Genuine inquiry . Maybe I am not experienced enough with the various federated platforms but I am an avid user of matrix, and have dabbled in lemmy. From what I have seen is federation is on the path to decentralization but not fully there. It creates fiefdom, little kingdoms . Great yes you may find one that suites you better, but users now can end up isolated to their island, switch island sure but now you are isolated for the previous island and maybe others. Its stupid. On matrix you need to know the other island(server) to even find its rooms(communities). Some rooms block users from one server while others block users of other servers. You either have to run multiple accounts or accept the limits. Add in you are at the mercy of your home server, you can lose your account have it immitated, and more. The performance is horrible not sure why, but content is slow to update and spread. Matrix has the problem because of its design most people are on the matrix.org server and so the point of federation is largely lost. They are moving to p2p where it seems the solutions for federation now dont apply.
Anyway why is federation not stupid? Are these problems only with Matrix? Cause I look at lemmy and it seems far worse.
Federation is stupid
So don’t use it.
convince me otherwise.
Perhaps it’s one of those self selecting things in life - if people bitch and complain and are those sorts of people that don’t want to research into a topic enough or are generally negative and will complain about everything, then they shouldn’t use Lemmy, and this would lead to Lemmy with the best people.
Win-win.
Well, win for us. They can stick to Reddit and threads.
Instance blocking and user blocking are features, not problems. Most users want to block trolls for example. If you dont like that, you can make your own instance without, or just go back to commercial social media. The rest of your rant seems exclusively to Matrix, not Lemmy. I think you should spend more time here (or on another instance) before you really judge it.
You are correct I need to spend more time here.
Realistically most people are not going to run an instance . Blocking does become a problem when you dont run an instance because most do some blocking or are blocked. Even if you run an instance at some point for any reason it could also be blocked.
Yes my rant is not solely targeted at lemmy put mainly Matrix because that is what I use. I should probably look more at mastodon and some others too. Federation has a lot of spins on the technology.
Its nice that you still get involved with users here, the matrix devs are also quite easy talk to. Have you considered how lemmy might intergate with other federations? I did mention in another post someone was developing forums on matrix. Hummingbard I think its called. I have wondered if chat and forum could be combined in an interesting way, such that they could both possibly flow together but be seperated by alternate uis or view modes. On matrix often discussions can take a forum type format and style with conversations having days or many hours between replies. It would be great if those could end up in more permanent threads.
It may be worth reaching out to matrix and element and seeing if some crossover can occur or even intergration. Their $30million and german health sector selection is a lot of potential dev work to be done.
Thanks nutomic!
I think instance blocking is pretty transparent, you can see the list of allowed and blocked instances under /instances. If admins do something you dont like, thats when you choose another instance.
Its funny that you mention how i “still get involved with users here”, because Lemmy is still really small when you compare it to Matrix or Mastodon (16k total users, less than 1k active in the last month).
Matrix uses a completely different protocol, so I dont think federation with them is realistic. And we definitely have enough work to do on Lemmy itself. But it might be possible to write a chat app using activitypub, or even as a Lemmy frontend as described in this post.
I am not arguing its not transparent just its not ideal. I think matrix users are way over stated.
Thanks
Federation is critical to the development of alternative social networks. The reason people use specific social networks is because other people are on them. If there’s an open source social network app that isn’t federated, then everyone using sites powered by that app is isolated. With federation, all instances of the social network can interact with each other. Different instances can pop in and out of existence and the ones that stick around can become popular.
It’s better for new social networks to be federated then isolated.
Replace “federation” with “email”.
Still stupid?
I want to be clear, that I disagree with his “federation is stupid” point, but email has problems right now.
Theoretically it’s federated, theoretically you can spin up your own mail server and self host.
But even if you do that absolutely perfectly (SPF, DKIM, DMARC etc), you can falsely end up on spam list, that effectively block delivery of your email to large segments of the network for days if not weeks.
Whilst theoretically federated, email falls under the broad dominion of google, microsoft and a couple of other large players.
As a cybersecurity pro who primarily deals with email security right now (in the sense that there is basically none)
STOP
CLICKING
LINKS
I mean yes, but that’s not a federation problem.
To completely strawman AND slippery slope what you’re saying:
As a car safety pro, who primarily deals with car crashes:
STOP
TOWING
TRAILERS
Agreed, dangerous, I don’t want numpties doing it.
But it’s a large part of why I have a car.
Federation is the most natural form of human society. We’ve developed to exist in small communities of a couple dozen people. Some groups talk to each other, some don’t, and every one has its own identity. Sometimes a member leaves and gets assimilated into another group.
Existing in large communities with thousands and millions of members - other species do that, like ants.
Ok weird analogy, but I think that’s the gist of it really.
Uniform places like Facebook, with one queen/master, perfectly organised, never really seeing outside, being just a cog - that’s a life of an ant.
Small, agile communities, sometimes a bit messy and complex, especially when it comes to outside interactions - we can handle that, because of our huge human brains.
It’s time for moving away from being ants back to being humans.
It’s definitely not stupid. P2P does have its own set problems and inefficiencies that don’t exist in federated servers.
Also, the separation in “fiefdoms” is not necesarily a trait implicit in the federation. Email is federated and I don’t think people using Gmail feel like it’s an isolated island separated from those using Hotmail.
Having to synchronize state across peers in a P2P network can be troublesome and has limitations, specially in groups with very few peers that might not be simultaneously online at the same time and/or have limited bandwidth (if you think content in federated Matrix is slow to spread, then don’t be surprised if it’s even worse in a situation like that were it be P2P).
I think you can find arguments for both methods and it just depends on what are your requirements. I think a hybrid model of federated servers and peer nodes where you can choose whether you want your community on a dedicated server or in your own peer node (that’s what I hope Matrix is doing) would be more flexible and preferable than just going full p2p or full federated. But neither of them are “stupid”.
P2P does have its own set problems and inefficiencies that don’t exist in federated servers. […] I think a hybrid model of federated servers and peer nodes […]
Indeed, I too think that eventually we’ll move to a model of hybrid decentralization, where both federation between servers/apps and P2P devices are intermixed. Both types of decentralization have their pros and cons and this allows a “best of both worlds” scenario.
I think a hybrid model of federated servers and peer nodes
Yes email is an interesting point, but aren’t many mail servers blocked for spam? or many automatically filtered into trash for spam? Mind you so many run their own domains for companies it cant be possible that only a few domains are whitelisted? I am not sure how gmail and others go about blocking for spam but it would have to have a reporting theshold type situation.
I think as federation expands feifdom should be less of an issue but then look at the split in mastodon with gab.
The other thing to think about email is it is user to user messaging, private with laws protecting it. I am not concerned with being blocked or censored in the modern messaging platforms for this because they use e2ee and cannot discern any of the communications. You are still at the mercy of the network you are on and hence p2p helps with that, depending on hardware how much though. Wouldn’t it be great if we could use the cell hardware in phones to mesh? Thats nuclear world ended or revolution type tech.
I think matrix p2p will have to take a multi type node approach, with them beginning the existing home servers will just act as peers but imagine more will have to be done. Yes propogation will be slow but the tech is quite new so will have to see where it can go. I imagine nodes can adapt or be set to provide more based on resources available much like torrents can. Like a hydrid approach but more dynamic.
Yes email is an interesting point, but aren’t many mail servers blocked for spam?
Aren’t there blocklists in many p2p networks? I know in bittorrent there are. I’m ok with blacklisting (which I expect even gmail does, like you are hinting), as long as it’s non-intrusive enough to not be detrimental.
My point is that it’s not federation what’s the cause of the “fiefdom” syndrome, the behavior of the p2p-swarms/federated-group is. In theory you could have the same problem in p2p network if a majority of peer nodes applied whitelisting. Each whitelist (and whitelist sharing could be a thing, much like blacklists in bittorrent are, so each community could all share the same source for its whitelist) would be its own “fiefdom”. This would kind of defeat the purpose of the openness of p2p in the same way as whitelisting in federation kind of defeats the purpose of an open federation. What causes the “fiefdom” isn’t federation itself, but the protective philosophy.
On the other hand, I can also understand that protectiveness is needed when we deal with hosting content that’s not private anymore but it’s open to the public. When hosting publicly you become responsible of the content you are publishing, even when someone else wrote it. If, for example, a country wants to ban a community that shares illegal content, then they might take action against whoever is hosting that content, whether the content is hosted in a federated dedicated server or in a personal p2p node, that’s no different. They can prosecute you either way… same way as many countries prosecute p2p bittorrent sharing for copyrighted works. So ultimately you do need some form of moderation so that you don’t end up hosting content you don’t want to be involved with as “partner in crime”.
I imagine nodes can adapt or be set to provide more based on resources available much like torrents can. Like a hydrid approach but more dynamic.
Yes, that’s what I was saying. I’d prefer a hybrid model where you actually have both dedicated servers and user nodes. Torrents have problems when there’s not enough peers, there’s a lot of dead torrents out there showing this.
Aren’t there blocklists in many p2p networks? I know in bittorrent there are.
Block lists with torrents were used to try prevent legal implication and a few other reasons all which are largely useless because peers on blocklists can just move to nonblocked ips, especially peers with resources looking to gather evidence.
My point is that it’s not federation what’s the cause of the “fiefdom” syndrome
I disagree but it is a point worth exploring. Thinking what would be the cause likely is that the federated instances exert control over the content rather than just storing and relaying it to the users to deal with.
In theory you could have the same problem in p2p network if a majority of peer nodes applied whitelisting.
Yes you could but it still depends how the content is distributed, peers on the white list could act as relays to those not on the white list. This would risk them being removed but when dealing with so many peers I just dont see lists like that working to segment the network too much. They may as well fork. The desire to remain connected to the larger network will be too much for most users. You could also have underlying blockchains or hash tables that make forking the network not very promising.
When hosting publicly you become responsible of the content
Section 230 and the dmca should still apply, enforcing becomes much harder with p2p but so does defending I think the legal system does make having free speech an interesting problem inspite that the constitution says it should be possible. I think having speech government and companies and many others don’t want is part of the plan for these networks. There is a great many authorities not happy at the freedoms the internet has allowed.
Torrents have problems when there’s not enough peers, there’s a lot of dead torrents out there showing this.
Yes but I think keeping the text part of these platforms alive should be much easier. Torrent only needs 1 peer(seed) with a full copy, a social network can easily have 50000 users which is a lot of resistance and redundancy.
Block lists with torrents were used to try prevent legal implication and a few other reasons all which are largely useless because peers on blocklists can just move to nonblocked ips, especially peers with resources looking to gather evidence.
Same can be applied to federated blocklists. You can just move a server to a nonblocked IP. The only way to truely block content is to use whitelists instead of blocklists.
Yes you could but it still depends how the content is distributed, peers on the white list could act as relays to those not on the white list.
Same could be applied to federated servers then. Servers on the whitelist could act as relays to those not on the whitelist.
Section 230 and the dmca should still apply, enforcing becomes much harder with p2p but so does defending I think the legal system does make having free speech an interesting problem inspite that the constitution says it should be possible.
This all depends on the country. I know first hand of cases where letters have been sent to particulars in P2P networks and have been called to court.
a social network can easily have 50000 users which is a lot of resistance and redundancy
This is why I’m saying it depends on the case. Those big communities maybe don’t need a dedicated server, but that might not always be the case. I might want to make a private community for my family or a small circle of friends. Much like how I might want to have a Matrix room with a few people that might only come online one at a time.
Good points you make, whitelisting is powerful that way making closed communities. I wonder if lemmy will ever have encrypted communities?
Funny I just watched you make 2 edits while I was reading which impressed me with lemmy’s update speed. Its making matrix look slow.
Yes use case is important I think adding features and choice and letting users work it out is a good path. Something good would be serverless communities, matrix needs that for some rooms especially with p2p.
The slow-ness is definitely a Matrix problem only, other federated systems are in fact often faster then centralized systems as they can distribute the network load more evenly. Try XMPP and see how fast it is :)
I will try xmpp. Matrix is horrible, pics and vids take forever, chat can seem to lag for 10minutes at times! You would think it could be fast, like just distribute the content but its not and I think the servers are overloaded hence p2p is needed
Which instances did you try? In another comment you mentioned Matrix.org as home server, which might be part of that problem.
several and performance varies room to room. Can you recommend a fast server?
If you already tried several and had varying results in different rooms, can’t really recommend any, although I have made a wiki page with a super small list: https://wiki.tilde.fun/guide/messenger/matrix
It’s good if they’re close to you geographically. There’s a ping bot which is mentioned in Matrix blog’s TWIM segments, maybe one of those?
https://matrix.org/blog/category/this-week-in-matrix#dept-of-ping-
I think a lot of these criticisms are pretty fair in terms of user experience, but I’m more optimistic about federated services.
On matrix you need to know the other server to even find its communities
Can you think of a good alternative that doesn’t fall victim to the issues of centralization? e.g. All communities are registered with a community index – but doesn’t that community index have a responsibility to hide hateful communities?
(I think distributed, curated community lists could work here – which could be a feature where you can see the communities / rooms / magazines another user is publically subscribed to.)
little kingdoms
I think this is a real concern, more than people are giving you credit. I lost my vlemmy.net account this week – it’s unclear if the admins could have prevented this (I bet I would do a worse job!), but it’s a situation of bad stewardship / “lordship” :) There is also the issue of political defederation.
I think the answer here is self-hosting, which I think solves all the issues you mention (let me know if I’m wrong). Self-hosting is not free in money, time, or knowledge – the PC side of me wants to say that this makes it not a viable solution for people in need. But another side of me thinks that - at some point - there is a cost to using a service that needs to be paid: either pay in self-hosting, or pay by serfdom.
performance is horrible
Can you be specific? I know Matrix folks are working on sliding sync to manage some big performance issues. https://github.com/matrix-org/matrix-spec-proposals/pull/3575
I think performance will never be 1:1 to a VC-funded, centralized tech company, but I don’t think federation inevitably means “horrible” performance (but maybe someone with more technical knowledge can weigh in).
I appreciate an actual insightful answer. I feel like most people speaking for or against federated systems don’t understand what it is they’re even talking about… So an absolute breath of fresh air.
I’d say the worst thing about the fediverse is people talking about it rabidly but stupidly. The number of posts demanding defederation between instances because of the XSS vulnerability was mind blowing. People with no clue what was going on but only one lever to pull just wanting to pull it. Crazy.
What happened to vlemmy.net? I’m OOTL, all I know is I can’t login or reach the server.
No one can log in or access it, and there is no communication about it from vlemmy admins :( https://lemmy.ca/post/1307147 seems to be a good summary.
Sorry, not related to the topic but… nice application of Cunningham’s law btw. https://xkcd.com/386/
lol I didn’t know its called that but I have seen the meme. Much of human existance seems to be about people trying to prove each other right or wrong.
Matrix is overall a spec, not just the main homeserver implementation, synapse. So things like P2P matrix, and lighter homeservers that could work locally are in development. Then you could control your blocklists, as opposed to relying on a server to do that for you.
IMO the reason matrix specifically is so reliant on a few servers is because:
- The deployment / installation, even with ansible, is fairly difficult.
- Synapse is a system hog that pretty much requires a dedicated server.
These aren’t problems with federation, but the implementation of it. Federation overall is the only thing that has the potential to break the silicon valley fiefdom ( which only doesn’t feel as isolated, because there are so many users there ).
Yes matrix’s new slim dendrite server is being worked on with the intention for it to be used in p2p. I get there is many possible implementations of federation and I am totally down to take on ‘big tech’ it’s just to do that I think you not only need to match them but bring something better and I am not entirely seeing that yet. In fact I am even seeing subpar things like discoverablity, fragmentation, performance, complications like to all the rules and instances.
In fact I am even seeing subpar things like discoverablity, fragmentation, performance, complications like to all the rules and instances.
All of these problems would definitely would be worse in a p2p network. In a federated network they are easier to fix.
Which follows the rule that centralization is easier than decentralization, because p2p is more decentralized.
Reddit mods would block you if had subscribed to other communities they didn’t like. People created multiple accounts and still got banned.
Reddit could block/ban your accounts just as easily as Lemmy mods could.
Anybody could pull the material you posted on reddit, run an LLM, and imitate you.
Don’t know about slow, I’m OK waiting a couple of seconds. Don’t see that as a problem, nothing is crashing on my end
Every system has its advantages and disadvantages. It is the same whether it is centralized, federated, or P2P.
They do and difficulty to implement is one of them and because of that neither federation or p2p has been developed the best yet.
I am looking forward to the day everyone can just run an app with no specific servers and the platform works, and content is quickly and optimally distributed
Started cracking up at fiefdom
P2P is probably better for direct communication, but I hardly see how a P2P publishing platform (forum, microblogging, …) would work. As far as I understand, this would involve a permanently running home server storing at least your full history, right?
Add in you are at the mercy of your home server, you can lose your account have it immitated, and more.
Indeed, the main criterion to choose an instance (server) should be that you trust the people who run it. If no existing instance fit that criterion (or any other of your criteria, for that matter), you are free to create one.
Great yes you may find one that suites you better, but users now can end up isolated to their island
… which is also a criterion to look for an instance (server). Some look for a broadly federated network, others for a “safe space” kind of experience. Servers exist for both use-cases.
but now you are isolated for the previous island and maybe other
which makes sense when one considers moderation. If you are running a “safe space” instance, you don’t probably don’t want your users to be exposed to less moderated content, and you certainly cannot moderate one by one all the sensible content from across the federation. So it makes perfect sense to federate only with instances who have a reasonable code of conduct.
You either have to run multiple accounts or accept the limits.
Note that there’s another solution allowed by the following property: an instance can be federated with two instances who are not federated with each other. Concretely, this means that you can create an instance who both has a CoC for its own content strict enough to federate with safe space ones, and federates with instances who have a less strict CoC to access their content.
Thanks for your well written reply
The Matrix p2p project is not even alpha. You should join the p2p room @matrix.org if you have some curiousity. Problems like storing distributing the data are not fully solved. I think you will have a mix of devices including heavier servers. They are trying to integrate each device as another home server but thats not ideal I think.
I do think p2p can be done and for forum type platforms. These are just not solved problems because as is said among dencentralizers centralization is easy! federation seems a natural stepping stone to p2p, after all instances or servers are really peers or nodes themselves it’s just when you push out to lowerpower devices and so many of them things become a bit different. I surpose that is why the Matrix’s p2p dev Neil is working on the server now.
I think p2p will need content to be distributed among nodes. I am looking forward to ipfs intergration on matrix as the media performance is horrible.
Note that there’s another solution allowed by the following property: an instance can be federated with two instances who are not federated with each other. Concretely, this means that you can create an instance who both has a CoC for its own content strict enough to federate with safe space ones, and federates with instances who have a less strict CoC to access their content.
You have lost me a bit on this. So are you saying one of the duplicate instances acts like a filter to interact with the other safe spaces instances?
If that is the case this is the beginnings of what can just naturally exist on discussions.app
Ok my first sentence was very naive and due to my ignorance on p2p data storage. I do understand from your answer that it’s nontrivial.
About those instances things, say there are 3 instances. One instance A that is huge, where a lot of people that you know are, but is very laxist on its CoC. One instance B who aims at being a safe space for its users, and hence is not federated with instance A. Then you have instance C, who has the same CoC as instance B, so it can federate with it. Instance C can also federate with instance A. Hence my statement : one instance C can federate with two instances A abd B who don’t federate with each other.
Note that instance C is not a duplicate of anything, it is just an instance whose users are open enough to the sensible content of instance A and polite enough to not publish anything that would be offensive for the instance B users. Note that I don’t know about the details on how comments/mentions between instances A and C are perceived by instance B.
Can you say more about how content moderation and codes of conduct work on a p2p network?
Note that I don’t know about the details on how comments/mentions between instances A and C are perceived by instance B.
So I am wondering that too. How does content interacted with on A by C affect B. I know how B would want that which is not to see it. To me all these server to server rules are not ideal and should instead be the users themselves organizing on the platform setting the rules. I know you have that with each community but I am talking also platform wide. The instance does that with its blocks but thats very authoritaian and lacks nuance.
Can you say more about how content moderation and codes of conduct work on a p2p network?
I know how I would do it and how matrix is planning to do it. Matrix is planning to keep it the same as rooms currently do it which is the same as communities here. Server based blocking becomes quite pointless in p2p I think which leaves me wondering how matrix will handle spam as their current main method is to block servers. In p2p you can potentially have new servers continually appear and attack you so…
I would use a moderation approach as like discussions.app is trying but that also really needs to also use how they organize content. The advantage is the approach is grassroots and nuanced with everyone getting the most unique moderation you could expect outside say some wonder ai doing it personally for everyone.
I know you have that with each community but I am talking also platform wide.
Maybe to simplify the discussion let’s talk about a platform who is not subdivised in communities, like mastodon or facebook. Communities are already some kind of federation.
should instead be the users themselves organizing on the platform setting the rules.
Ok but nobody has the power to enforce the rules right? How do you deal with trolling and spamming? Does every user have to block every troll one by one?
I would use a moderation approach as like discussions.app is trying
which is?
Communities are already some kind of federation.
This is an excellent interpretation. On federated platforms there is federations inside federations. Its superfluous really, but it gets worse where communities get repeated.
Ok but nobody has the power to enforce the rules right? How do you deal with trolling and spamming? Does every user have to block every troll one by one?
Well it could depend on what you mean by enforce. There is moderators and they can enforce rules, it’s just each user is put kind of in an admin position to pick and choose the moderators. You can also have groups of moderators controlled by other users for any user to use, which is some part of how discussions.app is doing their ‘communities’ the other part is each community chooses what #tagged content to use. Aswell the instance could enforce certain moderators and infact some moderators must be set as default to keep the platform clean. The idea is these could be changed individually or with lists of moderators, or users just live inside the curated communities. The other thing is different instances could do it differently with a different set of default moderators.
The whole point of this type of platform/moderation is to solve problems seen on others. You won’t be able to own a topic or community ‘as such’ as users must consent for you to have power. The reality is people are lazy, stupid and will consent to crazy thing as seen from the last year. So because moderators provide a service stopping spam and abusive behavior then people will use them and they will have power. What this really prevents is moderators being bad actors and also people not having good moderation or situations with no moderation when its needed. Because anyone can moderate there should be a much higher supply of moderation and types of moderation. The types of moderation I think is where things can get interesting because there is a heap of behaviours that could be hidden and platforms would be much nicer places, but really it’s up to the users what they want and how they experience things.
There is moderators and they can enforce rules, it’s just each user is put kind of in an admin position to pick and choose the moderators.
Wait, if there are globally set modarators, how is this not a centralised network ? I mean ok it uses p2p technology so the data is not physically centralised on a single server. But the network itself, the graph of interactions, it is a single blob where every node is connected to every other. Or do I miss something?
Moderators are globally set just on that one instance (or really it’s an interface or frontend), but also changeable by the users but also depending how the instance is setup. The way they are doing the data it is all shared in blockchains so in that sense it is a blob, but depending on each user and how they have that content curated would change the interactions user to user. In theory you will have groups oblivious to other groups but users within them that cross over between and become social bridges. The network is the people. My idea is that these human bridges will eventually lower barriers between users blocking each other and change minds. The end result is a more connected less divided social platform and so too society. Following the axiom that communication solves all problems. By putting the barriers to communication into the users hands instead of third parties he will have control to remove them.
I think the best part of federation is there are no islands, you can make an account on an instance and follow content from the rest of the fediverse, you can even host your own instance just for you and follow everything else
