I very much think this is a pragmatic approach. I also truly wish for a LinkedIn and Facebook without a single corporate entity profiting in the middle (decentralized/web3). However, for this to ever truly work, the UX just has to be heads and shoulders above this approach, imo.
Dont underestimate the value of a community of enthusiastic avid folk! This can work & do great things & have amazing value & be awesome, without having to scale to a billion people.
I agree a lot that this is a super pragmatic approach. Loqseq users are, by and large, enthusiasts, who already are willing to go far. Giving them some conventions for how to express social metadata, and then letting the existing graph based product do it's thing, building some books of queries people can use: it seems like a wonderful organic strategy. It invents very little, and rides on the strength of what already is.
Logseq has a super powerful plugin system that can have whole react apps in it.
Once people get set up with public profiles, a plugin could be built to make it easier to create profiles, friends, browse, etc. Just some simple buttons and fields to start.
Dear HN: Please stop making this comment, or phrase it differently. People are allowed to be different and things are allowed to exist without being practical.
Please stop making comments asking people not to make comments whose intent you can't discern.
Either: The author hasn't heard of twtxt, and might be intrigued at the possibility of compatibility, or they have heard of it and may have reasons for being different which may add to the discussion.
Your comment assumes the worst intent (I never said or suggested differences shouldn't be allowed) and deadends good discussion.
All these attempts are not as easy to use as the dominant network.
Friction-free is very hard. Are the masses going to switch purely for privacy preservation?
See our Internet-deployed social network from 13 years ago. Without any server and no gatekeepers. Its fully self-organising. It lacks polish. You simply can not compete against the thousands of engineers which polish the dominant social network. https://www.researchgate.net/publication/221015322_A_Gossip-...
So twtxt is a text protocol for sharing social media posts which can be shared by hosting a plaintext file. I bring it up because it's extremely close to what the author proposed.
It doesn't look that close; the model of interaction seems to be completely different. twtxt is a command line tool with flat files and no obvious way of linking content, and logseq is an outliner with hierarchical files and bidirectional links created with wiki-like syntax.
Someone without advanced knowledge of the command line would not know how to start using twtxt, while using logseq is as simple as any plain text editor.
Twtxt is more of a sharing standard that can be statically hosted rather than a command line tool: Yarn.social, for instance, is a web app that uses Twtxt format for social microblogging.
What do you see about twtxt that seems similar to you?
Logseq doesn't have "posts" as such, nor is there a protocol here for sharing. There's nothing graph related in twtxt. It also doesn't integrate with the personal knowledgebase people are already using. I'm not sure why twtxt would come to mind as similar.
Since everything is just data, if you wanted to write a logseq query to transform and export your graph in this format to be compatible with their CLI, you could (especially with the new graph as api tools)
Unless filtering algorithms are transparent, can be configured by users and errs towards allowing speech that is acceptable globally instead of just in Silicon Valley, then it just incites feelings of biased censorship.
> More spam, more misinformation, more hate speech, more everything bad.
I don't think you would get all that from a network like this. You only see content from people you follow, there is no algorithm pushing viraliced posts and advertisements on your "feed" of "things you want to see". If you are careful about who you follow, it could work like a BBS or a web forum.
The "echo chamber" is a made up thing. We've always mostly socialized with people close to us. What makes it toxic is the existence of reward systems that specifically target the most outrageous statements because we need to goose those enragement numbers.
Misinformation gets pushed because it generates clicks. As a result, it floats up in everybody's stream. More eyeballs -> more clicks -> more visibility -> more eyeballs.
Will there be small groups that traffic in misinfo? Yes, absolutely - but the large scale impact is gone once you remove the ability to artificially boost visibility.
This seems like the exact opposite claim to the top post. First we said, sorry, you don't have permission to innovate & explore & connect with people unless you have moderation and control. Ok, yes, this is opt-in following: that's how we moderate.
Now we're saying, sorry, you don't have permissions to innovate & explore & connect with people unless you also admit in the public & get exposed to randos.
In general I think both of these views impose unreasonable limitations & constraint upon people getting out there & trying things. Not all value that is created has to be perfect value, in my humble opinion. Should we try to keep an eye on these factors? Absolutely. To you I'd say, blogrolls (what this effectively is) is generally a pretty good way to explore, that in the blogging era different people would have different blogrolls & it was always fun to cruise around, & check out new people you hadn't checked out before. Even if you didn't end up following that person, they might have their own blogroll that had some interesting people on it. Connectivity, being able to go to different places, is a precondition for diversity, and this unqualifiedly enables connectivity. Worrying about that connectivity being an insular one is premature a concern.
I personally prefer to use a chat system that uses an open standard, is not moderated and requires me to moderate the channels I manage/own. Being an open standard means I can use a generic client with regex filters that could be conditionally applied to usernames, chat text, links, emojis, whatever I want. That is just my own preference but should say something given how lazy I am ... that I would be willing to take the time to ban something myself.
At least one of the IRC daemons [1] for example allows people to toggle any of their channels to have a word filter on/off. +G/-G Channel operators can contribute to word filters or ask an IRC admin to remove one. Since I am linking an IRC daemon I should link one of the web front-ends [2] as so few these days use IRC. I assume these days both of the linked projects are probably dockerized and mostly pre-configured.
As a side note and real world example anyone can already do this today with anything that uses HTML because it is an open standard. [3] I have not seen the word "butt" in ages.
From what I remember, before we had highly moderated/censored social monopolies we had a lot less spam and misinformation on the old forums and chat spaces.
I'd much rather be on HN, biggerpockets, or some auto forums than on twitter or facebook.
I'd readily give up the spam-free and misinformation-free networks of Facebook and Twitter for a moderator who knows what satire is, doesn't get nudes and dunes confused, and can generally parse context.
Private communities tend to have less of all of that, unless maybe you're building them with the worst of humanity, in which case I guess it's good that Facebook is there to keep you and your friends from doing Everything Bad. Yikes.
Out of curiosity: what level of AI or human-driven moderation would it take to prevent you from linking your own site to, let's say, a Hitler Fan Club page? Is that something you feel you might do by accident without Facebook there to tell you that is Bad?
Or, say a random site said it's Good to drink bleach - would a browser extension be sufficient to prevent you from linking to them, or would it require an actual human like Keegan-Michael Key to show up and slap the mouse out of your hand and explain to you in-person that this is Stupid and Bad and Dumb?
I fear for people who rely on Twitter Safety to keep them from doing a Big Dumb.
Ideally you would have a platform where you could enable or disable filters for spam, hate speech, etc (independently)... to make sure that they don't filter out stuff that you want to see.
I think we need to see more implementations of the already existing federated protocols. We need that one app that will truly explode and bring the fediverse to the masses. Mastodon is okay but it's just a microblogging platform; it's not really too innovative.
I would agree, but encryption (let alone end-to-end) is totally absent from specs like https://www.w3.org/TR/activitypub/. The word isn't even mentioned once.
I don't just want federation, I want to know that my/friend/family's information (or at least private messages) is private because I know they won't "self host".
Maybe the solution, instead, is to create a simple browser plugin that accepts ciphertext and keys and decrypts blobs of text. It's ugly, but you could render it in place with some creative tagging of the blobs.
1. define a schema for the blob. recipient pubkey, signature, hash, etc
2. define a delineator or symbol of some sort to encapsulate the serialized blob so our plugin recognizes it
3. parse blob into readable text for the user
4. provide interface to manage/create keys/messages and load the blob to the clipboard for users to paste into whatever.
voila. please hold your applause, I'm going to need one of you to MVP this by the end of the week.
Let's do it. I have similar ideas and we can use blockchain to anchor some stuff there so that the extension doesn't need any servers to get data, only Blockchain node to connect to
If the future popularity of the fediverse depends on a killer app instead of the various benefits of federation then it's doomed. As soon as such a killer app appears, a non-federated competitor could copy the idea and out-compete the fediverse version because it is not hamstrung by having to stick to the (inherently slow-moving) fediverse standards.
A similar thing happened to IRC, which _still_ lacks heavily used features that other chat apps like discord and slack have had for years and years. Simlarly, end-to-end encryption for emails is unlikely to ever become a thing despite the privacy advantages.
IRC isn't and never was federated. It is a centralized chat protocol where you connect to isolated networks.
Similarly Matrix is making massively huge gains in the chat space and is federated. It's not taking over slack or discord yet, but is gaining features and ground.
I feel like Mastodon's standout feature is also its biggest drawback when it comes to adoption. The vast majority of end-users don't understand or care to understand the need for different instances or which fediverse they should join, nor do they have interest in setting up their own.
> Mastodon is okay but it's just a microblogging platform
Why would we a social network to be anything other than a microblogging platform? In case it lacks some conveniences you would like it to have, why not just add them as optional features?
There are so many ways to build decentralized social networks. It’s a 1000% solved problem at this point. They never get traction because they always require more learning and work than 99% of people are willing or able to fit into their lives. You jump through more hoops to reach fewer people.
This isn't a conventional social network. This is a knowledge-graph system, building conventions to link to other knowledge-graph systems.
This effort is to enhance an already existing cherished information system, to make it more expressive, and to seed some very basic interconnectivity which was not possible. Logseq is a wonderful digital garden[1], a personal place people tend, and I don't think anyone in this space would tell you this is a solved problem, that we are done exploring & understanding how to do this knowledgebase work, how to digital garden.
These personal-knowledgebases are rich & personal platforms. But so far, they have largely been alone & isolated. Different patches. FedWiki[2] is perhaps the lone counter-example, of an interconnected knowledgebase, but it has it's limitations with what it is & how it interconnects. Taking another best-of-breed product, like Logseq, and layering in some minimal, small conventions for how to define interconnect, how to make a blogroll: it seems pure & un-objectionable to me.
Nice! I couldn't have said it better myself. I did use a linkbaity title like "Decentralized social network" to get people interested. It's more of a simple link ring.
But since profiles are public queryable structured static data, you can do a lot more with it in the future.
That's right. What the current batch of social networks are missing is good tools for knowledge consolidation; there are no tools more complex than the single thread intermixed with comments.
A knowledge graph would work more as a Wikipedia of personal projects and topics of interest, with the possibility to build special data visualizations on top of it. For example, this could be the perfect medium for building distributed discourse aggregators of opinions and rational debates, like Kialo[1] and Parlia[2] but without a single company controlling the content. Social networks transform these debates into confrontations, while a structured knowledge tool could create a structured knowledge base from this kind of exalted conflicting positions.
Building a new decentralized social network reminds me of building yet another Linux desktop distribution. While you might find a niche, you'll never get enough traction to compete with the big players, being Android and Windows for desktops and Facebook, Twitter and others for social networks.
I've been playing with Logseq for something similar - using the Zotero integration to share research notes (and sometimes blog). The attributes and queries are super clever, too.
I'm not sure it'll ever replace Facebook/Twitter/et al, but it is nice to have a semi-private network of people who can have grown-up discussions, or express thoughts deeper, or with more nuance, than 100-some characters.
There were only a few Facebook groups that I ever got any value out of, which were mostly small professional groups that banned political bickering. My mental health has definitely improved by going back to old-school forums, and I've noticed a lot of others doing the same over the past year or so - especially with Lemmy, which is maybe even a bit easier to set up.
Hey Everyone! Really appreciate this discussion. Here's some extra context.
Logseq is an open source notetaking app (similar to Roam) that has a feature where you can publish your notes as a static site.
It has some advanced features where you can attach data to pages and bullet points (like some basic bio info, like name, site, twitter, tags) that you can query.
There are some public graphs out there, but they're hard to find and figure out who made them. The main goal is just to create a "link ring" where we can find each other.
This spec suggests a standard page on each graph called [[logseq-social/profile]]
That way, when you find a public graph, you can go there to find out about its author and who they follow. You can download that page to your notes, and a simple logseq query makes the bio information of each user you save show up on your follower list.
The idea is really simple, just agreeing on a convention for a "profile" page on logseq that has some standard tags.
Logseq has a really powerful plugin system that can have a whole react app in it, so down the line you could make a UI for friending, browsing, etc when more people have public graphs.
Plus, there's now tooling to host your logseq notes as a programmatic api.
So having some metadata standards opens up some interesting possibilities.
There's lots of other stuff out there like this, but since this is just public static data there's hopefully a way to make it interoperable with existing solutions. If anyone knows of a good existing "bio" or link ring schema, I would be interested in taking a look.
I have other fun ideas like hosting graphs on IPFS, or even some day having every logseq bullet point block be an immutable record on ipfs for use by everyone.
Another example out there is agora, which seems to be somewhat compatible with markdown flat files that logseq uses https://anagora.org/agora-editor
Love the broad direction, and I'm interested in digging into this area myself (decentralized social networks and applications). Recently been studying Lens Protocol a bit (https://lens.xyz/) which has a social graph system that records to the blockchain (Polygon in this case). Check that out and let me know what you think of it.
I like the idea of hosting logseq graphs on IPFS. It does seem that with other decentralized networks I've looked at, that scaling up to Facebook level would not be practical as the size of the graph grows. It almost seems like you have to sacrifice decentralization for a centralized database. Do you think logseq would perform on that scale or would it be limited to niche social networks?
Either way, signed up for the newsletter, looking forward to following your progress on this! And always down for a chat with an old friend!
Hey Chris! It's definitely an area I'm interested in. The super decentralized IPFS blocks is a little bit of a pipe dream, but it seems like it could work in theory.
I like logseq because the "primitive" where they store/organize the data called a "block" which is pretty well thought out. It's just an object with content and a pointer to a parent. Seems like you could content hash it and store it on ipfs forever. Haven't fully thought out what that would look like yet though. It could be developed as a plugin for logseq. Seems like it could scale, lots of other problems like search though.
{:block/uuid #uuid "62a8318e-ff36-4f97-9ef5-f19c3f3041bf",
:block/left {:db/id 67152},
:block/format :markdown,
:block/content
"Users create a bio under a page called `[[logseq-social/profile]]`",
:db/id 67144,
:block/path-refs [{:db/id 20}],
:block/parent {:db/id 67160},
:block/unordered true,
:block/page {:db/id 20}}
53 comments
[ 2.8 ms ] story [ 83.8 ms ] threadI agree a lot that this is a super pragmatic approach. Loqseq users are, by and large, enthusiasts, who already are willing to go far. Giving them some conventions for how to express social metadata, and then letting the existing graph based product do it's thing, building some books of queries people can use: it seems like a wonderful organic strategy. It invents very little, and rides on the strength of what already is.
Once people get set up with public profiles, a plugin could be built to make it easier to create profiles, friends, browse, etc. Just some simple buttons and fields to start.
Either: The author hasn't heard of twtxt, and might be intrigued at the possibility of compatibility, or they have heard of it and may have reasons for being different which may add to the discussion.
Your comment assumes the worst intent (I never said or suggested differences shouldn't be allowed) and deadends good discussion.
See our Internet-deployed social network from 13 years ago. Without any server and no gatekeepers. Its fully self-organising. It lacks polish. You simply can not compete against the thousands of engineers which polish the dominant social network. https://www.researchgate.net/publication/221015322_A_Gossip-...
Someone without advanced knowledge of the command line would not know how to start using twtxt, while using logseq is as simple as any plain text editor.
Logseq doesn't have "posts" as such, nor is there a protocol here for sharing. There's nothing graph related in twtxt. It also doesn't integrate with the personal knowledgebase people are already using. I'm not sure why twtxt would come to mind as similar.
https://twtxt.readthedocs.io/en/latest/user/twtxtfile.html
Since everything is just data, if you wanted to write a logseq query to transform and export your graph in this format to be compatible with their CLI, you could (especially with the new graph as api tools)
This is the most similar project I've seen to what I'm thinking https://anagora.org/agora-editor
More spam, more misinformation, more hate speech, more everything bad.
Sure, you get, maybe, some more privacy. But you can get that by just staying off social media.
I don't think you would get all that from a network like this. You only see content from people you follow, there is no algorithm pushing viraliced posts and advertisements on your "feed" of "things you want to see". If you are careful about who you follow, it could work like a BBS or a web forum.
The "echo chamber" is a made up thing. We've always mostly socialized with people close to us. What makes it toxic is the existence of reward systems that specifically target the most outrageous statements because we need to goose those enragement numbers.
Misinformation gets pushed because it generates clicks. As a result, it floats up in everybody's stream. More eyeballs -> more clicks -> more visibility -> more eyeballs.
Will there be small groups that traffic in misinfo? Yes, absolutely - but the large scale impact is gone once you remove the ability to artificially boost visibility.
Now we're saying, sorry, you don't have permissions to innovate & explore & connect with people unless you also admit in the public & get exposed to randos.
In general I think both of these views impose unreasonable limitations & constraint upon people getting out there & trying things. Not all value that is created has to be perfect value, in my humble opinion. Should we try to keep an eye on these factors? Absolutely. To you I'd say, blogrolls (what this effectively is) is generally a pretty good way to explore, that in the blogging era different people would have different blogrolls & it was always fun to cruise around, & check out new people you hadn't checked out before. Even if you didn't end up following that person, they might have their own blogroll that had some interesting people on it. Connectivity, being able to go to different places, is a precondition for diversity, and this unqualifiedly enables connectivity. Worrying about that connectivity being an insular one is premature a concern.
At least one of the IRC daemons [1] for example allows people to toggle any of their channels to have a word filter on/off. +G/-G Channel operators can contribute to word filters or ask an IRC admin to remove one. Since I am linking an IRC daemon I should link one of the web front-ends [2] as so few these days use IRC. I assume these days both of the linked projects are probably dockerized and mostly pre-configured.
As a side note and real world example anyone can already do this today with anything that uses HTML because it is an open standard. [3] I have not seen the word "butt" in ages.
[1] - https://www.unrealircd.org/docs/Badword_block
[2] - https://github.com/thelounge
[3] - https://addons.mozilla.org/en-US/firefox/addon/foxreplace/
I'd much rather be on HN, biggerpockets, or some auto forums than on twitter or facebook.
Private communities tend to have less of all of that, unless maybe you're building them with the worst of humanity, in which case I guess it's good that Facebook is there to keep you and your friends from doing Everything Bad. Yikes.
Out of curiosity: what level of AI or human-driven moderation would it take to prevent you from linking your own site to, let's say, a Hitler Fan Club page? Is that something you feel you might do by accident without Facebook there to tell you that is Bad?
Or, say a random site said it's Good to drink bleach - would a browser extension be sufficient to prevent you from linking to them, or would it require an actual human like Keegan-Michael Key to show up and slap the mouse out of your hand and explain to you in-person that this is Stupid and Bad and Dumb?
I fear for people who rely on Twitter Safety to keep them from doing a Big Dumb.
Privacy is a completely different topic though...
I don't just want federation, I want to know that my/friend/family's information (or at least private messages) is private because I know they won't "self host".
1. define a schema for the blob. recipient pubkey, signature, hash, etc
2. define a delineator or symbol of some sort to encapsulate the serialized blob so our plugin recognizes it
3. parse blob into readable text for the user
4. provide interface to manage/create keys/messages and load the blob to the clipboard for users to paste into whatever.
voila. please hold your applause, I'm going to need one of you to MVP this by the end of the week.
The real issue is the actual exchange of messages
A similar thing happened to IRC, which _still_ lacks heavily used features that other chat apps like discord and slack have had for years and years. Simlarly, end-to-end encryption for emails is unlikely to ever become a thing despite the privacy advantages.
IRC isn't and never was federated. It is a centralized chat protocol where you connect to isolated networks.
Similarly Matrix is making massively huge gains in the chat space and is federated. It's not taking over slack or discord yet, but is gaining features and ground.
Why would we a social network to be anything other than a microblogging platform? In case it lacks some conveniences you would like it to have, why not just add them as optional features?
This effort is to enhance an already existing cherished information system, to make it more expressive, and to seed some very basic interconnectivity which was not possible. Logseq is a wonderful digital garden[1], a personal place people tend, and I don't think anyone in this space would tell you this is a solved problem, that we are done exploring & understanding how to do this knowledgebase work, how to digital garden.
These personal-knowledgebases are rich & personal platforms. But so far, they have largely been alone & isolated. Different patches. FedWiki[2] is perhaps the lone counter-example, of an interconnected knowledgebase, but it has it's limitations with what it is & how it interconnects. Taking another best-of-breed product, like Logseq, and layering in some minimal, small conventions for how to define interconnect, how to make a blogroll: it seems pure & un-objectionable to me.
[1] https://maggieappleton.com/garden-history
[2] http://fed.wiki.org/view/welcome-visitors
But since profiles are public queryable structured static data, you can do a lot more with it in the future.
A knowledge graph would work more as a Wikipedia of personal projects and topics of interest, with the possibility to build special data visualizations on top of it. For example, this could be the perfect medium for building distributed discourse aggregators of opinions and rational debates, like Kialo[1] and Parlia[2] but without a single company controlling the content. Social networks transform these debates into confrontations, while a structured knowledge tool could create a structured knowledge base from this kind of exalted conflicting positions.
[1] https://www.kialo.com/
[2] https://www.parlia.com/
https://briansunter.com/graph/#/page/62a6cc7b-e921-4890-813a...
I'm not sure it'll ever replace Facebook/Twitter/et al, but it is nice to have a semi-private network of people who can have grown-up discussions, or express thoughts deeper, or with more nuance, than 100-some characters.
There were only a few Facebook groups that I ever got any value out of, which were mostly small professional groups that banned political bickering. My mental health has definitely improved by going back to old-school forums, and I've noticed a lot of others doing the same over the past year or so - especially with Lemmy, which is maybe even a bit easier to set up.
So it's good for the POSSE philosophy, in which you're writing about it on your site and sharing/commenting it on social media.
https://briansunter.com/graph/#/page/62a6cc7b-e921-4890-813a...
Logseq is an open source notetaking app (similar to Roam) that has a feature where you can publish your notes as a static site.
It has some advanced features where you can attach data to pages and bullet points (like some basic bio info, like name, site, twitter, tags) that you can query.
There are some public graphs out there, but they're hard to find and figure out who made them. The main goal is just to create a "link ring" where we can find each other.
This spec suggests a standard page on each graph called [[logseq-social/profile]]
That way, when you find a public graph, you can go there to find out about its author and who they follow. You can download that page to your notes, and a simple logseq query makes the bio information of each user you save show up on your follower list.
The idea is really simple, just agreeing on a convention for a "profile" page on logseq that has some standard tags.
Logseq has a really powerful plugin system that can have a whole react app in it, so down the line you could make a UI for friending, browsing, etc when more people have public graphs.
Plus, there's now tooling to host your logseq notes as a programmatic api. So having some metadata standards opens up some interesting possibilities.
https://github.com/logseq/nbb-logseq/tree/main/examples/fly-...
There's lots of other stuff out there like this, but since this is just public static data there's hopefully a way to make it interoperable with existing solutions. If anyone knows of a good existing "bio" or link ring schema, I would be interested in taking a look.
I have other fun ideas like hosting graphs on IPFS, or even some day having every logseq bullet point block be an immutable record on ipfs for use by everyone.
Another example out there is agora, which seems to be somewhat compatible with markdown flat files that logseq uses https://anagora.org/agora-editor
Love the broad direction, and I'm interested in digging into this area myself (decentralized social networks and applications). Recently been studying Lens Protocol a bit (https://lens.xyz/) which has a social graph system that records to the blockchain (Polygon in this case). Check that out and let me know what you think of it.
I like the idea of hosting logseq graphs on IPFS. It does seem that with other decentralized networks I've looked at, that scaling up to Facebook level would not be practical as the size of the graph grows. It almost seems like you have to sacrifice decentralization for a centralized database. Do you think logseq would perform on that scale or would it be limited to niche social networks?
Either way, signed up for the newsletter, looking forward to following your progress on this! And always down for a chat with an old friend!
I like logseq because the "primitive" where they store/organize the data called a "block" which is pretty well thought out. It's just an object with content and a pointer to a parent. Seems like you could content hash it and store it on ipfs forever. Haven't fully thought out what that would look like yet though. It could be developed as a plugin for logseq. Seems like it could scale, lots of other problems like search though.
https://github.com/logseq/logseq/blob/d0755ef161eb826e452b92...The other dapp "primitive" I'm interested in is dChat. Seems like there could be some overlap between blocks and chat built on the same primitive.
For now, I'm mostly focused on getting technical users to just make their own personal sites since it's really easy with logseq and netlify.
I'm not a dapp expert at all (though I do host my traditional static site on ipfs https://gateway.pinata.cloud/ipfs/QmRDPhzAjBdBovPhhPKt4tanFp...)
Happy to hear from you and I'll definitely look into the lens protocol!
If anyone sets this up definitely contact me on twitter so I can add you to my following list.
And I believe the self-hosting option inevitable. Including your own name (domain).