Once a community is known to an instance it is available via the search feature. Thus this really doesn't improve discoverability at all assuming the person adding it to the feed is already using the instance.
What it however does is moving the conscious choice of looking for and joining a community to an opaque follow feed button that makes someone subscribe to a lot of communities they know nothing about other than that someone else thought they somehow fit to a single word tag (and it is worse than hashtags on Mastodon as it is not the person making the post that adds them, but a totally unrelated 3rd party).
Yes the All feed has the same problem, but posts need to be significantly more popular for them to even register in the All feed. Thus most small communities currently fly under the radar of the All feed, and if they do get a popular post it nearly always becomes a moderation nightmare.
Hashtags on Mastodon have a similar problem, having given rise to the universally dreaded "reply guy" issue.
I think most people on Lemmy haven't really thought this through and what the implications of such a feature are once it becomes widely used.
And no, the one that is doing the opt-in is the person creating the feed without asking the community that is being forcefully opted-in. Giving them the option to veto that is better than having them realize that they have been opted into something they don't agree with by being flooded with trolls and off-topic comments.
If I don't misunderstand then you can only add communities to these feeds that are already known to your instance, thus I don't really see how this solves the federated discoverability issues which are ultimately due to instances not being aware of each other at all.
Yes but only moderation actions on specific users, it doesn't say who did the moderation action.
That is unless you are an instance admin. It used to be different, but this was relatively quietly changed a few releases back and many long term admins haven't realized that as for them it still looks the same as before.
This will reduce the discourse quality significantly as it will bring in more drive-by comments from people not subscribed to the specific communities in question.
I hope there will be some way for communities to opt-out from this or maybe better require them to opt-in.
Just a note: you as an instance admin can see who did a moderation action, but this is not generally visible to other users. I personally don't mind this, but some moderators and instance admins consider this leaking of information that can lead to harassment. Might be better to not post screenshots like that.
Charity is not the same as mutual aid anyways, even though I have also seen "mutual aid" requests on the Fediverse that were clearly asking for charity.
Otherwise not really. Three years ago Conversations was quite good already, although the newer forks Cheogram and Monocles added some nice convenience features.
Xmpp itself works great. The slidge.im bridges are relatively new and your mileage will vary. Matrix, Discord and Telegram works ok, Signal & Facebook messenger have issues right now, WhatsApp is a bit tricky to set up properly.
Matrix servers have the problem of highly variable resource use.
Basically if you only use it for some light chatting with friends and family and some niche topic public rooms it isn't very heavy.
But if any user of your homeserver joins any busy rooms or uses the bridges to join busy public Telegram channels or such, it will quickly outgrow the resources of a reasonably priced VPS.
Personally I would rather recommend you to set up an xmpp server, which can include a gateway to Matrix and other services, but architecturally is much more lightweight and has better mobile clients.
Yes having that option more easily accessible would be much apprechated.