Indeed, Commons pages don’t produce nice embeds on Discourse. I don’t know what would need to happen on the Commons side to make this possible out of the box, but I guess it’s related to oEmbed or Open Graph support.
In the absence of this, it is perhaps possible to extend Discourse with OneBox custom code to support Commons.
For general embedding, I think Commons needs to be exposing better metadata (in the form of opengraph or whatever) and then it’d work with Onebox and lots of other systems (like WordPress’ embeds, at Diff).
I tried making a WordPress plugin (embed-wikimedia) for doing this a few years ago, but ultimately it feels like it’d be better to follow existing standards of metadata interchange than create separate fixes for each system that wants to embed this content.
In my opinion, even if Commons supported embedding, using only Commons links for embedding, would simply be extending the problem of being a “walled garden,” in a lot of ways.
I support posting videos on Youtube, I support embedding items from youtube on MS forums, I support pasting youtube links on group chats in Telegram, and generally I endorse, applaud, admire, and appreciate, any tangible efforts by WMF, to extends its reach, to expand our community, and to post items from WMF far and wide on multiple platforms, in order to extend our community as greatly and as widely as possible.
by the way, below is a barnstar which seems relevant. here is the actual description, as a direct quote, italics added, QUOTE: The Motivation Barnstar is awarded to users who
do a considerable amount to promote Wikipedia beyond the site, or
do wonders in “gee-ing up the troops” either through general motivation of Wikipedians, or
motivate Wikipedians by way of tools, contests, and other such things.
It’s not at all clear exactly which metadata system (or systems) we should support. It seems like oembed might be a bit of an unfashionable standard these days, and more sites support the Open Graph Protocol. There are other ways to expose the same metadata, e.g. via Dublin Core tags (which overlap a lot with OG).
So there needs to be investigation into that, and how it’d fit in with existing metadata (e.g. the PageImages extension and Wikibase’s SHORTDESC parser function).
Then the code has to be written, undergo security review (I presume, as it’d likely be a new extension), and be deployed.
It’s not massively complicated stuff, I think, but the devil is in the detail and I feel like there are a bunch of different things to be sorted out. And at the moment, it doesn’t sound like there’s anyone working on it.