Page MenuHomePhabricator

Feedback on Wikistats 2 new edits pages
Open, Needs TriagePublic

Description

Here is some (late) feedback on new Wikistats 2 edits pages. First I'm glad to see new pages added.

  1. Top editors (a.k.a most prolific contributors)

See example for Italian wikipedia:
https://stats.wikimedia.org/v2/#/it.wikipedia.org/contributing/top-editors/normal|table|1-Month|~total

1a What is group bot vs name bot. I can't find explanation and I'm not familiar with this breakdown.

1b I can't navigate to previous month, right? (Wikistats 1 couldn't either, for many metrics), expect via Wayback Machine (with irregular intervals)

1c presenting name 'Anonymous user' on breakdown for top editors filtered for 'Anonymous' makes this list not so meaningful, right?

1d1 Filter by page type, 'content' vs 'non content' . I assume content is articles and non content is all else (talk pages, help, category etc). May not be obvious for user, and I can't find explanation.

1d2 Does it follow API list which contains 'countable namespaces' per wiki?

1e A nice (minor) addition could be direct link to UC = user contributions for any given user name.
My hunch is this is what people want to see next. Knowing LauBot is most active in Oct with 127,298 edits get more meaning by a quick glance over https://it.wikipedia.org/wiki/Speciale:Contributi/LauBot (user can do this via side menu, so adding UC is only an added convenience).

1f Top editors is a useful metric. Top article creators would be nearly as important. Could be an extra filter: Filter by create/update. Some bots add hundreds thousands new articles. Might be extra useful to patrol those once in a while. Wikistats 1 does make this distinction. Compare https://stats.wikimedia.org/EN/BotActivityMatrixEdits.htm and https://stats.wikimedia.org/EN/BotActivityMatrixCreates.htm

1g Button 'Table' doesn't do anything

  1. Top edited articles (pages with most edits, not most contributors):

Again, example for Italian wikipedia:
https://stats.wikimedia.org/v2/#/it.wikipedia.org/contributing/top-edited-pages/normal|table|1-Month|~total

2a 'Top edited pages' seems to me a not so useful metric. Top scorer is Joe123 with https://it.wikipedia.org/w/index.php?title=Utente:Joe123/Sandbox&action=history
So this person massively plays in the sandbox. Who cares? The next 30 or 40 are administrative pages edited by bots.

2b For comparison please look at ZeitGeist table: https://stats.wikimedia.org/EN/TablesWikipediaIT.htm#zeitgeist
This tells which subjects were hot in any given month. It doesn't count edits, but editors. Again I'm less interested in how many edits any user made to https://it.wikipedia.org/wiki/Vichai_Srivaddhanaprabha. I do find it interesting that no other topic got more traction in Oct 2018 on wp:it. That's comparable in a way to the top most viewed articles which Signpost publishes and analyzes in every issue. This ZeitGeist deserves a page of its own. It's the kind of stuff reporters find interesting as well. Keeping the 'pages with most edits' will do little harm except it dilutes the overall relevance of the stats page. (same can be said of columns in Wikistats 1 pages, some columns are on hindsight better omitted, for sake of brevity and conciseness).

Event Timeline

ezachte created this task.Nov 23 2018, 8:33 PM
Restricted Application added projects: Internet-Archive, Analytics. · View Herald TranscriptNov 23 2018, 8:33 PM
Restricted Application added a subscriber: Aklapper. · View Herald Transcript
ezachte updated the task description. (Show Details)Nov 23 2018, 8:39 PM
Nuria added a comment.EditedNov 26 2018, 7:10 PM

1a What is group bot vs name bot. I can't find explanation and I'm not familiar with this breakdown.

Having in mind that we still have pending our tasks about supporting annotations in wikistats: https://phabricator.wikimedia.org/T178015

The bot classification comes from wikistats1 but clearly we might need better names for it? https://wikitech.wikimedia.org/wiki/Analytics/AQS/Wikistats_2/Metrics_Definition#Editor-type_-_{anonymous,_group-bot,_name-bot,_user,_all-editor-types}

1b I can't navigate to previous month, right? (Wikistats 1 couldn't either, for many metrics), expect via Wayback Machine (with irregular intervals)

It is not possible to look at previous month in the UI. That functionality will become available once we deploy a time selector. Metrics are, of course, available in API since the beginning of times.

1d1 Filter by page type, 'content' vs 'non content' . I assume content is articles and non content is all else (talk pages, help, category etc). May not be obvious for user, and I can't find explanation.

Right, that is documented here: https://wikitech.wikimedia.org/wiki/Analytics/AQS/Wikistats_2/Metrics_Definition#Page-type_-_{content,_non-content,_all-page-types but until we add annotations UI will be less than optimal

1d2 Does it follow API list which contains 'countable namespaces' per wiki?

Yes, that information is derived from site matrix upon ingestion and metric calculation.

1e A nice (minor) addition could be direct link to UC = user contributions for any given user name.

Indeed. We have created a task for this:T210422

Top editors is a useful metric. Top article creators would be nearly as important.

Thanks for the suggestion. We have created a task for this: T210423

Button 'Table' doesn't do anything

Indeed. We have created a task for this: T210424

2a 'Top edited pages' seems to me a not so useful metric. Top scorer is Joe123 with https://it.wikipedia.org/w/index.php?title=Utente:Joe123/Sandbox&action=history
So this person massively plays in the sandbox. Who cares? The next 30 or 40 are administrative pages edited by bots.
2b For comparison please look at ZeitGeist table: https://stats.wikimedia.org/EN/TablesWikipediaIT.htm#zeitgeist

You made this point before and we totally missed it, "top edited pages" should had been done in terms of editors not "edits". Reopened ticket: https://phabricator.wikimedia.org/T204965

group-bots vs name-bots, Ah I see, makes sense to makes this distinction, at least internally. Not sure if it's relevant to the end-user. AFAIK using the string 'bot' in user names is solely allowed for actual bots since long (users try to be funny with user names all the time, so not-really-a-bot names were a common occurrence but no longer). How about explicit(ly defined) vs implicit(ly defined) bots? My 2 cents.

Milimetric moved this task from Incoming to Radar on the Analytics board.Nov 29 2018, 5:56 PM