Fosstodon is now invite only
hub.fosstodon.org
hub.fosstodon.org
In your parlance, each Mastodon instance is "an app". The real apps, however, many of them can log into multiple accounts and change which account you're posting from.
It also isn't what people actually want, though.
Don't like it? You just excluded yourself as part of the intended userbase: Mastodon is not for people who don't grok federated service, and Mastodon doesn't want people who can't grok federated services.
Think of understanding simple concepts like this as sort of a "you have to be this tall to ride this ride" skill check gate. Either you get it, or you don't get in.
This is a big reason why I'm excited for Bluesky and ATProto: theirs includes many more touchpoints for federation and for control of the user experience without falling into the "server federation is so point to point that the entire thing feels like an empty landscape" problem encountered by many of the folks I know (myself included) who tried Mastodon and then bailed. Bluesky right now is pretty good, and when graph systems and personal data stores are federated along with the existing feed creators (which are a really awesome feature to use even in an early form), it'll be an interesting thing to witness.
Trying to conflate that with "xorg.conf sucks" misses the fact that Twitter is basically the X here (literally and figuratively), while Mastodon is the Wayland. A lot of X-lovers spend their time shitting on Wayland in HN comments because they don't understand the problem Wayland is solving, and don't understand it successfully solved it and is becoming the default, and it just makes them angry.
Also, I don't see the point of Bluesky. If their specifications prove to be useful, then federated services will start using those protocols. Anything they can do, Mastodon can do, too.
That's not true, though. Twitter offers valuable affordances that Mastodon does not support, and I don't mean quote tweets--I mean the ability to organically encounter conversations between people, one of whom you follow and one you maybe don't, and drift in and out of interesting conversations.
(On Twitter that is largely facilitated by quote tweets; Bluesky does support quotes, but also always surfaces replies from people you follow even if you don't follow the person to whom they were replying. This "overhearing effect" is awesome and helps gradually expand your social graph and find more neat people.)
> A lot of X-lovers spend their time shitting on Wayland in HN comments because they don't understand the problem Wayland is solving, and don't understand it successfully solved it and is becoming the default, and it just makes them angry.
What I mean is that there has been a ton of Mastodon evangelization, followed by fairly predictable and understandable user problems, followed by "well just edit xorg.conf" (or in our case here, "have multiple accounts on multiple servers and juggle all that"). It's the You Don't Need That, or the This Workaround Is Fine For Me And So It Must Be For You, or the Pull Requests Accepted thing.
There is a trap in consumer-facing OSS, where the evangelists don't realize that to compete you do have to be better in terms of the things that the evangelized care about, and because they've evaluated it doesn't mean that they're ready to contribute back, and that's the vibe I still get a lot here.
> Also, I don't see the point of Bluesky. If their specifications prove to be useful, then federated services will start using those protocols.
Maybe! Mastodon could do it, but I think that they'd do it by not being recognizably what is ActivityPub today. There's a fundamental difference in how ATProto envisions the use of the (federated, run-your-own-if-you-want) Big Graph Services[0]--a vision of which accessible today, in the small, with distributed feed generators that filter down the currently extant BGS--that the point-to-point peering of ActivityPub, the "here's a giant pile of mailboxes you're all sending stuff back and forth to", just doesn't really support. Firehose and filter, not mailbox.
The "ovehearing effect" is a big differentiator and is the straw that stirs the drink for Bluesky, for me and for a lot of people. Granted, there have been attempts to do something similar with AP relays, but that still requires server-level participation and, if they moved to per-user, you could happily crush most Mastodon servers under the load of receiving AP traffic from relay sources. Also the Mastodon frontend doesn't seem to really have a good way to integrate it into your own personal timeline, and I haven't seen movement towards it happening.
[0]: https://atproto.com/blog/federation-developer-sandbox#bgs
If course there's no reason to log in, read and log out of every account, every time you read, that would be silly when most people never bother logging out of any of their accounts to begin with.
Also, it's entirely feasible to just follow people on other accounts from whatever account you're on.
Heck, I had two twitter profiles because I had two different sets of interests. This isn't uncommon.
I hope you realize that this is a hilariously out-of-touch answer with respect to what most people want from a Twitter-alike.
If you don't care about that, then _you don't have to do those things._
To me it seems like the main question is the amount of friction encountered in switching accounts in clients. In more polished clients like Ivory this is just as easy as it is with Twitter.
The first two I have replicated on Mastodon. I have stopped using all the Twitter accounts, so I have now 'slimmed down'.
This gets the closest to having control over your online identity but, honestly, the ROI for self-hosting your own Mastodon server just isn't worth it. The requirements are too demanding for just a one-person (or a few more) instance.
Even with Pleroma/Akkoma/[other platform], you still have to jump through some hoops of complexity for someone new to the fediverse, let alone a Twitter refugee (even on the more tech-literate side).
Fully self-hosting anything is a big step for anyone without sysadmin experience, though managed hosting is a good intermediate option for those willing to pay for it.
I've already tried deploying Pleroma on Fly.io using a Dockerfile and successfully did it with someone's guide. I was hoping to get away with the Fly.io's free tier but it's still a lot of work to maintain for something that should be simple. With Pleroma and it's derivatives there is not even an option to reimport "backups".
Of course, it would be nice if somebody made a full-featured fediverse server that was self-contained with no dependencies, did backups through a web UI, and was generally as easy as possible for someone without expertise to self-host. Pleroma isn't designed for that. Neither is Mastodon.
Administrators have the limited scope of only their instance.
But that's the thing, I as an end user don't want to and shouldn't have to. Hosting my own instance is a pain in the ass that introduces needing a server to run it on, OS maintenance / updates, Mastodon instance maintenance / updates, paying for everything, ... It's a huge waste of my time and energy when decoupling identity from content solves it much more smoothly, for a much wider audience. If the goal of Mastodon isn't to attract that sort of audience, that's a separate issue, but that doesn't seem to be it from what I can see. Identity and content can both be federated without tying them together.
We don't have that yet. It'd be nice if we do someday.
So, Bluesky? :P (I know it's not federated at this point in time)
Of course, replicating this effect would be a dump truck of load on every Mastodon server, but it’s also the experience most people seem to want, so that’s unfortunate.
Yes, there is some locality aspect to the instance you belong to, but in general the instance you log on to has comparatively little impact on what other instances you can interact with.
This isn't absolutely true. The characteristics of a given instance do have an impact on what technical features are available to you. There are a number of supported advanced features such as toot length, Markdown and LaTeX formula support, etc., which apply to only a subset of instances. My primary instance, toot.cat, is based on Glitch.soc, a Mastodon fork, which supports both long posts and Markdown, both of which I (ab)use frequently. Mathstodon, hosted by our own @ColinWright, is amongst those supporting LaTeX-based maths formulae, for some inexplicable reason....
Another is that the moderation practices your instance follows will be reflected in how remote instances interact with yours. Ineffective or beligerent moderators are a fast route to being limited or outright blocked (there is a continuum of potential restrictions, it's not all-or-nothing) by others. And similarly, your local instance might have its own set of criteria for determining what traffic or patterns are or are not permissible in remote instances, again resulting in varying levels of moderation on that basis.
The "local community" element mostly shows up in the Local timeline. I ... hardly ever have that open, relying instead on a set of curated, priority-based Lists, or pinned Hashtags.
Lists are based on specific profiles you choose to add, and I've found that the best organisational method is not topical but rather is based on interest level. So I keep a relatively small top-priority list, a somewhat broader secondary, and for profiles who are overly chatty, a third which I view rarely. I follow nearly 1,000 profiles, but effectively see perhaps 50--100. I have a small set of topical lists, largely press/journalists and admins, though I use those rarely.
I'll also pin hashtags that I'm currently interested in, which might be for anything from a few minutes to months on end. You can also subscribe to hashtags as of recently, though I've yet to do this.
There are other topical tools including <https://gup.pe>, which function somewhat like a mailing list or forum. (There are newer variants which offer more powerful moderation options.)
But in general, the instance you're on, so long as it's well-managed and well-behaved, has little impact on who else you can interact with.
The misinformation, and disinformation, on this point becomes tedious with time.
And then you go on to enumerate various important ways in which it is not a gross misread at all.
"The characteristics of a given instance do have an impact on...":
"...what technical features are available to you"
"...how remote instances interact with yours"
"...criteria for determining what traffic or patterns are or are not permissible in remote instances"Secondly, the issues around moderation also matter, but:
- Blocks are reasonably rare, but generally implemented with good reason. Well-administered instances which play well with others tend very much not to get blocked.
- Increasingly, moderators are sharing blocklists through Fediblock (<https://fediverse.wiki/wiki/Fediblock>) or similar mechanisms. So if a block exists on one well-administered instance, it's likely to exist on numerous others.
- Blocked instances typically have a small number of members, though there are a few exceptions. (Large Japanese instances on which Loli features predominantly is the principle exception of which I'm aware, or alt-right sites such as Gab and Truth Social.)
- Instance moderation is on a continuum. "Limited" simply means there is limited federation, but that content won't appear on the Federated timeline, and that follow requests require an explicit approval. (I see this largely for profiles on infosec.exchange.) That's a relatively low bump.
- Toot.cat does moderate instances extensively, with 476 moderated instances listed at <https://toot.cat/about>. 138 of those are "Limited" (see above), 338 are suspended, meaning no federation occurs. The top 20 moderation rationales are:
50 Reason not available
18 Freeze peach instance
17 Transphobic admin
11 Admin using a wide variety of slurs
10 Admin using racist slurs
10 Covid denialism
8 Edgelord admin
8 Racist admin
6 Lamp instance(s)
5 Admin shitposting to #fediblock
5 Anti-moderation admin
5 Cryptocurrency-oriented instance
5 Harassment, slurs, fascism, etc.
5 Transphobic harassment from admin
4 Adming spamming #fediblock
4 Anti-LGBTQ+ admin
4 Conspiracy theory
4 Covid denialism and other disinfo
3 Admin using ableist slurs
3 Anti-moderation instance
I haven't (yet) checked to see how many profiles exist or are active (w/in the past month) on the blocked instances, but I strongly suspect that most are quite small, or no longer active. Just for grins, I'm looking at a random selection of 10 instances blocked by toot.cat ('sort -R | head -10' off the list of blocked instances):
neckbbeard.xyz: Server not found
hermitmountain.top: Timeout on Web access
dobbs.town: Timeout on Web access
ortodogs.info: Unable to connect
touha.me: Single-user
nnia.space: 282 active users (blocked as pedophile instance)
gamers.exposed: Appears inactive? Video site?
neenster.org: Unknown (users not reported). Principally Nina Paly's personal instance as I understand.
cybercriminal.eu: Timeout on Web access
mastinator.com: Read-only implementation, presumably no users.
Keep in mind that blocks are appealable, both by the remote instance and local users. And the threat of blocks is a strong incentive for both admins and users to execute and expect effective and socially-conscious moderation. Edgelords can play, but they'll play in a vastly reduced space.
That's a naive take, but I think it's worth moving on to something much more important...
> Blocks are reasonably rare, but generally implemented with good reason. Well-administered instances which play well with others tend very much not to get blocked...
I notice that you keep using weasel words like "reasonably" and "generally" and "tend" and "typically" with casually dismissive motte-and-bailey appeals to "well-administered" and "good reason" and "small number of members". You do not appear to be engaging in good faith.
Please address the fact from your own statements that your participation globally depends on whether another instance administrator likes your instance's administrator, and explain why you believe that's not an issue relating to separation of communities and server administration policies, and explain how that's not a direct result of demanding that identity providers be communities where absolute consensus dictates both administrative policy and also administrator personal behavior.
> The top 20 moderation rationales are:
> 50 Reason not available
> 18 [Someone who isn't you did something that someone else didn't like]
> 17 [Someone who isn't you did something that someone else didn't like]
> 11 [Someone who isn't you did something that someone else didn't like]
> 10 [Someone who isn't you did something that someone else didn't like]
> 10 [Someone who isn't you did something that someone else didn't like]
> 8 [Someone who isn't you did something that someone else didn't like]
> [etc.]
I hope you understand the absurdity here of binding identity provision and restrictions on who you personally are allowed to communicate with to disputes between distributed instance administrators that may have nothing to do with you personally purely by virtue of deciding that identity providers must necessarily be consensus communities.
The Fediverse presently has something north of 14 million registered accounts: <https://mastodon.social/@mastodonusercount/11110302515627643...>
I'd responded with what factors are distinguished between specific instances, noting specifically which of those affect broader community interactions and which don't. That's based on my own participation on Mastodon since 2016-7, across a number of different instances.
For the rest, I'm describing both in narrative and using actual instance and sampled data what the impacts of moderation are. You don't seem to like either my use of data, or how I narrate it, but I can in fact assure you that both correspond strongly to my experience over the past seven years. In particular. I'll note that the moderation rationales I'm reporting come directly from toot.cat, and that though I'm friends with the admin I'm not a moderator there myself. You're the one who's editorialised those rationales.
The impact of blocks is best measured by how many active Fediverse profiles are affected by them. That's something which instance block counts addresses poorly. My sampled check helps validate my own gut impression, again, that most blocks are against small, and often inactive instances. You're ... completely ignoring this fact in your umbrage.
I'd also make the argument that harassing people for their backgrounds, beliefs, and lifestyles which are in no way harmful to others is highly disruptive. You may choose to deny this, but that's not well supported by extensive experience and research.
A key function and attraction of the platform is that individual instances can make their own individual federation options. And yet, even with extensive moderation the total impacts seem fairly minor. I'll concede that current mechanisms make this less fully transparent than it might be, but the information is available, and in general issues around instance-level moderation is an ongoing area of Mastodon and Fediverse development. But if you're so opposed to such a fundamental aspect of the Fediverse, you'll probably not find the overall concept particularly attractive.
I'm interested in question of the magnitude of blocking in terms of affected profiles, by the way, and am looking for better ways to get a broader read of how widely blocks are used on different instances, and how many active Fediverse members are restricted by such blocks. Right now, that involves my manually looking up individual instances in a GUI browser session, though I believe there may be APIs which can help with that.
Just for grins, I'm looking at infosec.exchange, itself moderated ("Limited") by toot.cat (on the basis i.e. hosts law enforcement and intel agency accounts). Jerry Bell doesn't strike me as especially socially sensitive.
Infosec.exchange moderates 145 instances, 137 are Suspended, 8 are Limited.
Reasons (complete list):
114 Reason not available
3 Hate speech
3 Unmoderated racism
2 Part of free/hate speech network
2 Violent extremism
1 Admin uses hate speech
1 Almost exclusively an instance for posting porn.
1 An instance dedicated to bots that autopost animated/anime porn. Pretty massive draw on resources at best. Going to get someone fired for what’s in the timeline at worst.
1 At owner instance request
1 Blocked to resolve confusing bleed over due to Glaceon blocking infosec.exchange
1 Bypasses blocks and makes all toots public without consent
1 Generally awful free speech instance supporting racial and homophobic hate
1 Highly offensive and violent hate speech instance
1 Highly offensive content
1 Hosts far right content, offensive racial memes, hate speech, and clearly attempting to out MAGA truthsocial.com
1 Mutual silence in place
1 Racism promoted by admin
1 Should be obvious
1 Significant source of hate speech, racial slurs, and targeted harassment.
1 Supports/tolerates hate speech.
1 The instance appears to be up to no good. Block in place due to shenanigans from a user on that instance, will reassess.
1 Twitter scraper creating account mirrors
1 Under review
1 Unmoderated homophobia
1 Unmoderated racism and hate speech
1 Unmoderated spam and other nonsense
One of the "Limited" moderations is on toot.cat, as a mutual silencing.
I've not compared the T.C. and I.E. lists for shared blocks, though I see a few.
One of the largest Mastodon instances is mastodon.social (operated by Mastodon's author, Eugen Rochko). Moderated: 235, Suspended: 220, Limited: 15, rationales:
68 Third-party bots
64 Hate speech
43 Inappropriate content
23 Harassment
18 Spam
4 Misinformation
4 Transphobia
3 MAPs
2 Reason not available
1 Bot that announces blocks
1 Botnet
1 Conspiracy theories
1 Mirroring other Mastodon accounts without permission
1 Misinformation and conspiracy theories
1 Preparing for announced shutdown
582 1 (blocked by only one instance)
77 2 (blocked by two)
40 3 (blocked by all three)
I notice that you keep using weasel words like "reasonably" and "generally" and "tend" and "typically"...
Yeah, well, if I'd used absolute language, the obvious attack would be "Hah! I've found one example among the ~25k Fediverse servers which fails to follow the moderation practices you describe!" Which is to say, of course, that with roughly 25,000 individually-administered servers, of course there is variation in moderation practices. Again, this is a key feature of decentralised networks.
But again: in most cases the factors which call for, erm, moderation in moderation mean that excessively broad-brush actions tend not to be undertaken by larger, or even modestly-sized, instances, and that the total affected populations are relatively small. That poorly-administered instances face backlash (or departure) by their own affected members is another element to this.
That said, there are general tools for moderation, which are discussed in Mastodon's technical documentation, see: <https://docs.joinmastodon.org/admin/moderation/>
Those tools are there for a reason, their design is inherent to Mastodon's function (other Fediverse apps and servers have similar capabilities), and as with all regulation, tools which are effective when manipulated with skill and wisdom are of necessity dangerous when used carelessly or maliciously. That applies to every tool starting with the first scratch-stick and hammer-rock.
There are also recommended guidelines and practices, though much of that remains somewhat informal. Mastodon's creator, Eugen Rochko, has written on the topic however: <https://blog.joinmastodon.org/2018/07/cage-the-mastodon/>. Though broad, and focusing on technical details, Rochko details why features such as global default search ( People and brands search for their own name to self-insert into conversations they were not invited to") and quote-reply (Heed, my followers, how I dunk on this fool!). That is, there is some direction for how well-administered instances are expected to operate. The server rules of Rochko's own instance can also serve as a rough outline:
1. Sexually explicit or violent media must be marked as sensitive when posting
2. No racism, sexism, homophobia, transphobia, xenophobia, or casteism
3. No incitement of violence or promotion of violent ideologies
4. No harassment, dogpiling or doxxing of other users
5. Do not share intentionally false or misleading information
<https://mastodon.social/about>
Toot.cat posts a code of conduct, generally restricting: Personal attacks, Trolling or insulting/derogatory comments, Public or private oppressive language or actions, Other unethical conduct, red flags (detailed in a link). Again I'll note that the admin someone I've known for a long time, and she's run, moderated, or helped design community criteria for a wide range of diverse communities over decades, with infinitely greater patience and tolerance than I'd show myself.
<https://mew.toot.cat/mw/Toot.cat/CoC>
There are also freestanding guides to moderation (this is, again, a decentralised federated universe), such as "A Guide to Moderating in the Fediverse: The Basics" <https://tenforward.blog/a-guide-to-moderating-in-the-fediver...>. And, I'm sure, numerous others.
As to the predictable, and extant, objection that those rules are vague and subjective, I'll note that this is similarly the case on the forum we're discussing this matter on, Hacker News. The guidelines are kept intentionally brief, and much of the actual guiding lore is to be found in the many, many, many moderation comments by dang (> 47k comments), sctb (~6k), and pg (~10k) over the years. (Totals are all comments, some concern matters other than moderation.) This is by design. As dang writes:
Here's a brief refresher: Curious conversation is good. Substantive comments are good. Thoughtfully sharing personal experience is good. Flamebait is bad. Personal swipes are bad. Ideological boilerplate is bad.
Those are excellent ideals for virtually any healthy community in my view.
<https://news.ycombinator.com/item?id=26613795>
More by dang on HN's brief guidelines: <https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...>
And no, I don't always agree with dang. That ... would be obsequious, pandering, and utterly unimaginative. But I very frequently do, and even where I disagree, can understand why he's moderating as he is.
See, e.g., <https://news.ycombinator.com/item?id=37566315> <https://news.ycombinator.com/item?id=37258356> <https://news.ycombinator.com/item?id=29493683>.
And finally: I really don't expect that BugsJustFindMe will be persuaded. I feel rather like the narrator of Tim Minchin's exquisite beat poem "Storm", and we'd all as well be back nine minutes in time. But I'm hoping that other readers might find reason and fair claim here, and reach an understanding and appreciation. As much as my faith in persuasion through rational argument has taken a profound hit over the past decade or so, that hope remains live for me.
I could be misunderstanding what you’re saying, but regarding Fosstodon, they’re not at all limited to one community — they’re not decoupling from the fediverse?
As for administration policies - find a server whose moderation policy you enjoy. Unlike on Xitter, you have a choice of these policies and have immediate contact to the admins and moderators of your instance and other instances, and you can consult them.
If you still feel limited, then the solution to your problem is to have multiple accounts for the mutually-exclusive communities you enjoy. That's what people do on Reddit, Facebook, and Twitter with their alt accounts. Keep that in mind whenever you encounter someone online using an anime profile pic.
The major factor I look for when choosing a mastodon instance is the business model. The vast majority are free, which leads to either limits on registration, ads, or spammy free-for-all instances. I want to join a mastodon instance that I can pay for, as a control over these types of problems. If it can sustain itself through membership, then the experience and policies should be relatively stable.
For the above reasons, I use mastodon.green.
Personally I think a lot of services on the web would be better if they were invite-only where invites are free to provide by any member. It seems one of the few ways to effectively cut down on spam, abuse, botting and astroturfing. Mastodon servers are an even better version of this because even if you can't find someone to invite you you can still participate from another server (and if needed you can run your own).
On that note if anyone wants a Fosstodon invite I'd be happy to provide one if I can tell from your posting history, blog or other online presence that you are a real person. (At my own discretion and offer may be revoked at any point.)
There are many one-person instances, for example.
And Discord servers are just totally opaque and you can't search its contents on the web (even if it is public). You have to join the server and use Discord to do so. This is generally bad for information longevity. This is why people say Discord should be avoided for hosting discussions on Open-Source software.
There's a very large segment of the mastodon population that thinks servers ad big as .social are very bad for the ecosystem. A lot of instances defederate them out of principle.
The argument is that .social is so big that it's not possible to moderate effectively. It also presents a very large and inviting place to launch attacks from. We frequently see waves of spam, abuse, and just bot accounts originating from .social.
A lot of severs are already defederated from .social for these reasons
Would you also know if it's commonplace for instances to deny by default, and then use whitelisting?
The fringe-er instances are more likely to block by default. It's almost an issue of fediverse fundamentalism. A lot of people think large instances like .social are antithetical to what mastodon is trying to accomplish. We have huge, enormous centralized servers in the middle of a decentralized network. I think you can see how that sentiment would come about
All of these “twitter but with fraction of users” -things won’t really mount to much
I'd rather a service have been a success amongst its smaller userbase than be an anti-success with billions of users worldwide but is a massively unprofitable trashfire that is implicated in multiple state psyops and infowars and is an otherwise unmitigated shitshow with a clown running the circus.