– DRAFT –
Data and Account Portability on the Social Web

Presentation by dmitriz

URI for presentation?

explanation that we have done portability in the past, e.g. on email

change currently not possible, data export. Archive example from X/Twitter

no export of social graph possible

no followers can be taken and exported

Social graph is the primary asset and normally not exported, not even upon GDPR request

Social WG has done a lot of work, based on that work some ecosystems were created

export screen from mastodon with lots of options and export of social graph is possible

want to be able to move your account profile and all metadata, interactions etc
… want to move our social graph

no option for import in traditional social media

mastodon has import screen. But can only import following, nothing else. Because it is technically hard.

can we do automated transfer from one service to another

id migration can be redirect

we can partially move the social graph

can do following

can partially also do followers. Can also send follower requests to the followers and ask them to follow again

move account doesn't transfer activities because it is hard

automated? Manual transfer is possible, but automatic is harder

live migration would be easier than export/import

so as long as both services still run, this could be done

UI question, automatic deletion of old account

could we add features that migration from dead services is possible and re-import on new provider

main difficulty is that authorship and identity is dependent on the server that may be dead

challenge: much of the identity is based on domains, but there is some limitation. We want to have non-URI names, what about the new cryptographic identity things coming up in W3C & IETF

disambiguate identifers for all authors

social services that track versions, need to import versions as well

standardise one or more export formats

If interested in the topic, join the social web CG.

standardise various formats for export

need a roadmap for technologies to standardise

<bengo> I wish this section wasn't dropped from CRTR. "(Linked Data Signatures are best used when authentication is meant to be "long lived" and attached to an object, such as verifying that an object truly was posted by this actor, and signed HTTP messages should be used when

<bengo> authentication or authorization is ephemeral.)"

unbundle the current instance model, need headless servers

chrishtr: mentioned extension to AP is portability, about Bluesky has that?

dmitriz: bluesky prot is wrestling with the same challenge, but have made some choices made. But Bluesky uses DID, allows for moving

evan: 10k servers and open protocol.

dmitriz: portability of objects, but??
… goal is the same: data portability

bengo: integrity was removed from the spec but should be put back?
… certain implementations do this

evan: implicit in this conversation, lose trust relation between user and service provider.

<dmitriz> link to the slides: https://docs.google.com/presentation/d/1SSFrS2ke2tqMZrRUGyIPsOXIfS62qZFu0JcyBVSz9rI

evan: difference in fediverse is that there is no such fixed relation as employment. People are much more nomadic because unhappy with moderation e.g.

but may change over time

<bengo> rigo: to clarify, I just mean that normative text about how to do client-side integrity/authenticity via (what was then called) "Linked Data Signatures" had to be removed after CR before TR because there was not a stable spec to refer to, but now there are better client-side signature vocabs to mix into ActivityPub JSON objects/requests

??: what defines the scope, want to work on even larger context.
… we want to have a larger scope, not only ACL thinking (we want access to this and that data)

rigo: question about larger use of AP outside social web.
… question about portability of moderation

??: different approach to portability, look for pragmatic approach via own domain.
… use with personal domain name

<dmitriz> https://about.iftas.org/

Lisa: Respond aobut moderation. Moderation is not individual and we are scoping individuals. But it is also a very important topic and shoudl be in scope

bengo: Federated ID based servers. Moderation of content of that post.

<bengo> do moderation of hash of content, not URLs to locations of content that can be changed over time

domain based moderation is limited and should go beyond. But non domain based object identifiers could be used

Laurens: Linked Web storage has similar challenges

pchampin: linked data

rigo: W3C, involved in social things since ever
… Seeing how ActivityPub has taken off, we must address the challenges, esp. in moderation.
… We have proven that moderation in a federation setting was possible.

chrishtr: do you mean Mastodon?

rigo: not specifically. I think 1/3 of my timeline does not come from Mastodon.

tantek: want to see a pain-point user scenarios for portability. The user stories really helped us to advance in the Social WG

<dmitriz> user stories document (in progress): https://codeberg.org/fediverse/fep/src/branch/main/fep/73cd/fep-73cd.md

tantek: want to go beyond technical details. Right now it a lot of work. Tantek told 2 user stories / minute and scribe wasn´t able to follow

evan: Presentation:
… Research Director of Social Web Foundation.
… how to we get from Specification to a reference implementation
… or proof of concept

bengo: Lifewire in the past. Protocol labs, DIDs in bluesky, conformance testing

@@ Google: just learning

Chris Harrelson, working at google and working on Chrome. Want to learn more. Interested in having the federated prot and joined the CG recently

Lisa Dussault: Data transfer initiative

AronCoburn: Working at Inrupt,

standardized storage will be key in Activity Pub.

client to server storage exchange

@1 worked in the Social WG in the first year. Social graph should be main effort. Basic user profile that can be done via @2 API
… standardizing social graph

<tantek> "focus and make the social graph portable" is literally what we did with XFN, back in 2003 https://gmpg.org/xfn/

MattLee: have 70k users and want to allow for users to move with their data

dmitriz: standardizing the social graph. How many times did people reivinent identification and contact?
… we need to take that into account
… we need to take that into account and extend and be involved in address books

<tantek> note that the latest version of vCard incorporated XFN: https://datatracker.ietf.org/doc/html/rfc6350

pchampin: new WG on mydata and participated the first call

discussion about vcard,

dmitriz: use subset of vcard 4.0

pchampin: one challenge was to find an alignment

<bengo> activitypub at mydata.org 2018 https://www.youtube.com/watch?v=ikCumzhfV9k

rigo: identity is key as we need ephemeral and less ephemeral ids