“[R]ecent revisions of an article can be peeled off to reveal older layers,
which are still meaningful for historians. Even graffiti applied by vandals
can by its sheer informality convey meaningful information, just like
historians learned a lot from graffiti on walls of classic Pompei. Likewise
view patterns can tell future historians a lot about what was hot and what
wasn’t in our times. Reason why these raw view data are meant to be
preserved for a long time.”
Erik Zachte wrote these lines in a blog post
<https://web.archive.org/web/20171018194720/http://infodisiac.com/blog/2009/…>
almost
ten years ago, and I cannot find better words to describe the gift he gave
us. Erik retired <http://infodisiac.com/back_to_volunteer_mode.htm> this
past Friday, leaving behind an immense legacy. I had the honor to work with
him for several years, and I hosted this morning an intimate, tearful
celebration of what Erik has represented for the Wikimedia movement.
His Wikistats project <https://stats.wikimedia.org/>—with his signature
pale yellow background we've known and loved since the mid 2000s
<https://web.archive.org/web/20060412043240/https://stats.wikimedia.org/>—has
been much more than an "analytics platform". It's been an individual
attempt he initiated, and grew over time, to try and comprehend and make
sense of the largest open collaboration project in human history, driven by
curiosity and by an insatiable desire to serve data to the communities that
most needed it.
Through this project, Erik has created a live record of data describing the
growth and reach of all Wikimedia communities, across languages and
projects, putting multi-lingualism and smaller communities at the very
center of his attention. He coined metrics such as "active editors" that
defined the benchmark for volunteers, the Wikimedia Foundation, and the
academic community to understand some of the growing pains and editor
retention issues
<https://web.archive.org/web/20110608214507/http://infodisiac.com/blog/2009/…>
the movement has faced. He created countless reports—that predate by nearly
a decade modern visualizations of online attention—to understand what
Wikipedia traffic means in the context of current events like elections
<https://web.archive.org/web/20160405055621/http://infodisiac.com/blog/2008/…>
or public health crises
<https://web.archive.org/web/20090708011216/http://infodisiac.com/blog/2009/…>.
He has created countless
<https://twitter.com/Infodisiac/status/1039244151953543169> visualizations
<https://blog.wikimedia.org/2017/10/27/new-interactive-visualization-wikiped…>
that show the enormous gaps in local language content and representation
that, as a movement, we face in our efforts to build an encyclopedia for
and about everyone. He has also made extensive use of pie charts
<https://web.archive.org/web/20141222073751/http://infodisiac.com/blog/wp-co…>,
which—as friends—we are ready to turn a blind eye towards.
Most importantly, the data Erik has brougth to life has been cited over
1,000 times
<https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=stats.wikimedia.org>
in the scholarly literature. If we gave credit to open data creators in the
same way as we credit authors of scholarly papers, Erik would be one of the
most influential authors in the field, and I don't think it is much of a
stretch to say that the massive trove of data and metrics Erik has made
available had a direct causal role in the birth and growth of the academic
field of Wikimedia research, and more broadly, scholarship of online
collaboration.
Like I said this morning, Erik -- you have been not only an invaluable
colleague and a steward for the movement, but also a very decent human
being, and I am grateful we shared some of this journey together.
Please join me in celebrating Erik on his well-deserved retirement, read
his statement <http://infodisiac.com/back_to_volunteer_mode.htm> to learn
what he's planning to do next, or check this lovely portrait
<https://www.wired.com/2013/12/erik-zachte-wikistats/> Wired published a
while back about "the Stats Master Making Sense of Wikipedia's Massive Data
Trove".
Dario
--
*Dario Taraborelli *Director, Head of Research, Wikimedia Foundation
research.wikimedia.org • nitens.org • @readermeter
<http://twitter.com/readermeter>
Hi everybody,
as you already know dbstore1002 is going to be migrated to dbstore100[3-5]
moving from a multisource (single host) scheme to a multi instance (multi
host) one. As part of the migration we have to upgrade the following DNS
CNAMEs (they all point now to dbstore1002):
s1-analytics-slave.eqiad.wment
s2-analytics-slave.eqiad.wmnet
..
x1-analytics-slave.eqiad.wmnet
Since they don't seem really used except from some old use cases (
https://github.com/search?q=org%3Awikimedia+s1-analytics-slave&type=Code),
so I thought to move to a different naming scheme (removing the word
"slave" where possible):
s1-analytics-replica.eqiad.wment
s2-analytics-replica.eqiad.wmnet
..
x1-analytics-replica.eqiad.wmnet
The new CNAMEs will point to the right dbstore hosts of course. Details in
https://gerrit.wikimedia.org/r/#/c/operations/dns/+/488004/
Please let me know if this is going to affect any of your scripts or ETLs
or etc.. I don't personally think that there are any but better safe than
sorry :)
Luca
