I find the nostr[1] protocol to be a nice alternative here. The core idea is the separation of message content from the relays which transport them. So you sign a message and publish it to several relays, and then anyone can retrieve your message from any of the relays where it is published.
As opposed to Mastadon, your identity is tied to your pubkey rather than the instance that you created your account on. The difference to Scuttlebutt is the differentiation of relays vs clients, and the lack of chaining messages together in the "blockchain"-like structure.
The result is a fairly simple protocol which is fairly low latency and seamless for many applications compared to other decentralized solutions.
> Learn about Nostr: A simple, open protocol that enables a truly censorship-resistant and global social network.
> Relays are like the backend servers for Nostr. They allow Nostr clients to send them messages, and they may (or may not) store those messages and broadcast those messages to all other connected clients.
I still haven't wrapped my head around these two words from the nostr site. How can it be truly censorship-resistant when used are entirely dependant on what their relays decide to store and share?
I'm not particularly familiar with the project, but:
The claim is that it enables social network/s that are resistant to censorship. It doesn't say that it is a social network that's impossible to censor.
It's a protocol that everyone is free to implement. Valid implementations of the protocol should be compatible with each other. Meaning you don't have to go to soshl.com or use the Soshl app. Relays can choose whether or not to store/broadcast your messages, yes. But they can't stop you from sending them and they can't fake or modify your messages. They also can't revoke your identity. I assume a relay can't force other relays to delete things, but I'm not sure of the details.
A relay choosing to retransmit your message or not could be considered censorship, but it is not the same censorship as a court ordering Youtube to take a video down, which is not the same as Twitter silently burying your posts, which is not the same as the police arresting protestors, etc...
Users publish their messages to multiple relays, and they can be followed by others on multiple relays. The idea is that if a relay starts censoring, then users and their followers can choose different relays, so the system as a whole “routes around censorship”. At least in theory.
I interpreted in a similar yet different way. It is not just that users can choose a non-censoring relay but that users are connected to many different relays and so it does not matter if a few relays censor stuff
This is an interesting detail to me and may just spend on how the community grows
In Mastodon each server makes their own decision on who to censor. Functionally it's grown into a bit of a hard mentality with entire servers being censored if they don't go along with the larger group
Messages are signed locally with entirely local private key, so relays can censor or ban all they like following local laws, without harm to your identity or social graph.
Hopefully someone, like a Raspberry Pi cluster on a derelict Soviet military space station on an international orbit, will take your message and relay it, :shrug:, and those messages can still be authenticated by the key.
That’s how I understand that part; Nostr uses DNS for blue badges and relay connections so “truly” indeed sound a bit of crypto talk though.
Do you know if nostr supports peer to peer messages, or did everything have to go through a relay?
If it's all through a relay it probably isn't very censorship resistant at all. My messages may be signed locally but if the network trends towards mob-based bans and censorship like Mastodon as soon as I get on the wrong side of the network I'll be the only one seeing my signed posts
I believe it's all through relays, there's no peer discovery or even inter-relay meshing. So that could very well could happen. Currently it's just each clients multi-posting to dozen relays and receiving dozen duplicates.
What I believe that "censorship resistance" actually means in un-cryptocurrencified talk is, it lets relay operators filter out locally illegal but not globally unethical contents(e.g. political speeches, pornography, certain URLs and strings) without banning users and/or fragmenting the network. And that is an improvement over Mastodon/ActivityPub architecture.
But boy those crypto guys knows how to hype it up...
>If it's all through a relay it probably isn't very censorship resistant at all. My messages may be signed locally but if the network trends towards mob-based bans and censorship like Mastodon as soon as I get on the wrong side of the network I'll be the only one seeing my signed posts
Well on their page they said"
To publish something, you write a post, sign it with your key and send it to multiple relays (servers hosted by someone else, or yourself)."
If you really want to prevent that then it is to also selfhost.
Good question. Not sure about Nostra, but that is how WebRTC is doing it. If there is only 2 people then it is peer to peer, more parties require some sort 'relay' to facilitate the chats.
However the main issue with peer to peer is not everyone is going to be online all the time, relay serves as a temporary storage place until one of the user/client goes online and get the message.
That's because bitcoiners created it and Jack Dorsey boosted it and abandoned Twitter for it. He's been extremely active there, https://primal.net/jack
And the idea of using nano transactions for upvotes has long been an attractive approach in this community for mitigating bot swarms from influencing conversations. In fact, the proof of work system that bitcoin was based on was originally proposed in the 90s as a solution to fight email spam. A micro cost of computation power (or a nano transaction) is neglegible to most legitimate users, but adds up quickly for spammers who depend on sending out thousands of emails for a single hit.
Then clients and relays SHOULD act as if the referenced event is deleted. This is different than MUST, probably because there’s no way to guarantee it happens. It’s also possible the delete event doesn’t make it to everywhere that received the initial event.
Social media sites are being siphoned up by various intelligence services, data brokers, quant trading firms. Everything you delete always has a window where it can be permanently recorded by someone. Even on HN. Anything you delete/edit isn't necessarily gone forever, even if it is no longer on the server you posted to.
It's a key reason why I don't attempt to delete social media posts. I've said a lot of stupid shit over the years, and whether I like it or not, that stupid shit is permanent; any attempt to take it down would only serve to make it more interesting and more worthy of preservation. Better to let that stupid shit linger unnoticeably - or even better, embrace that I'm (hopefully) a better person than I was $N years ago.
> It's a key reason why I don't attempt to delete social media posts. I've said a lot of stupid shit over the years...any attempt to take it down would only serve to make it more interesting and more worthy of preservation.
Only if someone cares enough about your social media to be watching for deletions, which I'd imagine is pretty rare. If no one's watching now, now is the right time to delete something.
> See also: the Streisand effect
I think it's important not to overstate that effect, as there are lots of scenarios where deletion will not draw more attention to something. For instance: you're a rando deleting something that's not interesting, not currently being observed, or something no one will notice when it's gone.
The trope namer got hit by it because 1) she's a celebrity, and 2) removal would result in a noticeable gap.
It would, if someone were looking in detail, but that isn't what the Streisand Effect is. That effect is the demand to stop/remove something that draws attention to it, not the actual removal (if it is indeed removed at all).
If the photo of her property had been quietly removed by the photographer I and most others would likely never know anything about it, much like you and most of the rest of the world wouldn't know I deleted a post on my blog that was so full of embarrassing typos that I couldn't be bothered to correct it. I know about it because of the heavy-handed demand that it be removed from that collection.
Right, but how do I know that nobody's watching? Indeed, the idea that nobody's watching seems far-fetched; even if that somebody is some Internet archival robot or NSA monitoring tool or advertising algorithm, it seems more likely than not that all but the most recent posts would draw more attention if deleted than if left alone, whether I'm a nobody or a celebrity or somewhere in between.
> Right, but how do I know that nobody's watching?
A better way to think about it is: do you have good reason to think someone is actively watching you or will go through the effort to collate your social media posts against some archive (that will be incomplete unless you're an active target)? If you don't, delete away.
If you're still concerned, delete some old stuff randomly to distract from the stuff you really want to delete.
> even if that somebody is some Internet archival robot
Those can't even scrape everything people actually want to keep, let alone comprehensively scrape every social media post from everyone. There are huge gaps.
I actually was peripherally involved in some volunteer efforts to archive a bunch of stuff around the Hong Kong protests and the Afghanistan withdrawal. IIRC, they didn't even try to scrape Facebook posts (because FB was doing so much to thwart scraping, and there may have even been problems with Twitter). There are well-developed tools for YouTube, through.
Exactly! It's amazing how few people understand this. Signal is the perfect example, there's a lot of technical magic wrapped up it, but my Dad can use it.
It's funny how common jailing-breaking iPhones used to be, I did mine immediately when it would come out and everyone else I knew did too. Now I don't know a single person that does.
You know what, that’s a great insight. I got the iPod touch precisely because it could be unlocked to be a pocket Unix with shell. 15 years later, it still is and isn’t. Like, lol not at all ha ha. It was a giant, near irrecoverable mistake to computing freedom.
> Append-only databases are acceptable for financial ledgers, contracts, and things we want to be persistent and immutable.
Neither financial ledgers nor contracts are things we should want to be persistent and immutable - as the crypto and smart contract folks have repreatedly demonstrated
You very much do want immutable and persistent records of financial transactions. That's the whole point of having a ledger in the first place. Error corrections should appear as actual corrections and be dated appropriately.
It depends on what the ledger does. If the ledger is tracking transactions that occur somewhere else and for some reason a transaction is entered incorrectly then immutability will prevent the ledger from being corrected to reflect reality.
I have seen way too many systems that model something that happens somewhere else and provide no way for correcting discrepancies between the system and reality, built on the assumption of either infallible humans or infallible integrations, when the truth is that all humans and all integrations will eventually fail to input data correctly and need the ability to correct anything (no field should ever become fully read-only in such a system, and no status should become fully unreachable from any other status).
> If the ledger is tracking transactions that occur somewhere else and for some reason a transaction is entered incorrectly then immutability will prevent the ledger from being corrected to reflect reality.
You just enter a new operation correcting the issue. If you can't do in in the external ledger, you do it on your own side and then cancel your correction using a new operation when the external ledger is finally corrected.
The point is auditability. You want what's inside the system to reflect what's actually happening.
Talking about power usage per transaction, instead of talking all the costs to secure a network? Check!
"Crypto is only for buying drugs", when anyone serious already said numerous times that the key value proposition of blockchain is not in privacy but its permissionless aspect? Check!
It looks like you just won Stupid Crypto Basher Bingo for today...
While I mostly agree with you (especially about the energy savings from using alternatives to proof-of-work), we do have to admit that at the moment, crypto-currencies aren't really used to buy much of anything apart from other crypto-currencies. There's not even all that much drug business going on.
Right, but this is not the fault of the people working on crypto. The (good) people working in crypto are working based on principles and values. The ones people keep talking about are the ones working to find a way to make a quick buck without any actual value built for society.
I spent a good part of the last cycle (from 2018 until mid last year) working on an open source self-hosted payment gateway for crypto (https://hub20.io). Even with my ridiculous ideas for how to run marketing and promoting my work and doing basically everything possible to stay away from the radar of the scammers and hype-riders, they were the ones getting to my matrix room and coming up with proposals for "partnerships". The quickest way to get to disappear was by saying "If you think that my project is such a good idea and so valuable, how about you sponsor my work on Github? I'll take even a contribution of the smallest $4/month tier as a signal of interest. If you do it, then I'll listen."
Chalk this one up to human nature and our lizard brains, I guess.
Take any system where we depend on some central authority (an institution, a state, a quasi-monopolistic company) that can deny you access to their services, and try to come up with an alternative system that could work for them.
If you don't need any of that and you can leave in a society that warrants you and your neighbors fair treatment, great. But consider yourself priviledged. The majority of people are not just as lucky as you and have to deal with corrupt or dysfunctional institutions, authoritarian states that free exchange of work and goods, companies and banks that take arbitrary decisions and leave merchants without access to their funds, etc.
Before you come with the tired-and-stupid argument of "blockchain will never replace any of these things": yes, we know. This is not meant to be some type of revolution. No one really wants to live in a decentralized utopia, at least none of the sane ones. We just want to have an alternative for the time and places where shit is hitting the fan.
So don't use it?
But the existing banking sytstem is no different in energy consumption –more still if you figure in supporting government agencies that crypto obviates.
Completely unaware of this social network. What problem, exactly, is a blockchain based social network trying to solve? Proof of origin? A conversational paper trail?
The latter would be interesting for areas where conversations should be on the record, however in cases that people want their conversations off the record they’d just use a side channel. The UK govt communicates using WhatsApp for Christ’s sake.
So it’d need to be something where both parties fundamentally benefit from the ability to verify that a post, or a conversation, took place. Where would that even apply?
> What problem, exactly, is a blockchain based social network trying to solve? Proof of origin?
(SSB is blockchain in the strict sense, but there's no proof of work or anything like that. Each feed is its own append-only chain with its own private key.)
For SSB, the aim was to be able to gossip feeds via untrusted intermediaries, with patchy network connections all round, and be sure that the intermediaries haven't added, changed or removed any posts in the feed.
The protocol's designer lives on a boat in New Zealand, and other collaborators live in remote areas in different time zones.
One adjustment to the protocol that seems to me like a quick win (but presumably has some technical hitch I can't see, since I don't recall anyone suggesting this) would be to not include the post's body in the “block” (in the message itself that gets hashed and signed by the next message), but rather as a “blob” (essentially an attachment) which others don't need to download in order to verify the feed. That way old messages could be effectively forgotten if all peers co-operate (and no-one took a screenshot, etc).
> For SSB, the aim was to be able to gossip feeds via untrusted intermediaries, with patchy network connections all round, and be sure that the intermediaries haven't added, changed or removed any posts in the feed.
why would that need a blockchain (even without proof of work), as opposed to simple signed, timestamped, indexed posts and having the reading client app check the signature and post indices?
> > For SSB, the aim was to be able to gossip feeds via untrusted intermediaries, with patchy network connections all round, and be sure that the intermediaries haven't added, changed or removed any posts in the feed.
> why would that need a blockchain (even without proof of work), as opposed to simple signed, timestamped, indexed posts and having the reading client app check the signature and post indices?
It depends on how the message is signed. If the message format's this:
Just as a reminder: The original stated task was to protect against "untrusted intermediaries" and "patchy network connections", and in particular to make "sure that the *intermediaries* haven't added, changed or removed any posts in the feed." … which can perfectly well be done with simple signed, timestamped, indexed posts and having the reading client app check the signature and post indices.
In contrast, this here, which is the only thing the blockchain struckture really adds:
> The signer (malicious or not) can resign an old message & effectively overwrite a previous message
… is a rather radical shift of the goalpost, as the only capability the blockchain really adds is to protect against modification not by untrusted intermediaries (which was the purpose and which would already be covered without a blockchain) but by the legitimate owner/editor of the feed.
In other words: The only thing the blockchain adds is that it makes it impossible for YOU as the user to edit YOUR OWN posts. I'm not sure that's something most people in the social network context (as opposed to, say, a financial transaction ledger) would see as a feature and not an anti-feature.
> why would that need a blockchain (even without proof of work), as opposed to simple signed, timestamped, indexed posts and having the reading client app check the signature and post indices?
Yeah, that's exactly what SSB does, plus each post includes a hash of the previous post.
I think that's so that even the holder of the private key can't retroactively replace an old post with an altered version of that post (but with the same index number and similar timestamp).
> One adjustment to the protocol that seems to me like a quick win (but presumably has some technical hitch I can't see, since I don't recall anyone suggesting this) would be to not include the post's body in the “block” (in the message itself that gets hashed and signed by the next message), but rather as a “blob” (essentially an attachment) which others don't need to download in order to verify the feed.
There's nothing to prevent you from taking this route, you just sign a blob hash instead of an entire message object.
I work on an experimental SSB-like-protocol in my spare time that does exactly what you've suggested: https://github.com/evbogue/bogbook
I don't know if this makes the network forget more, but the aim is to reduce the time it takes to sync and get started.
I'm working on something in this space, though god knows if I'll ever finish it. And that's one of two approaches I considered to that problem. Another is to have a signed deletion marker saying something to the effect of "there was a block here with hash d9841a, but now there isn't anymore, and I'm its replacement".
To my understanding, it allows messages on the network to have a total order, even when fully decentralised. Without a hash chain, clients could not tell what order events happened in, and might display them ordered differently, which would change a conversation's meaning.
The traditional alternative is logical clocks, but a malicious participant could forge those, so they're not suitable for a fully distributed system run by random users.
AFAIK it was intended to support the use case of being occasionally connected to the internet (one author spent a lot of time on a small sailboat in the south Pacific), while maintaining an order of messages.
That sounds like it could be useful in legal and policing contexts. One complaint I’ve often heard about body cameras in policing is the possibility of footage that was created and did indeed exist being “lost”, with the implication being that legal entities intentionally destroyed evidence of either the footage existing, or its contents.
To the extent that this centralized authority issue applies to textual data or metadata, decentralized records of messaging times and contents would potentially be an invaluable tool of transparency and promoting systemic confidence in a visible shared audit trail.
> What problem, exactly, is a blockchain based social network trying to solve?
The same as every other blockchain application: the problem that no-one has found a practical use for Blockchain despite 14 years of effort, but that blockchainers continually need the next wave of hype so that their coins don't lose value.
> What problem, exactly, is a blockchain based social network trying to solve? Proof of origin? A conversational paper trail?
I would say the main interesting problem to solve is how to have a social network with a shared state without requiring a all-powerful central party like the existing solutions. Blockchain solutions allow a consensus about what is in the network without this.
If you don't care about the social network state diverging endlessly that's not a problem, but part of the "social" part of social networking is that users expect to see broadly the same messages or at least that all messages are drawing from a common well in the sense that I can choose not to follow a particular person or conversation, but if I do follow them/it then I see the same set of messages as others who are following. This is how we have a shared conversation.
Edit: Not saying this is the only or best way to do this, but it is one way and I believe this to be the problem blockchain social networks are trying to solve. eg these guys are building a blockchain social network and it's explicitly one of their goals https://www.projectliberty.io/
While not anything blockchain-based, I had a similar thought of just doing social media through a single text file that you can add to/remove from whenever, and re-post the whole thing[0]. Literally HTTP POST it.
Edit, amend, add, delete, schedule, etc., all from a single text file. I haven't worked on it in a few weeks but I intend to get back to it soon.
I'm pretty sure, OP is aware of that. Someone found a passion project for themselves. Nothing to be snarky about, right?
People do stuff for the sake of doing stuff all the time.
Haven't been posting as much as I want to yet, I'm working on the editor from https://markwhen.com to hopefully port it over to https://post.mw as well
Worth considering, but the conclusion that permanency is "definitely not what you should look for in your social network" goes way too far. People use different social networks for different purposes.
There are certain privacy features like self-deleting messages that you can only get in a closed ecosystem. Does this mean you shouldn't use any kind of web-based social media, since everyone you communicate with can easily and automatically archive everything using a browser extension or userscript?
Unfortunately lawyers at least in Russia would have an opinion here.
Let's suppose that you posted something publicly to an append-only social network, and it was completely OK at that time. Then, laws changed (and you could not predict this), and this post is no longer legal according to the new rules. What the lawyers say is that the very fact that the message is still online is now a crime. Yes, there is a universal notion that laws do not apply backwards in time (retroactively), but here they indeed don't: the message is still online AFTER the law got into existence, and so this is a lasting crime that started when the law was enacted.
Normally you would be required to delete the post between the moments when the new law is announced and enacted, but you can't, and it is still yours for others to find and read, so you go to jail.
P.S. It would be interesting to know the opinion of lawyers from other countries.
Mostly agree. You can have strong cryptography even with mutability/overwritability. You can simply issue soft deletes by issuing a tombstone write.
However, you always risk someone else (like a scraper) storing messages including your signature forever. Deleting is virtually impossible for practical communication purposes. Best you can get is “I promise you I deleted my copy and thus wont distribute it further”.
I'd generally consider that different because whatever people scrape could be randomly reassembled but its not forming a canonical proof. A social platform that digitally signed content to have irrefutable proof of what you scraped would be closer and that would probably be adding the anti pattern they criticize for little reason.
It sounds right, but would you mind spelling out the full conclusion? The level of indirection is a bit high for my lazy brain.
To me public communication has the trade off between trusting the communications platform (like most social platforms today) vs trusting the end user (signed messages). If you have trusted end users, scraping a third party repo of “forever history” becomes more plausible, as they can both prove that you(ish) signed it and are unlikely to respect your request to delete their copy. Is this different from what you’re saying?
If you post something public it can be attested to by whoever that it once looked like X which is almost as irrefutable as it being self proving like a block chain. Scuttlebutt is trying to help you obscure your identity in posting and one could be sending encrypted content to particular other parties (maybe in its protocols but certainly just using it for the medium) where those parties could reveal irrefutably that content and that you made that content as that identity in that historical context.
Take instead the example of sending content encrypted with shared session keys over a live mesh network, where you send session keys to your specific friends to decrypt instead of revealing an identity to them.
Any "friend" could prove that they had some content at a specific time, but they could never prove that they didn't just make it themself since you gave them the symmetric key that let's either of you make shared content. This is a simple way to fail to achieve irrefutable though other properties as well..
If for example you are accused of a crime with no statute of limitations and a friend has since died, there should be a distinction between whether their hard drives prove something or their hard drives are hearsay without something else, like their testimony that they didn't tamper with contents in some long forgotten practical joke.
This article like many others claims a mass exodus from Twitter, but Musk claims usage is at an all time high. I wonder which is true? I’d be inclined to think Musk is lying. He has clear motive to do so. But on the other hand, he has access to the actual numbers unlike any of these writers who are relying on anecdotal evidence. The writers also say hate speech is more common, but my anecdotal evidence says the opposite is true. Genuinely torn on this one.
Didnt they allow all the banned users to come back + removed the systems that automatically ban bots?
If you let the bots come back, then you will have much more "users".
They unbanned some controversial accounts (Trump being the most famous by far, although he hasn't posted anything since being unbanned), but it's pretty hyperbolic to say they allowed "all the banned users to come back"
I rejoined Twitter mostly to hang with writer friends, but I’ve noticed that their engagement has gone down a lot since Musk took over and with it mine as well. It used to be that I would pop into Twitter several times a day, now I’ll go days.
I can’t speak to hate speech (other than seeing reactions to it showing up a lot in trending topics) as I’m liberal with my use of block to keep Twitter as non-toxic as possible.
The third option is that the mass exodus is limited, and Twitter usage dipped just a bit. Mr Musk has obvious reasons to lie, but his users or people who don't like him can also lie.
If the mass exodus has caused any major user count decrease, people should be talking about "Why would you still using Twitter?" than "Hey, I've closed my Twitter account, cool uh?"
I deleted 12 years of Facebook posts after downloading a backup one by one. This was before they had multiple deletes. It took me a few weeks so I did it when I was bored. What's funny is that you can't actually delete everything, because something from the past will show up again. Just one or two things. I assume when databases located around the world sync up there's a few things that didn't get deleted.
I think they are supposed to delete them by law, but they could ignore that. Assuming they do delete them, it seems to me they miss a few that show up as the DBs around the world sync up
Mastodon and SSB have terrible user interfaces. Centralization still makes the user experience way better and smoother, a necessary condition for a network effect and mass user adoption. I don't see these products going mainstream.
Also, I wouldn't call the Snapchat UI terrible. I would call it difficult for first time users. This creates a "buy in" effect with users... Lots of powerful tools have this concept, similar to the Bloomberg terminal.
- Your identity isn’t a username and doesn’t depend on a domain name or service provider. Instead, your username is a meaningless cryptographic hash that consists of random-looking numbers and letters. Your friends only need to know your hash to follow you, and you can keep it as public or private as you want.
Yeah, that sounds like a really positive user experience. Hard pass.
No one has or owns a "true name", collisions on names are expected and natural with a set of billions.
Crypto keys aren't supposed to collide, but on a set of billions this could happen. -- If it does have a contingency for replacing the PKI part of the identity. This also covers loss of key material, etc...
Identity is ephemeral and in flux.
An indexed web of trust, who knows who, etc might work.
> The technology that powers SSB is entirely different from Twitter and Mastodon. You don’t need to register an account anywhere to get started. Your identity isn’t a username and doesn’t depend on a domain name or service provider. Instead, your username is a meaningless cryptographic hash that consists of random-looking numbers and letters. Your friends only need to know your hash to follow you, and you can keep it as public or private as you want.
I'm not familiar with SSB. And, to me, this article's headline is five words too long (harumph)
Question though: Does this imply that I can follow someone without them knowing about it if their address hash gets leaked? If so, in what way could it meaningfully be described as being private? I assume there's some way to control who you broadcast to, or approve someone who follows you, and this description is just not mentioning it.
~Everyone can tell how many of these you sent, when, and how large were they, but not to whom (incl. how many recipients they had unless there were too many and you had to send multiple copies[1]) or what was in them.
[1] The copies will still be unlinkable, but if someone sends 5 messages of ~same size within a second, it's a pretty good guess that they're sending 5 copies of the same message.
I treat all of the web as being an append-only store and so should you. With storage being as cheap as it is (and it's still getting cheaper), your published information is going to be stored somewhere. Make decisions about posting something as if unpublishing it were impossible.
Methinks any social network that prioritizes its tech over user experience will not take off. It doesn't mean that new tech is automatic grounds for failure, but if that's all you have to show for it right now you barely have anything.
In SSB it's a public key. Most users announce their own display names (which needn't be unique) and some clients let you set a pet name for other users (like a phone's contacts list).
After hard-avoiding social networks, I spent some time thinking about this problem when I wanted to share pictures of my kids. One of my requirements was privacy, another was _not inventing a protocol_. I landed on RSS and built Haven[1], but I made the rookie mistake of putting my principles above usability. I want it to be hard-decentralized so everyone runs their own server--but nobody wants to run their own server.
I still have vision for what Haven could be, and supporting self-hosting remains important but I think I'll end up with something like the Matrix/Element model where there is a (free?) centralized server you can use, or you can host your own with full interoperability.
Avoid building new protocols. However if it really can't even be shoehorned into an existing protocol, or if there are extremely clear problems (to the point that the existing closest option is even unsuitable as a compatibility / bootstrap interface); then go ahead and make a protocol that really covers the use case, if you can also make an open source (as in free beer) library anyone can use to talk with it.
As opposed to Mastadon, your identity is tied to your pubkey rather than the instance that you created your account on. The difference to Scuttlebutt is the differentiation of relays vs clients, and the lack of chaining messages together in the "blockchain"-like structure.
The result is a fairly simple protocol which is fairly low latency and seamless for many applications compared to other decentralized solutions.
[1]. https://nostr.com/