The Wikimedia Research Hackathon on August 6 and 7 takes place parallel to
the general Wikimania Hackathon in London.
Wikimania Hackathon information is available at
Research Hackathon information is available at
>From the Research Hackathon info page: this "is an opportunity for anyone
interested in research on wikis, Wikipedia, and other open collaborations
to meet, share ideas, and work together. It's being organized by
researchers in academia and the Wikimedia Foundation, but we want anyone
interested in research to participate. Whether or not you consider yourself
a researcher, or would ever want to be one, come with questions, answers,
data, code, crazy ideas... or just your insatiable curiosity."
Local participation will occur at Wikimania London and in Philadelphia, PA,
US. Remote participation is possible and will include researchers and
community members globally.
Please see the Research Hackathon information page for scheduling and
Further questions may be directed to Aaron Halfaker (ahalfaker(a)wikimedia.org)
or Leila Zia (leila(a)wikimedia.org).*
*A $1 fine will be imposed by Oliver Keyes on anyone who misspells Leila's
name or misdirects emails to the WMF Executive Director.
Back in May I announced that we were starting a new project to try and ask
more anonymous editors to join the community and register accounts.
Since then, we've finished up our first A/B test on English, German,
French, and Italian Wikipedias. Yesterday we also launched our second A/B
test, this time just in English.
As a refresher, the first time we tested two different workflows. In the
first, called "pre-edit", we prompted people to sign up right after they
clicked edit. In the second, called "post-edit", we told people about the
advantages of signing up after they completed an edit as an IP.
To summarize the test results:
- The post-edit version was seen by much fewer people, and only gained us a
incremental number of newly-registered Wikipedians. The plus side is that,
because it didn't interrupt people, it caused no negative impact on the
volume of edits by IPs. The few editors we gained were more likely to reach
5+ edits than users in a control group too.
- The pre-edit version was much more interesting. It caused a very large
increase in new registrations: +200% in English and German, +300% in French
and Italian. These users who registered were also significantly more likely
to complete a first edit. Unfortunately, this version also decreased the
total volume of edits by IPs by ~25%. We have several theories about why
this was, and the A/B test we just launched is an iteration on this
pre-edit design to rule out some of those theories.
You can check out our full research report so far at:
1. Original announcement
Due to an issue with a smoke detector beeping me awake at 3,4,5 and 6am I
had to unexpectedly go into landlords office. running late now likely to
miss standup .
Mostly did code review yesterday and began moderated states design. Gonna
pair up with May today and get that and hopefully make a few additional