"Mobile devices" in the sense of T189554, i.e. clients that by default see the mobile site (more technically: that our Varnish code, according to a user agent regex there, will redirect from e.g. en.wikipedia.org to en.m.wikipedia.org unless a cookie indicates they opted into desktop).
Limited to the wikitext editor because those edits are easier to detect/count in the webrequest table, at least for the desktop interface.
Here is a quick and dirty query for a related question: The ratio of wikitext edits using the desktop interface made from iOS and Android devices (a good approximation of the "mobile devices" in the sense of this task). It is around 5% on average.
This will need to be compared with the number of mobile web edits (ideally also filtering out those which were made, conversely, from desktop devices, although that might be trickier to determine).
year month day mobiledevicesratio 2018 8 1 0.0436471068561674 2018 8 2 0.04785027686078897 2018 8 3 0.04889076420120272 2018 8 4 0.052364802730797244 2018 8 5 0.04910686305233469 2018 8 6 0.04664318673825397 2018 8 7 0.047794931664016424 2018 8 8 0.04737463087516748 2018 8 9 0.046592882821020316 2018 8 10 0.04738172826453639 2018 8 11 0.053164619044115846 2018 8 12 0.04893385422354363 2018 8 13 0.04622028477725041 13 rows selected (1674.773 seconds) via SELECT year, month, day, SUM(IF(user_agent_map['os_family'] = 'Android' OR user_agent_map['os_family'] = 'iOS', 1, 0 )) / SUM(1) AS mobiledevicesratio FROM wmf.webrequest WHERE year = 2018 AND month = 8 AND day <=13 AND http_method = 'POST' AND uri_query LIKE '%action=submit%' AND http_status = '302' GROUP BY year, month, day ORDER BY year, month, day LIMIT 10000;
Here is a quick and dirty query for a related question: The ratio of wikitext edits using the desktop interface made from iOS and Android devices (a good approximation of the "mobile devices" in the sense of this task). It is around 5% on average.
@Tbayer just to clarify, is that 5% of all edits get made from mobile using desktop mode?
@ovasileva I mentioned to Tilman that I think it'd be useful to know, more generally, what percentage of mobile page views happen with desktop mode, since we're interested in non-editing editor/contributor behavior. Thoughts?
Not quite - the 5% does not refer to the share of these desktop-mode-but-mobile-device edits among all edits, but to the share of mobile devices among edits made using the desktop mode wikitext editor. (A bit more precisely, among those edits that are not tagged as a mobile edit, or as a VisualEditor edit, or were made via the API like most bots or automated tools do.) That said, these desktop-mode-wikitext edits form the majority of non-bots edits on most projects (cf. this data). So the ratio you refer to is probably not much lower than 5%, either. Keep in mind though that I only used an approximation of "mobile devices" here so far (namely, Android and iOS user agents),
@ovasileva I mentioned to Tilman that I think it'd be useful to know, more generally, what percentage of mobile page views happen with desktop mode, since we're interested in non-editing editor/contributor behavior. Thoughts?
As mentioned in our conversation, that is the converse of T189554: Determine current percentage of the mobile site among desktop device pageviews, i.e. we'll be able to derive it from that result once we have it.
@ovasileva I mentioned to Tilman that I think it'd be useful to know, more generally, what percentage of mobile page views happen with desktop mode, since we're interested in non-editing editor/contributor behavior. Thoughts?
As mentioned in our conversation, that is the converse of T189554: Determine current percentage of the mobile site among desktop device pageviews, i.e. we'll be able to derive it from that result once we have it.
I think what @alexhollender is referring to here is rather the percentage of pageviews from the desktop site among mobile device pageviews (I think it's possible we can use the same setup to get this result as well?)
I don't quite understand this objection. By "converse", I did not mean that this is the same number, but that it can be derived from it (since we already know the number of mobile site pageviews and desktop site pageviews from our regular metrics, and mobile devices are complementary to desktop devices). In any case we should conduct this discussion over at T189554, as this task is about edits rather than pageviews.
@kzimmerman - I think we're okay here for now. Might request this again once we're deploying advanced mobile contributions for all wikis, but can open a separate task for that later.