Oh I've heard.

But it's not surprising. Mastodon mainly sees the fediverse as a mastoverse with "other things attached" and gargron mainly sees mastodon as his personal project with "other people attached". It's a bit harsh, but there's enough truth to it, especially for an ecosystem/platform that has as its core ideology cooperative diversity and connectivity.

All of that being said, I think a good amount of blame falls to the protocol. With a good enough protocol and set of guarantees around what compliance with it entails, it shouldn't matter whether someone like gargron behaves in this way (where to be fair, he's built by far the biggest platform over many years now), as compliance with the "standard" protocol should preclude his ability to arbitrarily push things around like this.

But the protocol is very soft. And I think a time is coming, if it isn't here already, where the creators of the protocol are going to rue how much they've left up to developers like gargron who will suck the "standard" out of AcitivityPub. I say this may already have happened already because there are calls (which are very reasonable IMO) to take the mastodon API and basically make it a standard, not least because other platforms are adopting as a de facto standard already. And of course the biggest, sometimes desperate/emotional push back against this I've seen has been from one of the authors of the ActivityPub (evan, if you know them), because there's apparently something in ActivityPub that should be used instead but no body as really heard of and which I've heard is kinda crappy and vague (not unlike the rest of the standard perhaps?).

Bottom line is that ActivityPub needed gargron and mastodon to prove that something good could be made with it and they may regret the bed they've made. What ActivityPub should have done from the start is dedicate efforts to the creation of standardised software, at least for testing purposes, so that there was some sort of standard at the functioning software level for developers to work against. THere are efforts along those lines now, but it may turn out to be too late and one of those things every fediverse developer is "supposed" to use and contribute to but don't bother because they're too busy getting their platform/app to work with mastodon.

EDIT:

And on that note, an interesting project in the fediverse is vocata (https://codeberg.org/Vocata/vocata), because it tries to make the server as generic as possible and instead require the client (ie browser or mobile app) to make all the decisions.

Well sure, glass half full or half empty. To be clear, my problem isn't with either lemmy or a.gup.pe ... they're open source projects doing great things IMO, and I hope I've not made any of their devs upset.

My problem is with what seems to me

as a mostly technically naive user that tries to pay attention to what others who know better are saying

a tendancy on the part of the fediverse to not live up to its hype or promise, which, AFAICT, is attributable to the nature and design of the protocol and how federation works on the fediverse.

And sure, we can celebrate something half working ... there is, as you say, definitely something cool in seeing federation happen.

But there's a reality here that is too easy for open source projects to ignore ... user experience matters.

Unless you've got a great big sign on the app/platform that warns any user that this thing will not work at any random time (which OSS users implicitly understand but not everyone does) anytime the UX breaks in unexpected, uncontrollable and incomprehensible ways, a user has an upsetting experience. Sure, it's a bit of a first world problem, but the promise of social media is to connect socially with people, which involves an expressive and emotional behaviour. Moreover, people are drawn to the fediverse for emotional reasons: join the "good place", do the "right thing", help build a better internet etc.

The moment something unexpectedly breaks on someone as they're earnestly trying to reach out and make this place work for them, something rather frustrating and a little heart breaking happens.

In this case, there's something iffy about the protocol itself that things like this can "half work".

You ask whether it's better that it doesn't work at all ... and I'd say yea, probably. For the simple reason that to some user "half working" won't be discernible from "fully working", creating an expectation of functionality that will inevitably be disappointed. What happens when someone posts to the group and it doesn't get federated and they get no replies or interaction? They're gonna think there was something wrong with their post. If it happens a few more times, there's gonna be some psychological effect, however much we don't like to think about how emotionally bound we are to social media. What if they praise that a.gup.pe is on lemmy and tell everyone they should use it to only realise down the line that they've told people to try something that doesn't work well and feel embarrassed and frustrated that they've dedicated time, energy and their personal endorsement on something that's clearly got problems in the foundations? That's not a bitter experience that turns people off of caring about the fediverse? Why wasn't there someone warning everyone that the system wasn't working? Why allow something to only half work without telling anyone about it?

If this doesn't make sense to you ... I can assure you it happens ... people have gotten very upset over things just breaking down here. And part of the difficulty isn't that the system is inconvenient and hard to use ... it's that when it breaks it breaks in mysterious ways for the user, in ways that are undocumented, unknown and sometimes misconstrued and misrepresented by others.

I don't mean to demean the developers and their projects. But I do think broken things looking like working things is awful user experience, and that ActivityPub in the way that it seems to be a vague and very flexible protocol contributes to exactly these sorts of problems and should be criticised for it.