Hacker News Reader

Yeah this is a really interesting question. It's the point Moxie brings up when people ask why signal isn't federated or otherwise decentralized. You basically have to solve multi-stakeholder collaboration. You want projects/companies to have the flexibility to add features, but you also don't want subtle incompatibilities to plague the system like "I can't see the picture on your tweet because you use the MediaEmbeds schema and I used the EmbeddedMedias schema." This is the topic I get nerd-sniped about a lot.

There are a couple of ideas I want to try. One is machine-readable schemas that get published on the network. That's kind of conceptually similar to what smart contracts represent; there's a program that essentially "lives on the network" and which people can build against, reference by hash/pubkey, etc. It's a coordination primitive. I like how intuitive that feels.

Another idea is schema negotiation, essentially metadata that says "this post uses MediaEmbeds, here's what needs to happen if you don't support it."

A third idea is using graphs of fact-triples like RDF, which gives you a lot of flexibility but I worry may not solve coordination well enough.

And a fourth idea is to use lenses like Ink and Switch talked about with Project Cambria.

I think this is a really interesting topic, I hope we find some good answers.

Talking about IRC in particular, the reason I'm on Discord is:

(a) UX far beyond anything I've seen with IRC

(b) super easy for people to create new Discord "servers"

(c) super easy to discover or link to Discord servers

Talking about social microblogging/commenting (aka Twitter), the reasons I joined then later exited the ActivityPub "fediverse":

(a) Twitter's discovery is massively better

(b) My "identity" was not portable between instances (that's different from user on Network A interacting with user on Network B), also no easy way to link multiple instance identities

If bluesky can solve all these identity/discovery/onboarding UX issues in the context of an open protocol, it'll be a massive step forward.

> My "identity" was not portable between instances (that's different from user on Network A interacting with user on Network B), also no easy way to link multiple instance identities

That is by design. It's a federated system! Just like you don't need multiple phone numbers to talk with people in different networks, or different email addresses depending on the recipient, you don't need different identities to communicate with different servers.

Unless you want to keep your identities separate, there is no point in having multiple accounts.

> Unless you want to keep your identities separate, there is no point in having multiple accounts.

Until an instance goes down either temporarily or permanently and it was your only account. Then you'll wish you'd had a few extras. How many email addresses do you have? From how many independent providers?

Accounts aren't portable but there's no inherent reason that accounts need to map 1:1 to identities. In fact they already don't, a single identity already commonly has multiple accounts. The software just isn't aware of this fact and has no standardized way to communicate it.

> Until an instance goes down either temporarily or permanently and it was your only account.

That's a technical problem, but not a UX or system design one. If a server goes down, my backup account is only going to allow me to keep participating, but it is not going to save the conversations I had on the lost one.

> Accounts aren't portable but there's no inherent reason that accounts need to map 1:1 to identities.

Again, you are talking about identity portability, which is not the same as "creating one different identity for each different server".

In any case, I totally understand the use case and I tried to get the Pleroma devs to focus on this idea as well: https://mastodon.communick.com/@raphael/106825107786781891

> my backup account is only going to allow me to keep participating, but it is not going to save the conversations I had on the lost one.

The same issue exists for other services including email, irc, and matrix. The obvious simple and immediate solution is to retain copies of things you care about locally instead of expecting a particular service to stick around forever. With email this used to be standard practice before the rise of webmail. Even now most desktop clients sync to and retain a copy of everything on the local device.

> > Accounts aren't portable but there's no inherent reason that accounts need to map 1:1 to identities.

> Again, you are talking about identity portability, which is not the same as "creating one different identity for each different server".

It seems like it is though? Servers need a way to handle authorization so they presumably need accounts. With a simple protocol, accounts map 1:1 to identities as far as the software and protocol are concerned. If you make a new account then you also make a new "identity" from the perspective of the software. The obvious way to solve that is to map any number of accounts to one (or possibly multiple) identities, where an account is the thing the server is concerned with and an identity is something else such as a private key.

There's an additional question about the mechanisms for account creation and authentication. Maybe you don't want to maintain an email and password based login for a long list of accounts that are each tied to the same backing identity. There are multiple approaches available there as well, from oauth providers to public key based authentication possibly including things like cross signing.

> The same issue exists for other services including email, irc, and matrix (...) expecting a particular service to stick around forever.

It doesn't have to be like this. This is the point where identity portability comes into play. If the identity belongs to the user and they carry to different servers, the user is independent from the provider.

In the case of email/irc/matrix, as long as you own the domain name, you can change the service provider whenever you want.

You seem to be describing hosting your own service, changing the underlying hosting provider, and tying your identity to ICANN DNS.

There are two separate things here, identity portability and data portability. If your identity is portable (say a key pair) but the service hosting the data goes down and you don't have a backup or there's no way to import the data to a new provider then you've got a problem. Similarly, if the service goes down and you do have a backup of your data and a way to import it to a new provider but your identity isn't portable (perhaps it was user@domain and you don't own the domain) you also have a problem in that case but it's a different problem.

And neither data nor identity is the same as account, at least the way I was using the words.

> You seem to be describing hosting your own service

No, it still gets to be hosted by someone else. And it is not just because I am using a custom domain that it requires a separate server. "Shared hosting" has been a thing for web and email forever, and there is nothing stopping* a matrix or an activitypub server to process requests for different users on different domains.

> tying your identity to ICANN DNS.

Not necessarily. If you see the linked conversation I had with Alex, it was about using Ethereum ENS. It could also be something based on OpenID. The important thing is that the actorid in activitypub can be anything that can be authenticated by the server, it doesn't have to be only the internal username.

* there is a current limitation on matrix synapse (where the homeserver can serve only one domain), I don't see anything on the protocol that forbids the implementation of a multi-tenant server. Perhaps it is not a top priority because such a server would incentivize big hosting companies to offer it as a service and possibly re-centralize it.

Most federated services DO heavily focus on identity as username@server though, and don't have any real account sync.

This is half-true. For plenty of services it is easy to add "custom domain" support, and then your identity is your domain.

It's not about having multiple accounts, it's about the fact that federation is a bit meaningless when only a megacorp can run a server.

Using a small server is just asking to lose everything when it goes away, and small things come and go all the time. Porting one account to a new server(Or better yet not tightly binding it in the first place) is essential.

> Or better yet not tightly binding it in the first place

Yeah, I agree. As I repeated elsewhere on this thread: having your own domain is the most straightforward way to guarantee that you can port your identity.

But this is a separate problem from this idea of multiple "accounts". This is mostly from Mastodon's weird culture where people make this association between themselves and the in-groups from the instances. I wrote about that on my blog

[0] https://raphael.lullis.net/federations-and-identity/

> Twitter's discovery is massively better

Do you mean recommendation algorithms, or just the number of users ie network effects?

> My "identity" was not portable between instances

Your identity isn't portable between email servers either. Or between HN and Reddit. Isn't that part of what Keybase is about?

I can imagine a couple ways of achieving this but it runs into a number of thorny issues regarding namespaces. Regardless of the approach taken truly portable identities would require lots of refactoring of lots of software in a FOSS ecosystem with fairly limited developer resources available.

Merely linking identities should be much easier but still each piece of software has to support it. Simpler just to link to your other accounts in your bio the same as on any other website.

> Isn't that part of what Keybase is about?

was

Fair point. But OTOH Twitter isn’t really innovating anymore either. Arguably the “move fast and break things” phase is over. Now we’re heading into the “monetize or commoditize” phase.

The wireless telecom industry does this on the regular. It requires tons of collaboration, tons of design & specification, tons of testing & verification, etc. etc. With all that, I suspect the reason why Bespoke Walled Garden 5G Wireless™ doesn't become a regular occurrence is because the barrier to entry in both buying spectrum and rolling out infrastructure is so incredibly high.

I have no idea what if any lessons there are to learn from that, but it is at least an example of where an entire multi-vendor, multi-layer industry migrates forward more akin to the "protocol" model mostly all together.

> because firms, that is to say top-down organizations can push features quickly

I'm not convinced that's the case. I think it's more an issue of resource investment. Firms are capable of funding things. Twitter at a low level barely has more "features" than email if you stop and list off what you actually need in order to do what it does. But it has reams of UX stuff built on top of the base functionality, recommendation algorithms, moderation tools, lots of servers, technicians to maintain those servers, etc, etc.

FOSS stuff is continually at a resource disadvantage in most domains.

Decentralization can push features very fast, and support all kinds of extensions and variants that spread and become standards.

ActivityPub and all the rest are either not decentralized, or they are blockchainful with all the associated disadvantages. Namely, it doesn't work on isolated LANs, throwing away one of the coolest killer apps of P2P. Federated protocols often tie your identity to some random hobby scale project server.

What stops P2P from adding feaures is just the fact that it's a lot of work and nobody seems all that interested, and most P2P projects are held back by a very narrow defined scope, wheras companies like Facebook go for the kitchen sink.

The other thing that slows dev is multiple unrelated apps. If you control everything(Which is totally possible in FOSS, if one group is the de facto leader that most people go with), you can update as fast as you want, as long as your protocols have primitives for arbitrary message types.

If Tox or Jami had a million dollars, or SyncThing were interested in expanding scope, or a few dozen people wanted to make a social network on BitTorrent they could do just fine, but everyone wants to build yet another Etherium token, or doesn't have the resources to support all platforms and continually add features.

All of these problems go away the moment that we stop accepting non-free software.

If we as consumers start saying "no, even if your product is better I will not sacrifice my long-term freedoms for the short-term advantage and convenience of your product", companies would still innovate, but they would be playing in a more leveled field.

How about free software creators start innovating first, instead of insisting people should simply be willing to put up with poor quality because "freedom?"

This way of thinking is the reason Gimp will never be a viable competitor to Photoshop. Meanwhile, Blender is actually used by professionals because its quality makes it worth the effort.

Who said anything about "putting up with poor quality"?

> This way of thinking is the reason Gimp will never be a viable competitor to Photoshop.

Have you contributed to Gimp in any way? How do you expect a barely funded project to compete with a multi-billion dollar company?

> Who said anything about "putting up with poor quality"?

you did...

>> If we as consumers start saying "no, even if your product is better I will not sacrifice my long-term freedoms for the short-term advantage and convenience of your product (...)"

>Have you contributed to Gimp in any way?

No. I don't have the time or the specific domain knowledge, and as a tool I don't find it useful enough to use. Have you? I didn't see a name on the contributors' page[0] that resembles your username.

>How do you expect a barely funded project to compete with a multi-billion dollar company?

I don't know. I do know that isn't an argument that's going to convince anyone using Photoshop to switch over to Gimp. Maybe one of the millionaires that hang out here can do it.

[0]https://www.gimp.org/about/authors.html

No, I did not. Making a priority to have free software in my stack is not the same as accepting anything "just because it is free".

> I do know that isn't an argument that's going to convince anyone using Photoshop to switch over to Gimp.

Gimp vs Photoshop is a somewhat bad example to the topic at hand (closed systems taking over free software that implement open protocols), so if you don't mind maybe we can go back to those?

In this case, I am a backer of Mastodon. I also contributed to Pleroma and to the developers of Conversations (XMPP client). Nothing much (~100€/year), but I feel a lot better by contributing to them that I would ever feel by paying for Slack or Discord.

The point is, it doesn't take a millionaire to do that. If companies took 5-10% of their IT budget to contribute to free alternatives of the closed software they use, perhaps they would quickly lessen their dependency on the closed software and free their resources.

FOSS was innovating a lot, and then it took a major hit when people decided code and features were evil and they didn't want to risk any extra attack surface, and then all the P2P types seemingly mostly moved to blockchain dev.

Plus, I suppose most P2P was piracy driven and piracy is no longer an everyday conversation topic, or something a lot of people would even think of.

That might also have something to do with why the privacy conversation moved away from things like neutrality and right to encrypt, to focusing almost entirely on extreme hate for telemetry.