Presumably this is blocked by T112171: Building Wikipedia portal mockups and prototypes?

If "knowledge engine" means what I think it means, then this is an attempt to b the all-singing-all-dancing solution to the Internet's problems. Surely the more 'domestic scale' issues like getting our own talkpages into the 21sr century, or Structued Data on Commoms, or even integrating Wikidata into Wikipedia infoboxes, are things that underpin any attempt to build a 'knowledge engine'? If we can't yet crawl, why are we trying to run?

In T114263#1780515, @Wittylama wrote:

If "knowledge engine" means what I think it means, then this is an attempt to b the all-singing-all-dancing solution to the Internet's problems.

"Knowledge engine" was the umbrella term used to refer to Discovery's roadmap in grant proposals. We don't use the term any more, precisely because of the confusion it caused; we use "Discovery" now. I'm not certain why people are still clinging on to the term in spite of this.

Surely the more 'domestic scale' issues like getting our own talkpages into the 21sr century, or Structued Data on Commoms, or even integrating Wikidata into Wikipedia infoboxes, are things that underpin any attempt to build a 'knowledge engine'? If we can't yet crawl, why are we trying to run?

Things like crawling and indexing other sources are on the roadmap, but they're years out, so there's a lot more thinking to be done first. Maybe we'll decide we don't want to do it anymore because things have changed. Maybe we'll decide we want to invest even more resources in it. We don't know yet, and it's normal to have that amount of fuzziness in a roadmap that's years out, because things change.

It's not correct to say that we're trying to run before we can crawl, because we are well aware that our own existing search system sucks and that we need to improve it. We're doing that, with things like the completion suggester beta feature, and with the many A/B tests we're running to improve our search for our multilingual users. Building a roadmap which assumes that smaller, earlier things on it go well and has other, more ambitious things on it further out is a totally normal practice.

And, on reflection, almost all of what I said above is pretty off-topic for a task management system like Phabricator. If you have any questions, feel free to email the Discovery mailing list. :-)

This appears to be largely duplicative of T112171: Building Wikipedia portal mockups and prototypes, so I'm marking it as such.