Let's build a decentralized social network together with Logseq
briansunter.com
briansunter.com
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.
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.
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.
{: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}}
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!
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...
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.
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
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).
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...