Given the fiasco happening on Freenode (1), with the acquisition and transfer of control, it would be best if all wiki-related channels were moved to irc.wikimedia.org along with cloaks and bots. Right now, irc.wikimedia.org is home to recent-changes channels, for example #en.projectname, and is view-only.
IMO move from freenode should be discussed on meta and/or some mailing lists and not Phabricator. Technical solutions based on community consensus can (and should) be tracked on Phabricator when there are some.
Having said that, hosting an IRC network is a fairly large task with some downsides (for example, SRE incident response being difficult if our network would be down), I'd much rather move to libera (founded by many ex-freenode-staffers) or some other network.
It is a reasonable thing to think about, but not a trivial task. The current irc.wikimedia.org service is in no way setup to actually be a read-write IRC network. The existing service is read-only for clients and only fed data from MediaWiki recent changes hooks.
Building a functioning IRC service with nickserv and chanserv and cloaks and spam fighting tools it possible, but a non-trivial amount of work. The movement could ask the Foundation to start some of that work, but it is also worth considering if IRC is actually what is most needed if we are going in a fully self-hosted direction. Freenode has been very useful not just because the Wikimedia movement is there, but also because a large number of similarly aligned open culture/open knowledge/open software communities are there as well. If we build an island for Wikimedians, I'm not sure that I would suggest that IRC is the best modern solution. IRC has a lot of great features, but it is also a platform that requires a non-trivial amount of personal investment in selecting and operating a bouncer and clients to provide a full featured experience. There are many competing libre and gratis chat platforms which reduce this complexity for the typical user.
I also fully support the move to IRC.Wikimedia.org over any other third (3rd) party options, I am sure that these projects have enough tech savvy volunteers that can set the server up and do it properly. Just because they aren't running as a read-and-write IRC today doesn't mean that they can't tomorrow.
@Aklapper , I can't participate in the discussion there and judging by how my last unblock request went I am sure that I won't be able to ever discuss anything there again.
The issue remains that putting faith in any third (3rd) party will mean that such a situation can happen again, and we could be having this discussion again a decade from now. It's not like the Wikimedia Foundation doesn't have the resources to create such a tool, nor is an IRC that data-demanding, it is an old lightweight system that can easily be locally created.