Page MenuHomePhabricator

Generate statistics on recent beta feature rollout of 2017 wikitext editor
Closed, ResolvedPublic1 Estimated Story Points

Description

In T176181, users who enable "Automatically enable all new beta features" now have the 2017 wikitext editor beta feature enabled. The number of users of the feature as listed on the beta features page has not changed, which appears to only count explicit opt-ins for that one feature. Let's find out more.

Statistics to find out:

How many users now have the 2017 wikitext editor beta feature as a result of T176181?

  • How many users had the beta feature before T176181 rolled out, e.g. on 2017-10-04?
    • 26k unique global accounts (31k local accounts, with duplicates caused by people enabling it for themselves on a number of wikis); data
  • How many users had the beta feature after T176181 rolled out, e.g. on 2017-10-06?
    • 🤷🏽‍♂️ I don't think this is obviously findable, sadly.

How has the edit rate with the 2017 wikitext editor been?

  • Daily edits with the 2017 wikitext editor, over the past few weeks
    • We don't track this globally, but I can give you data for enwiki and dewiki (and with VE for comparison that the general edit rate didn't just spike):
day2017 wikitext editor edits, enwikiVisualEditor edits, enwiki2017 wikitext editor edits, dewikiVisualEditor edits, dewiki
2017-09-2563954622551880
2017-09-2640355991441746
2017-09-276176296961680
2017-09-2860859861111713
2017-09-2970654261781473
2017-09-305624513941191
2017-10-0151949112081492
2017-10-0280259852441546
2017-10-0370060232061615
2017-10-0471658541871608
2017-10-0582164182151863
2017-10-06100856441981780
2017-10-0782746552091381
2017-10-0897049213701618
2017-10-0998359573421908
2017-10-10106461833321840
  • Note that this is an under-count, because edits with multiple tags won't show up in my simple query.
  • Monthly edits:
    • with 2017 wikitext editor
month2017 wikitext editor edits, global
2016-11-010
2016-12-0115105
2017-01-0130063
2017-02-0138888
2017-03-0151981
2017-04-0151824
2017-05-0162697
2017-06-0166823
2017-07-0170932
2017-08-0168366
2017-09-0176269
  • with visual editor (nice to have)
monthVisualEditor edits, global
2015-01-01421706
2015-02-01383412
2015-03-01430406
2015-04-01423720
2015-05-01418767
2015-06-01410361
2015-07-01367321
2015-08-01404758
2015-09-01475512
2015-10-01569435
2015-11-01596792
2015-12-01602275
2016-01-01704706
2016-02-01691000
2016-03-01776487
2016-04-01741169
2016-05-01744527
2016-06-01693734
2016-07-01651368
2016-08-01679732
2016-09-01697150
2016-10-01776219
2016-11-01822716
2016-12-01791157
2017-01-01894827
2017-02-01844265
2017-03-01986061
2017-04-01852119
2017-05-01894564
2017-06-01809426
2017-07-01787130
2017-08-01767475
2017-09-01762554
  • with 2010 wikitext editor (nice to have)
    • We can't track this, sadly. My rough estimate is ~50% of non-bot edits, but that's based mostly on experience and there's no hard data.

Event Timeline

Deskana moved this task from To Triage to TR0: Interrupt on the VisualEditor board.

This is important information to be included in the quarterly review slides.

I can definitely give you monthly data on edits with the 2017 wikitext editor tag.

It should also be relatively easy to count the number of user receiving the beta feature. It sounds like this is (number of users explicitly opted in) + (number of users who automatically enable all beta features). This data would naturally come per-wiki; do you need a single global number? That would add some work for deduplication.

In T177597#3665536, @Neil_P._Quinn_WMF wrote:

I can definitely give you monthly data on edits with the 2017 wikitext editor tag.

Ideally I'll need daily edits too, for the past two weeks or so. T176181 only rolled out yesterday, and the quarterly review is next week, so monthly wouldn't help with that.

It should also be relatively easy to count the number of user receiving the beta feature. It sounds like this is (number of users explicitly opted in) + (number of users who automatically enable all beta features). This data would naturally come per-wiki; do you need a single global number? That would add some work for deduplication.

Per-wiki would be fine. I don't think we need to spend effort de-duplicating.

I'd also be interested to know the global number too, if it's easy to get.

@Neil_P._Quinn_WMF I've updated the task description to make sure we're on the same page. Does that look good to you? Thanks so much!

Daily edits would be harder, since I haven't pre-aggregated daily data the way I have with monthly edits. I can try for that, but given everything I have to do between now and next Thursday, I don't think I can promise it. It might be a bit easier to do it only for a couple key wikis; are there a few you would particularly like to see?

In T177597#3666349, @Neil_P._Quinn_WMF wrote:

Daily edits would be harder, since I haven't pre-aggregated daily data the way I have with monthly edits. I can try for that, but given everything I have to do between now and next Thursday, I don't think I can promise it.

No problem. I will appreciate whatever you can give me. :-)

It might be a bit easier to do it only for a couple key wikis; are there a few you would particularly like to see?

The largest wikis would be my priority, so English and German Wikipedias.

Deskana set the point value for this task to 1.Oct 12 2017, 2:19 PM
Deskana added a subscriber: Jdforrester-WMF.

Thanks @Jdforrester-WMF!

I'll close this task once I've got these suitably incorporated into the appendix in the slides.

Deskana claimed this task.

I never ended up incorporating these statistics into the slides, but they are useful nevertheless. I'll close this task now since there's nothing else to do here.