garden/distributed press 2021-05-19.md at master

mirror of https://github.com/flancian/garden.git synced 2026-08-07 15:06:19 +00:00

Files

T

flancian a13cfb1dfd [[snapshot]]: 2021-05-19 19:06:49

Affected files:

.obsidian/workspace
"distributed press 2021-05-19.md"

2021-05-19 19:06:49 +02:00

7.5 KiB

Raw Permalink Blame History

an instance of distributed press
- 2021 05 19
benedict lau
- hypha
  - interested in building a tool that publishes through many different protocols
    - sharing social messages seamlessly
  - flancian
  - paul frazee
  - guo
    - matters.news
    - serving mostly chinese communities
    - looking at ipfs, hypercore
    - working on a protocol similar to unwalled garden, but also thinking of integrating rss
      - also interested on webmentions
  - vera
    - working on agora
    - background in distributed social systems, worked on (sp?) status, before mastodon
  - fauno
    - (mic incoming) :)
paul frazee
- distributed vs federated
  - a bit stuck here currently
  - the way to conceptualize
  - federated ~ mastodon
  - distributed ~ ipfs mesh
    - completely open system, if you're in the mesh you're in
    - technologically, user experience wise is great
    - trying to figure out how to deal with the cost structure
    - in the case of social media, you're making your own twitter
    - how do you get economies of scale worked in there?
      - filecoin?
      - gateway model
        
        ipfs gateway: there is more than one of them
      - if your mesh contains social data, you get a social network
    - ben what are the costs involved?
      - paul unsure but they all seem associated to a side business so far
    - advantages
      - ?
      - serverless clients are possible
    - ben
      - less well maintained gateways become issues for user experience
      - censorship
      - guo yes, gateways become a centralized point
        
        gateway should be a backup after serverless client
      - paul frazee
        
        spent 6-8 months working on a social network back in beaker
        
        some stuff that stops you:
        
        links that are a new protocol are an issue
        
        ux killer on stuff like twitter
        
        didn't have a push messaging system back then
        
        didn't want to have to deal with spam
        
        ux killer: no notifications unless mutual follow
        
        aggregator
        
        try to bring in positive properties of the system
      - flancian
        
        metcalfe's law
      - ben
        
        web mentions
guo
- content feed + social graph as a distributed publication and social network
- people can store their content feed and social graph anywhere
- user control their data
- search
- separating storage and aggregation
paul frazee
- database vs datamesh
- federation puts application logic in the wire protocol
flancian
- datamesh should probably be treated as a commons
guo
- we could avoid many governance issues at the database layer
- ben this sounds like the scuttlebutt and hypercore model?
  - paul when last involved there was no global entrypoint, it was gossip protocol
  - pubs were store and forward relays
  - information would flow 100% on the basis of the social graph
  - led to complexities; people could give you keys and still be unreachable
flancian
- paul
  - how do you end up in a situation in which governance is required?
- guo
  - datamesh seems like it does not require governance, but the aggregator does
  - (interlay vs overlay?)
  - ben aether (sp?)
    - distributed reddit
    - distributed moderation
    - liquid democracy-like
  - ben
    - we need to agree on:
      - schemas
      - distribution protocol
      - aggregator censorship
    - what happens when your computer ends up hosting hate speech?
      - you can stop the flow in the mesh and that applies negative evolutionary pressure
    - paul aether does default allow, I do default deny
  - flancian
    - minimum exchange format
  - guo
    - choice of markdown vs html
    - chose html over markdown, it supports arbitrary media
    - encryption
      - exclusive content
      - hard to define protocol-level encryption standard
      - bake the encryption message into the html
        
        mdx
  - ben decided to first publish/store things as html
    - most people want to have something ready-to-publish
    - but a consumption pattern arises: you copy/paste whole articles
    - on the dweb there is an opportunity for this as we can publish different formats of the same concept
    - get content in more than one format
  - paul
    - schemas in general
    - I would describe what you were discussing as "rich text format"
    - orthogonal to schemas
      - fighting with schemas since secure scuttlebutt
      - pissed off rdf people
      - reached https://ctzn.network/comment
      - scuttlebutt: did not impose any schema obligations; disastrous approach
      - unwalled gardens: enforced types
      - ctzn network: urls identifying the schemas with a machine readable format
        
        rejects messages that don't comply with the schema
        
        believe in this approach
        
        if you don't, you end up in the tragedy of the commons or failure in collective action (scuttlebutt)
        
        compatibility with the schema dictates everything you can do
        
        you have a feature, you add a dependency
        
        had a post type
        
        would be nice if you have a topic
        
        if you are not subscribed to cats, you won't see it
        
        but you can't rely on it as many clients won't support it
        
        you might end up spamming old clients
        
        fatal ambiguity
        
        once you see this, schemas start seeming risky to change
        
        you can do extension points, but would suggest against
        
        current approach:
        
        well defined schemas
        
        people can determine their own schemas
        
        control the ctzn network schemas
        
        can have extension tables
        
        ben
        
        feel like scuttlebutt has a schema, otherwise you couldn't have more than one client
        
        what it lacked, it seems, is versioning?
        
        how do you manage upgrades?
        
        paul the problem is multiple clients controlled by different devs
        
        flancian what do you really need for features -- usually you'd want to have the minimum minimum schema requirements?
        
        paul frazee
        
        web mentions: exploring now
        
        unwalled garden
        
        difference between it and ctzn network
        
        suppose you want to have the ability to have mentions in your comments
        
        you have to keep parsing for entities if you store text or text-like
        
        using hyperbee in ctzn
        
        it can list directories or keyspaces in order
        
        more like a database technology
        
        for distributed press/unwalled garden: file oriented approach can work over http
        
        guo in the comments problem, who owns the post data?
        
        the post it's on could 'own' it
        
        the comment writer could own it
        
        this seems safer?
        
        fauno
        
        if we allow for comments, who is responsible for keeping them? hateful comments/censorship/etc
        
        webmentions: fascist websites pinging you
        
        people don't have time to moderate
        
        ben same thing is seen in wordpress
        
        fauno web mentions and how to keep bad parties out
        
        flancian how will moderation work in ctzn network?
        
        paul would like to have a great answer :)
        
        reputation systems
        
        no one size fits all
        
        this lets you scale the work
        
        reputational graph approach
        
        subjective moderation
        
        elected moderators

7.5 KiB Raw Permalink Blame History

7.5 KiB

Raw Permalink Blame History