Page MenuHomePhabricator

Identify highest priority key docs
Closed, ResolvedPublic


Of the docs with “yes” in the “needs review” column in the key docs spreadsheet (restricted access), which are the highest priority for us to make significant improvements to before launching the Dev Portal? Let’s try to keep the list to around five docs.

Event Timeline

Thanks, the suggestions make a lot of sense to me. Personally I'd love to also see T293771 but that looks blocked by WMF's lack of interest/prioritization to coordinate docs across teams.

My votes: (sorry there are 6 of them)

I agree that Research:Data ( could use some love, but I think to do that thoughtfully and effectively with the Research team would require a longer timeline than we have for this; we should prioritize it in the next round. Same for Wikimedia infrastructure (‏‎ (106 links).

Data I looked at as part of this comment:
A couple months ago I used and to get data about the pages on (I'm not going to redo the spreadsheeting now b/c the numbers aren't that old and I need to save time: (restricted spreadsheet with that slightly-stale data)). Caveats and disclaimers: a page being heavily linked-to isn't necessarily a direct indicator of its importance or utility, there are nuances that are probably inflating these numbers overall, and one of the functions of the dev portal is to help people find the most useful docs, which may be buried and thus NOT linked to heavily. That said, the number of links going to a page is at least a data point to compare docs where we otherwise lack such data.

Today, I also looked at, with the thinking that revisions counts are another signal of page importance and utility. (Caveats: bots, vandalism, etc etc etc). And, I repeated the aforementioned process (minus the spreadsheet creation) for wikitech, meta-wiki, and wikidata.

Taking all of that into account, and cross-referencing it with our list of key docs, here is my reasoning that resulted in the prioritization above:

These mediawiki API docs are all important but they're also in relatively good shape compared to some of our other docs to review:

  • API:Main page‏‎ (2,591 links)
  • API:Client code‏‎ (2,099 links)
  • API:Picture of the day viewer‏‎ (1,949 links)
  • API:Article ideas generator‏‎ (1,948 links)
  • API:Nearby places viewer‏‎ (1,947 links)
  • API:Holidays viewer‏‎ (1,946 links)

Manual:Pywikibot‏‎ has 938 links while other Pywikibot docs have...

  • Manual:Pywikibot/Wikidata‏‎ (752 links)
  • Manual:Pywikibot/Scripts‏‎ (709 links)
  • Manual:Pywikibot/Overview‏‎ (687 links)
  • Manual:Pywikibot/Installation‏‎ (685 links)
  • Manual:Pywikibot/Communication‏‎ (677 links)
  • Manual:Pywikibot/PAWS didn't show up in the top 5 linked to subpages in the collection.

Manual:Bots‏‎ (not one of our key docs) is heavily linked to, with 27 interwikis; it links to both Manual:Creating_a_bot and Manual:Pywikibot. But, the latter is in much better condition -- Creating_a_bot is quite overwhelming but contains really useful info.

New developer things:

Wikitech docs that showed up on the most-linked list, but still seem less crucial than other docs:

  • Wikimedia infrastructure‏‎ (106 links)
  • Deployments‏‎ (95 links)
  • Help:Cloud Services introduction‏‎ (77 links)
  • Portal:Toolforge/Quickstart‏‎ (65 links)
  • Help:Toolforge/Developing successful tools‏‎ (64 links)(review already in progress)
  • Blubber‏‎ (19 links)

Considered but not very useful info:

  • How to report a bug‏‎ (32 interwikis)
  • Phabricator‏‎ (24 interwikis)

Footnote: I also reviewed but didn't get any useful insight from:

apaskulin claimed this task.

Thanks, Andre and Tricia! Based on your comments, it seems like we agree on these four docs:

Let's call these our tier-1 priority docs. If we have extra time, we can put some work into these two, tier-2 docs:

Some notes on pages not included here:

Research:Data (

I agree that this page requires maintainer outreach, so we should address it when we have more time.

Developer account (

This is one of our most translated pages, plus it's already concise and well-formatted, so I don't think it needs significant improvements.

Resolving this task, but feel free to comment or re-open!