Does the plan for deployment of this feature include getting community consensus in advance?
I ask because I was considering starting a discussion on this at the English Wikipedia, but I don't want to mess up anything if someone with more involved here was planning to do it. If it's not planned, I'd also like to avoid a situation of this being started too early, causing discussion to circulate around remaining major bugs, instead of the real merits and issues of the feature as a whole.
In my experience, these kinds of discussions are far easier to have in advance of deployment, as the community atmosphere tends to be not at all conducive to calm discussion regarding the pros and cons of a change, when the discussion starts following a major change being pushed forward without sufficient discussion and consensus.
(See for example the enwiki discussions regarding the watchlist bolding, in which the same community reacted in two polar opposite ways to the same proposal, when first given the opportunity to discuss the proposal in advance (enthusiastic response) and when reacting to its unexpected implementation (about as angry as that community gets, strong "kill it with fire" attitude).)
Since the development seems still ongoing, and it looks like it might get a deployment date set independently by the developers, the discussion would probably be regarding whether, once the deployment date arrives, to keep the feature or disable it (either by Phabricator request or direct disabling).
However, I ask that you please consider simply not having a set "deployment" date at all, and have Hovercards as an available feature to be enabled upon request, once development has progressed far enough (marked "stable"). While discussion will probably successfully propagate to a number of wikis in either case, many will still be caught off-guard if an non-consensus-backed deployment is implemented. Communities in that situation tend to be more likely to respond with a "shut it down first, then we can discuss it" attitude, and it's generally much harder to get consensus for re-enabling a disabled feature than it is to get consensus for turning it on in the first place. Discussion in advance allows for useful feedback and suggestions, and allows for a much less contention between the communities and developers.
I am willing to help out in whatever way would be useful.