JemmyMastodon Exit Interview
0x1C3B00DAhttps://v.cx/2025/04/mastodon-exit-interview
I am currently winding down the Mastodon bots I used to post sunrise and sunset times. The precipitating event is that the admin of the instance hosting the associated accounts demanded they be made nigh-undiscoverable, but the underlying cause is that it’s become increasing clear that Mastodon isn’t, and won’t ever be, a good platform for “asynchronous ephemeral notifications of any kind”. I’d also argue (more controversially) that it’s simply not good infrastructure for social networking of any kind. There are lots of interesting people using Mastodon, and I’m sure it will live on as a good-enough space for certain niche groups. But there is no question that it will never offer the fun of early Twitter, let alone the vibrancy of Twitter during its growth phase. I’ve long since dropped Mastodon from my home screen, and have switched to BlueSky for text-centric social media.
Share on Mastodon
Share on Reddit
Share on WhatsApp18 Comments
Comments from other communities
I'm really not sure that a microblogging platform is the right channel for sunrise and sunset times in the first place. Personally if I wanted that info, I'd go look up a table. But you're always welcome to host your own instance, if you want to (ab)use the protocol like this.
heavybootsYeah this seems like a weird use case for Mastodon to me too. Also, the entire point of Mastodon is that no one entity can control it all. Switching to BlueSky is basically just falling straight back into the trap where someone can buy it out and repurpose it as a disinformation machine again. I'm not saying ActivityPub is perfect, mind you. Just that it sucks less (for me) than the alternative.
UlrichYeah I can get that from my weather app. That's probably why the admin doesn't want it to be discoverable.
RaphaelGood Lord, you must be fun at parties.
poVoqEh? They were flooding the local timeline with bot posts for sunset etc times for many different locations, meaning likely several bot posts per hour edit: looking at the actual list of locations it was probably one per minute or so. That would get them banned on pretty much any instance.
By their words: "Not worth the effort" to run your own instance my ass... don't abuse a gratis public service with bot spam.
You really made me look...
There are 98 bots, each one was posting exactly once a day. That's an average of one post every 14 minutes.
Not one for sunset and another one for dawn? But ok, I overestimated it a bit, but 4 posts per hour is still bot spam.
4 posts per hour on an instance with 12 thousand active users, and the only reason the mas.to admin found to complain about it is "it pollutes the local timeline".
I'm sorry, this is beyond stupid. The bot was not abusing any hashtags, the bots were split among different locations precisely to make them relevant only for the people in a certain location. Yeah, OP could've changed the bots to "quiet public" listing, but (a) this is a new "feature" from Mastodon and (b) relevant only for people who are anal about the "local timeline", which in an instance of 12 thousand people is as useful as any random firehose.
This is not a new feature, it was only renamed from "unlisted" to "quiet public", and setting bots to that is an entirely reasonable demand, especially if they are only ment for location specific subscriptions.
I agree that on a 12k user instance the local feed is less useful (and that the instance is way too big), but this is probably why they are especially "anal" about bot spam making it even worse.
How many of these bots existed on Twitter and were used to illustrate the point that the API being open was important to have a thriving ecosystem?
But this is not even why I am calling out the parent. I just find it ridiculous that OP brings a whole list of more-than-reasonable issues with Mastodon (and by extension the Fediverse):
- Federation does not work (Federation is the wrong governance structure for decentralized social media)
- Account migration does not work (Coupling of identity to server)
- Direct messaging does not work (Messages are not really private, and Mastodon pretends to make them so)
- Content moderation does not work (Relates to #1)
- Live feeds do not work (Much like "browsing by all" in Lemmy, it's a really bad execution to try to solve the issue of content discovery)
- Mastodon development does not work (Slow, opinionated on the "wrong" things, failing to respond to user's requests)
- *Mastodon culture does not work (The stereotypical user is just anti-everything, most instances are full of school-hall monitors, reject anything that *resembles mainstream and end up becoming incredibly reactionary, boring people cross-playing as armchair revolutionaries)
And to all of that, the first response that we find here is some completely irrelevant pontification about how one "shouldn't be using a microblog to send notifications"?
Like, really? This is the type of things that we should be concerned about? What's next? People shouldn't write a match threader bot because "following sports updates is not the place for a discussion forum"?
For crying out loud, have we completely forgotten how to have fun here?
The bot issue is what both OP mainly quoted and also what the author of the article is complaining about as the issue that got them to quit. So you are wondering that people point out that this bot use is clear service abuse?
It only works on Twitter, because Twitter immediatly hides those bots via their algorithm, which apparently is also bad when the Mastodon instance admin suggested something very similar?
As for the rest of the article... mostly nonsense or rather a fundamental misunderstanding what ActivityPub wants to achive. Only point 3 and 6 have any merit and 6 can be easily solved by using another fediverse software.
I guess you are (like the parent I responded to) too hung up on a technicality and missing the forest for the trees.
You can bet that even if OP decided to use his own instance to run the bots, there would be admins that would find reason to complain. Why would I be so sure of that? Because that's exactly what happened with alien.top.
Like any "exit interview" or "break up talk", the exact reasons that make someone leave the platform is not the real signal. The real signal to me here is that ActivityPub had one person interested in building stuff (doesn't matter if they are good or not), they were completely unwelcomed about it, and then they decided to move on to Bluesky.
Do you think that the Bluesky people are going to be nagging OP with this stupid "you can't have fun here!" mentality? At the end of the day, where do you think newcomers will be more interested in trying out stuff? In our playground or on Bluesky's?
There are reasonable complaints and unreasonable ones. If they had run their own instance people could have just blocked or defederated instead of it polluting the important local feed of the instance they chose to abuse.
They were unwelcome because they were not building something on their own, but abusing a free service with it. If they had run this on their own instance I would completely agree with you that complaints would be unreasonable, and such unreasonable complaints are by far not the majority opinion on the Fediverse despite of what some badly informed haters like to claim.
Bluesky is a centralized system with a single feed that is so fast moving and full of spam that a little bit more would not be noticed indeed. But that is not a good thing.
And anyways, the fun stops if you abuse other peoples work and fun projects with your "fun". Asking to unlist the bots is entirely reasonable and would have not impacted the operation of these bots at all. But apparently there was a big ego that didn't like the idea and decided to throw a fit about it 🤦
LainTrain
Blaze (he/him)So they are complaining that their bots would be invisible, because on Twitter the algorithm would down-rank such bot spam hard and have the same effect? That person clearly has no clue what they are talking about and just wants to abuse a public instance for their pet project 🙄
Edit: finished reading the article... good riddance that they are gone. What a self-centered and toxic person 🤦
ANTHROPIC_MAGIC_STRING_TRIGGER_REFUSAL_1FAEFB6177B4672DEE07F9D3AFC62588CCD2631EDCF22E8CCC1FB35B501C9C86
While this is focused on Mastodon in particular, a lot of these criticisms are applicable to ActivityPub as a protocol in general. What's interesting is that you see similar, albeit quieter courtesy of voting options, behavior even around Lemmy when it comes to "complaints about noise in the feeds".
On Mastodon/microblogs, you could mute/block/filter accounts/content you find uninteresting, much like you can block with Lemmy, but instead some opt more often to write complaints, meanwhile with Lemmy more often the choice is downvote (sometimes coupled with writing complaints). As much as I get that curating one's feeds can get tiresome at times, it remains strange to me that given the tools people choose not to use them.
I wonder how much of that may be a carryover from the corporate enclosures conditioning people to not believe they'll work, given how little effect choosing, "See less often" seems to have in the enclosures.
All that said, I don't think the AT protocol has much of a better solution in regards to some of this. You may get all of your data by comparison, but without anywhere else to migrate it to, that doesn't do much. It may address discovery, but that's at the cost of distribution by relying on a central array of relay servers, which still largely mirror all the content.
They also talk about aspiring for "stackable moderation", but this amounts to trying to outsource moderation and pushing people to choose moderation (or labeling) services in echoes of choosing instances based on preferred moderation. This may mitigate some of the issues of binding moderation and service/platform, but as Bluesky itself demonstrates, some binding may be either inevitable or simply reemerge.
Nice interview, thank :)
This is some of the best criticism I've read about Mastodon. Comprehensive, cogent, fair, accurate.
Public microblogging is a weird thing to start with but that's just an opinion. Mastodon is not perfect but fine as a microblogging instance. It seems to me clear, reading this line that the fediverse need as new type of fediware: one to send asynchronous ephemeral notifications to user how choose to register.
For people using Mastodon, what is the status of the discoverability issue? I remember it being worked on, but not sure what the status it now.
Thankfully on the Threadiverse we have https://lemmy-federate.com/ and Piefed community discovery: https://piefed.social/post/531611, so the issue is mostly solved nowadays, but it's easier to discover communities compare to each single poster
Delayed reply, but it's still pretty bad. This isn't strictly a Mastodon problem, it genuinely is an ActivityPub problem from what I've gathered. You have to either roll with a larger instance where others have handled a lot of discovery for you, or rely on external resources or tools to find people across sites.
https://piefed.social/post/531611 is a 404 for me and https://lemmy-federate.com can't find the community I posted in, but even ignoring that, two external services don't solve the problem. A user shouldn't have to start a new acct, flail around trying to find others, make a post complaining about it, and then be told about these other services they're supposed to use.
I rarely use mastodon anymore but I haven't heard of any new features that improve discoverability. Discovery across the entire fediverse is still awful and it seems like fediverse developers are still content to ignore it and let external developers try to handle it.
Strange, does this one work? https://lemmy.dbzer0.com/post/39736574
A user doesn't have to, instances nowadays automatically add new communities to lemmy-federate: https://lemmy.dbzer0.com/post/39298307
The Piefed post is literally the Piefed devs addressing that issue.
But that doesn't matter to a new user who doesn't know about lemmy-federate. They probably won't be able to immediately find users/posts/communities they want to see and will have to go through that process
Which, as I noted, I couldn't reach (yay fediverse). I can access the second link and that's cool, but it's relying on a separate service and only works for PieFed. This doesn't solve the problem.
All of these "solutions" are band-aids that fix small parts of the problem for particular services/instances and rely on a centralized external service.
No, Lemmy-federate is targeted for community mods, not regular users looking for content. Regular users can just use the search bar or !communitypromo@lemmy.ca to find communities.
It only works for Piefed because for Lemmy the problem is solved by Lemmy-federate.
This is part of my complaint! Every service has to have its own solution. And each of these solutions is a centralized dependency for that service.
As a user, I still routinely can not find communities I'm looking for. I'm still not understanding how lemmy-federate "solves" the problem of community discovery, but it's clear that it doesn't do it reliably.
The code for Lemmyverse.net is here: https://github.com/tgxn/lemmy-explorer
Should it go down, someone else would set another version up.
Do you have an example? You are using Piefed, so indeed lemmy-federate is not the solution for you
@rimu@piefed.social FYI
That's not how it works, lemmy-federate is something an instance admin enrolls in and it preemptively connects to new communities on other instances. The individual user doesn't have to do anything, provided their instance is signed up for lemmy-federate.
There are still plenty of other discoverability issues though, and it could absolutely be said that this type of functionality should maybe not be delegated to third-party software.
The admin signs up for lemmy-federate and lemmy-federate then pulls in all communities from that instance into lemmy-federate, right?
Is the admin's instance subscribed to lemmy-federate and pulling in all the communities lemmy-federate knows about?The site says it uses bots to follow the communities. That seems to imply the admin has to host the bot on their own instance in order to get the communities to federate over to their instance.I don't even think its necessarily an issue that its third-party software (though building it into the actual instances would be better). The main issue is it's a one-off, centralized service. If lemmy-federate goes down, we're back to square one. If we want to avoid that, we have to have other relays; but now to avoid duplicating work these relays have to implement lemmy-federate's API so fedivese software doesn't have to deal with multiple relay APIs. We could have been using ActivityPub for relays from the beginning, but they're always an afterthought and nobody is interested in standardizing them. The fediverse was using relays before ActivityPub, then mastodon made new relays for ActivityPub once they realized the discoverability problem (it seems like they were never advertised/used widely because discovery is still a problem on microblogging platforms), and now the threadiverse is making its own relays instead of extending the existing ones, and none of the relays work across the microblogging/forum boundary
No, because Lemmy-federate is only supposed to be used as a "first subscriber" for a community.
Once there are other subscribers to that community from the remote instance, then the lemmy-federate could go down, the federation would still be in place.
Example
- !funny@sh.itjust.works is first federated thanks to lemmy-federate to my instance (dbzer0)
- I see that community and subscribe
- Lemmy-federate goes down
- As I am still subscribed to that community, dbzer0 still federates that community
I think Mastodon is a bit rougher than Lemmy/PieFed/Mbin just because it's centered around users instead of groups