Mastodawn

Gonna try and negotiate an acceptable implementation with the masto devs (if u see this no pressure! Ik ur busy), will take a bit of background work like to implement an integrity proofs FEP, so may take a bit, but if you wanna write this with me I would love to work together on this. hopefully its implementable, it's my first spec document ive ever written so idk what I'm doing rly I just think I can write it and wanna move my posts around and break down the feudal fedi

Also if anyone wants to write about the social/political/governance problems caused by not being able to move your stuff with you when you move for motivating context, I'd be glad to put it in, I haven't had the energy to write that yet

Also it treats post backups as posts (I mean they should anyway, importing a list of posts is the same as posting a list of posts, and backdating is not a crime), and if instances support exporting your private key with account backups, then you can create a Move activity without needing the source instance, so account backups are just... your account in a file, so you can restore if your instance goes down or becomes hostile.

CBOR and CAR files and Merkle Search Trees are cool, but so is a zip file of JSON.

(This is spec'd as a "should" BC there are some problems to work out incl. Key rotation and etc. but the mechanism for remapping is general and driven by the new instance)

also it theoretically allows you to do a chain of moves, including moves without changing actor URI to rotate keys, so you can deal with repeated moves or old posts. idk it may help with key rotation but I know enough about cryptography to know that I don't know enough to say that confidently.

Speaking of which, shoutout to @ansuz for taking the time to give input and info on some of the cryptography stuff (if there are places where I am super wrong about things that is all me though)

silverpill Nov 30

@jonny Hey, are you still working on this FEP?

I wrote a short review a couple of weeks ago. Sometimes SocialHub email notifications don't work, so here's a link to my post in case you missed it: https://socialhub.activitypub.rocks/t/fep-1580-move-actor-objects-with-a-migration-collection/8111/22

FEP-1580: Move Actor Objects with a `migration` Collection

Supporting instances MUST indicate their support of this FEP by including its namespace in the @context of affected Actor objects. I think it would be better to use a regular property (e.g. supportsObjectMigrations) because most Fediverse platforms don't use JSON-LD and don't parse @context. Some may not want to add @context to their objects. Another option is FEP-844e. A "proposed move" is an Offer[Move] activity emitted by the source instance addressed to the target instance's shared inbox (an...

SocialHub

jonny (good kind)Nov 30

@silverpill
Yep! Just was on vacation and had a conference to do. It's on my list for Monday to do some followup and updates

despite receiving probably the most notifications of anyone on fedi, i have been told that this may be of interest to @pluralistic , having written extensively about the importance of the power of being able to leave re: preventing feudalism and lock-in

jandi Oct 15

@jonny No response? Maybe you have to invoke him the Beetlejuice way: @pluralistic @pluralistic @pluralistic

Look Cory, better account portability FEP!

Dan 🌻Oct 15

@jonny this is awesome!!

Along with load all posts, you're single handedly (well, not really, ty other fedi devs, but you get what I mean hopefully) improving fedi usability by like 200%

Fedi developer of the year award winner right here

Ty for doing this!

@danvolchek thank you. all i know how to do is collections, enumerate they collections, eat hot collections, and collection.

Jenniferplusplus Oct 15

@jonny just as a note for the future: content addressed resources would be technically tricky, but would also have a lot of desirable technical benefits. The biggest problem with implementing that is social, though. You need to know how to route reports about a given resource. Right now, that's inferred from the host name, and that likely breaks down under content addressing.

It's something I'd be more than happy to talk through some time

@jenniferplusplus yes of course! we are gonna start experimenting with the distributed trust idea that we talked about nearly a year ago by next week when we finalize two last sciop things and move to federation work. definitely not doing away with instances, just shifting their role, and i think doing reporting when an actor can be a part of several instances will be a good and hard puzzle

David Chisnall (*Now with 50% more sarcasm!*)Oct 15

@jonny Thank you for this but also for the list of things that are out of scope in the proposal. I initially skimmed it and had a load of 'how do you do X' things, and it was great seeing a list of 'X is out of scope, it's handled in this other FEP' things, which avoided my need to ask stupid questions.

@david_chisnall i appreciate you saying so, and also if you did have any remaining questions i promise you i will not think they are stupid.

David Chisnall (*Now with 50% more sarcasm!*)Oct 15

@jonny I have only a very vague understanding of ActivityPub, but I found this easy to follow and all of my questions were either answered or explicitly not answered as being either future work or covered on other specs. It was a great read, thank you!

@david_chisnall i was not expecting to hear that this document was an enjoyable read, that is high praise and thank you for it.

David Chisnall (*Now with 50% more sarcasm!*)Oct 15

@jonny

I might be a statistical outlier in the kinds of things I enjoy reading.

@david_chisnall the world is richer for its outliers :)

django Oct 15

@jonny looking forward to Digging into this!

Ben Pate 🤘🏻Oct 15

@jonny

Very cool. We absolutely need a full-fidelity move that’s supported by as many apps as possible, and you FEP looks very well written.

I’ll catch up on your work and share my thought on codeberg :)

Could you compare/contrast this with the LOLA work going on under the W3C. It’s a solid plan, but it can be hard to decipher in places: https://swicg.github.io/activitypub-data-portability/lola

I think it’s super important for us to define migration well, once. Competing specs could easily derail the whole thing.

LOLA Portability for ActivityPub (0.2)

☝🏻@lisarue

@benpate @jonny
These are very consistent in the stuff that both do - it is clear that ActivityPub is already *almost* able to do content portability just by serving up content the way it does. We have a task force within the SocialWebCG working on account portability - jonny would you join a call if we schedule one soon?

jonny (good kind)Oct 16

@lisarue
@benpate
Absolutely! DM me here or email [email protected]

ValorZard Oct 18

@jonny @lisarue @benpate
Excited to see where this is all leading!!!
Hopefully this will let fediverse servers act in a more PDS style manner a la bluesky

jonny (good kind)Oct 18

@valorzard
@lisarue @benpate
Yep, there's never been a technical reason activitypub can't do mobile identity, even truly delocalized identity, its just a matter of nudging the existing implementations to a point where you can make a jump

https://codeberg.org/fediverse/fep/src/branch/main/fep/ef61/fep-ef61.md

ValorZard Oct 18

@jonny @lisarue @benpate
Do you think @silverpill portable objects proposal can compliment what your up to?

fep/fep/ef61/fep-ef61.md at main

fep - Fediverse Enhancement Proposals

Codeberg.org

jonny (good kind)Oct 18

@valorzard
@lisarue @benpate @silverpill
It is indeed the medium term goal

ValorZard Oct 18

@jonny @lisarue @benpate @silverpill
If you need any help implementing any of this I’d be down.
Idk wtf I’d do, but I’m down
(Either portable objects or migrations)

jonny (good kind)Oct 18

@valorzard
Once we get a nod from masto devs on whether they'd accept it I can sketch out a few sections, and in the meantime I'm gonna get rolling on glitch

Ben Pate 🤘🏻Oct 20

@lisarue @jonny

If you get something scheduled, I'll do my best to attend :)

jonny (good kind)Oct 20

@benpate @lisarue thanks for reminding me to respond to that email 🫣

julian Oct 20

@[email protected] @[email protected] please keep in mind that simplicity is key. I think that's Jonny's FEP's strength.

The more complicated it is, the less buy-in, and you need all the buy-in you can get!

jonny (good kind)Oct 21

@julian
@lisarue
Thank you for saying so! I tried to keep it as simple as I could and take care to carve out space for low-resource authors to implement gradually, though when all the necessary considerations were on the table it feels more complicated than I would like. Glad it still feels simple to others :)

CoolBlenderKitten Oct 15

@jonny
Well, that sounds amazing! Great work, everybody and thank you all

Andy Piper Oct 16

@jonny how does this relate to the LOLA proposal from the SWICG? That has been in progress for quite some time and I’d like to see a comparison / how this integrates or handles the cases and issues identified.

jonny (good kind)Oct 16

@andypiper
I heard of the LOLA proposal after drafting this, and still havent found it but now knowing it comes from SWICG should be enough to. I'll take a look. I would defer to protocol experts, but lemme take a look

jonny (good kind)Oct 16

@andypiper
Brief read: looks nearly compatible, some things are more explicitly spec'd here, some more explicit there, this relies more on target instance and attempts to minimize role/burden on source instance, theirs is more source instance centric, this relies more on cryptographic proofs and exclusively AP, theirs seems to be oauth and webfinger. But it seems like reconciling shouldn't be hard

flere-imsaho 🇺🇦Oct 16

@jonny it's here. i gave it a quick read, and i don't think these are mutually exclusive.

LOLA Portability for ActivityPub (0.2)

@jonny Thank you for putting in so much work to put this together. I hadn't realised you were the person who had shared it in the Github thread, when I saw it there. It looks really good, and it's great to see that it's compatible with earlier FEPs like https://codeberg.org/fediverse/fep/src/branch/main/fep/ef61/fep-ef61.md, too.

It's good that this includes some ability for the new/target instance to decide whether it wants to accept content—target instance admins were the main source of pushback I got when building my (very basic) ...

fep/fep/ef61/fep-ef61.md at main

fep - Fediverse Enhancement Proposals

Codeberg.org

@jonny ...migration tool a few years back. Most of the objections were unfounded and based more on fear than reality.

Drawing on that experience, I wonder if by letting target admins decide whether or not to accept content separately from accepting a the overall migration, or which parts of a user's content they will or won't accept (allowing them to strip out DMs, for example, or media), is going to cause problems. Primarily that will be a moderation burden because admins fear they're going...

@jonny ... to be deluged with problem content when the reality is it's going to be a pretty small part of it. I'm concerned the consequence of this is going to be friction against the whole feature, and a knee-jerk reaction that says "this just shouldn't happen at all" rather than working with you to find a way to make it work.

It could also create a situation where a user can think their content will be migrated but afterwards discover that only half of it is there. ...

@jonny .... Really, if a target instance isn't going to accept a user's content that should be flagged at the start, not once the account has been migrated but the moderator is later considering whether to accept the content as well. (Am I misunderstanding the proposed sequence of events?)

In general, while I appreciate there needs to be a way to prevent abuse and accommodate admin's concerns about problem content, admins are a big part of why this feature doesn't exist, and I'm concerned ...

@jonny ... giving them the ability to accept only part of people's content and/or make that decision after the user has otherwise migrated is going to empower them to basically torpedo what you're trying to do, by causing bad experiences for users (whether or not intentionally), demanding the ability to moderate and then not being willing to handle the moderation traffic, or just hindering people's ability to move their content. ...

@jonny ... There's also privacy concerns. Few admins will read posts or people's DMs unless something is flagged as abuse, but it looks like what's proposed here puts all of people's content in front of admins and asks them to review it, which is much more invasive than typical use of the instances otherwise.

If instances accept people signing up and posting without review, I question the grounds on which they want to be able to review all of someone's posting history before allowing migration.