OK. I'm less worried about aggregate data than retaining information about an identifiable device or user. For example the information that the English Wikipedia article "Crimea" has 251 watchers is ok to record and publish, but we probably shouldn't log that user:BarackObama logged in at 10:04:34 EST with UA xyz after linking to the Wikipedia page from the the Huffington Post, watchlisted the article at 10:05:40, and typed "Angela Merkel" into the Wikipedia search box, and ended his session at 10:06:12. This concern may get into subjects that are covered by the new Wikimedia Privacy Policy and I would be interested in hearing from an expert about what non-edit data is logged and tied to a specific user including non-logged-in users who are browsing Wikimedia sites.
If no changes are being made to mobile that would affect checkusers then should we move this discussion to Analytics-l only?
Thanks,
Pine
From: jalexander@wikimedia.org
Date: Thu, 27 Mar 2014 23:58:03 -0700
Subject: Re: [WikimediaMobile] Mobile and CheckUser (Was [mobile-l] Wikipedia App User Agents)
To: deyntestiss@hotmail.com
CC: analytics@lists.wikimedia.org; mobile-l@lists.wikimedia.org
I'll let someone more knowledgeable about current practices then me answer that question to a better extent so that I don't butcher it but I know that some of that info gets in with differing levels of sampling and filtering and retention. The UA data, for example, is how we know information like what browsers visit the site (
http://stats.wikimedia.org/wikimedia/squids/SquidReportClients.htm ).
James