down individual users correctly, since the database only stores de IP address, and many "real" users may be hidden behind the same IP.
The problem may be the same for logged users (one real user with multiple accounts) but this is less common, and even discouraged due to possible suspicious behaviors:
http://en.wikipedia.org/wiki/Wikipedia:Sock_puppetry
Cheers,
F.
--- El lun, 17/11/08, Desilets, Alain Alain.Desilets@nrc-cnrc.gc.ca escribió:
De: Desilets, Alain Alain.Desilets@nrc-cnrc.gc.ca Asunto: RE: [Wiki-research-l] "Regular contributor" Para: glimmer_phoenix@yahoo.es, "Research into Wikimedia content and communities" wiki-research-l@lists.wikimedia.org Fecha: lunes, 17 noviembre, 2008 4:14
Interesting. So, in summary:
- Most edits done by a small core
- But, most of the text created by the long tail
- However, most of the text that people actually
read, was created by
the small core
Is that a good summary of what we know about this
question?
I think so :).
BTW, what is the status of anonymous edits in those studies?
Are anonymous edits excluded from these studies altogether?
If they are included, is anonymous treated as a single highly productive "user", or is it treated as being part of the long tail?
Alain