Here's the first pass at this:

I spoke to @egardner about this yesterday and he suggested a tab layout for the "media from other articles" images, which I think makes a lot of sense:

• It's extensible and gives us some interesting new IA opportunities to play with in the future, potentially.
• Usually I wouldn't use tabs because tapping requires more cognitive load than scrolling, but in this case we want to hide the other images behind a button press anyways, so this makes it more discoverable while still requiring reader action to see them.
• We can "hint" about the number of images the action would reveal.

I still have a couple open questions, though:

• Roughly how many other images will come back in this section?
• What's a good name for this section?

TBD:

• What is the text under the image?

Another open question: how do we handle images that appear in multiple language editions?

e.g. "From German Wikipedia and 13 other language editions of Wikipedia"

We want to avoid a combinatorial explosion in i18n messages for these strings.

In the example:

From German Wikipedia and 13 other language editions of Wikipedia

We'd need specific translations for every individual wiki name in every possible language we support, as well as the rest of the string overall.

We should do what we can to avoid this. If we can get by with the wiki domain (which will be the same across languages) instead of a formal name (which must be translated separately, because it is a variable), we'd be better off.

So something like:

Used on de.wikipedia.org and 13 other language editions of Wikipedia

Would be much simpler to internationalize. The de.wikipedia.org part could even link out to the exact image on the relevant wiki if desired.

I think it's worth to highlight that we can leverage 2 types of image suggestions:

1. article-level ones are meant to serve articles that don't have images (as a general rule, with some exceptions), so they shouldn't be suitable for other articles. For instance, this query lists all English Wikipedia articles for which article-level suggestions are available;
2. section-level ones target article sections, so they can also cater for articles that already have images. This query is the section-level equivalent to the above one. We usually don't have many suggestions here, due to some heavy filters that remove sections for which we don't want images, typically tables and lists.

We can build different explanations based on:

• article-level source
  • Wikidata - Used on Wikidata item ...
  • Wikipedia - Used on ... and ... other languages
• section-level source
  • sections aligned across Wikipedias - Used on the equivalent section on ...
  • blue links in sections - Found on a blue link in section ...
• section index or heading for section-level suggestions, i.e., we can link to the relevant section.

CC @JScherer-WMF.

I spoke to Olga on this again this morning. We're going to try a lazy load for the first release of this. If I recall, @egardner this was your initial preferred solution as well. Olga made another good point, which is why I also want to add a "back to article" button to the bottom of the image list to avoid a "dead end" in the experience if they reach the bottom.

Rationale:

• the kinds of users we're hoping to serve with this feature are more likely to want to scroll and we assume they won't give an extra action to click a tab or load button in order to see content from other wikis.
• The sensitivity around viewing images from other wikis may be mitigated by the fact that readers will need to scroll through all of the available images in the source article before seeing any images from elsewhere.

In T402269#11128009, @mfossati wrote:

We can build different explanations based on:

• article-level source
  • Wikidata - Used on Wikidata item ...
  • Wikipedia - Used on ... and ... other languages
• section-level source
  • sections aligned across Wikipedias - Used on the equivalent section on ...
  • blue links in sections - Found on a blue link in section ...
• section index or heading for section-level suggestions, i.e., we can link to the relevant section.

CC @JScherer-WMF.

Thanks for this @mfossati !

That's really good to know for us moving forward on this. I don't think we need to surface the complexity of article-level versus section-level suggestions to readers. I don't think they'll grasp the significance of the different sources.

I think we can just say...
If image is used in one place: From xyz.wikipedia.org
If image is used in multiple wikis: From xyz.wikipedia.org and XX other language editions of Wikipedia
If image is from Wikidata or one non-wikipedia source: From Wikidata
If image is from two non-wikipedia sources: From Wikidata and Wikivoyage
If image is from more than 2 non-wikipedia sources: From Wikidata, Wikivoyage, and Wikiquote

Does that make sense?

In T402269#11129165, @JScherer-WMF wrote:

I don't think we need to surface the complexity of article-level versus section-level suggestions to readers. I don't think they'll grasp the significance of the different sources.

I think we can just say...
If image is used in one place: From xyz.wikipedia.org
If image is used in multiple wikis: From xyz.wikipedia.org and XX other language editions of Wikipedia
If image is from Wikidata or one non-wikipedia source: From Wikidata
If image is from two non-wikipedia sources: From Wikidata and Wikivoyage
If image is from more than 2 non-wikipedia sources: From Wikidata, Wikivoyage, and Wikiquote

Does that make sense?

Yep, agreed!