Tent.io brings an important question. Can distributed social work as well as email?
Imagine a scenario where you sign up on service A and your friend signs up on service B and you're both able to message each other without worrying about things like SSO or making profiles on those social networks.

Also, can ADN act in such a manner? Can social networks built on ADN allow frictionless cross posting? Will that be a model that'll just work?

The main data issue is still not solved by ADN. ADN has control of it. You can export in and then it's totally useless since you can't take it anywhere.

I suppose if ADN was using Tent (they may be I'm lazy) then you could take it to another server, if you can even find one.

So who is using Tent.io now?

Tent.io is still in the beginning stages. I'm not really talking about data exporting. That's beyond the point of this discussion. I want to know what you think about email style social networks - data can flow anywhere. A copy stays on each server and anyone can setup a server without it being an open source software.

I think it is quite simple to make realtime update work as *well* as email in a federated environment. I guess for me, that isn't sufficient for interesting use of the medium. :)

Realtime update is interesting because in a lot of ways it works better than email. And I think it would be very difficult to maintain those properties in a true open federation.

I agree that realtime updates are something that are very implementation specific. That's why Facebook holds that knowledge so dear to it's heart and that's why I suggested that ADN can better implement it because all social networks based on ADN will be following the same protocol...

So, naming seems like a problem with distributed services. Surely @alanhogan is nicer than alanhogan@myprovider.domain. However, this is also a benefit: A flat namespace eventually reaches practical exhaustion of good names. And users do not need to be displayed as provider-username pairs; just like email, once you “friend” someone, you can usually just see them as a name and avatar. “Alan Hogan.”

I'm just not sure that an email style social network can succeed. People want realtime so I think that to succeed in any meaningful fashion there would have to be some realtime component.

If I was building on App.net I'd still be worried that they hold all the keys. There was a day when Twitter would go on about how open their API was, and we can see how that's turning out.

I can just see my not to tech savvy wife trying to find a friend@provider and totally failing. She even hates facebook but stays around because that's where her friends are.

Alan, you're right. Gmail changed to show the correct name of the email sender a long time ago, even if it was from a non-Gmail account. So it's not difficult to implement a username@provider system and then show the right name, the correct profile pic and a link to the corresponding social network (all this data can be pulled by one server from the other).

Also, ADN would be a flat network below a whole lot of other networks, so getting the right usernames wouldn't be a problem.

Can you also imagine it working, Curtis?

What if she used a popular client that allowed her to, like Facebook, search by name. It would search all the registered providers, and attempt a graph (FOAFs first) + popularity (Real Lady Gaga first) sort.

Curtis, working with friend@provider would be a pain unless we implement a protocol like (mailto:myemail@me.com) [so we have msg:friend@provider] and people just use these links everywhere on their websites.

IMO, the big problem is not the hierarchical naming system. The naming system really need not even be visible to the user. (i.e. when I say 'Dave', it knows to send a message to my Dave).

The big problems are:

(1) If services like app.net accept messages from outside users to inside users and vice versa then suddenly they've removed the incentive for people to be users and that is app.net's current business model.

(2) A true federation bring back the specter of spammers.

John,
You're right about the business model angle. That's why I think we should consider two separate angles - app.net based social networks that only allow app.net based users to cross post between social networks.
And separately a new protocol (perhaps tent.io) that allows the building of email style social networks. These are separate case studies.

For me to believe that tent.io would work, I would have to be convinced that there is a business that can be made out of running tent.io servers. If we are to rely on volunteers donating a box in their basement as a server, we run into all the same issues that DHTs have.

Curtis is right about twitter's API problems that might extend to app.net one day. This website (branch) is made by people who started twitter. So this website too can become evil any time. That's why it makes sense to have a standard that's implementable.
John, your point about business is correct. Which makes me think - whats the profit for google to run Gmail for free? Adverts right? Same goes for all other email providers, with only a few of them asking for money to avoid the ads.

@Jonathon

(1) Not sure why we are necessarily focused on ADN, but for-profit federation members like ADN itself could compete by having open & generous API terms, good UX, good anti-spam, and simply brand awareness. Or, maybe part of the federation model is that each provider agrees to require at least a $1/yr fee.

(2) Providers could have a meta-spamminess score. Providers could expose their own estimation/confidence of spamminess for each user, perhaps, and users could ignore messages from non-friends with high spamminess scores (or low confidences, if they are OK with false positives).

Yes I could also see it working if it was a totally 'flat' naming system.

I would love a standard for everyone to use, I just fail to see a solid business incentive for it. It totally makes sense for up and comers since they have no user base and it would give them access to the users of the established players, but how is that good for an established player.

At some point all these 'open' discussions have to come around to putting food on the table for people. There are lots of high minded stances I'd drop if it meant my kid could eat.

Ideally we would have an open solution that allowed people to make a decent living and allowed users to do what they want as far as connecting and using it. Just seems like to many 'ideal' requirements.

The reason I'm thinking about ADN is that I was contemplating the likelihood that ADN would itself join a federation. This also goes to Curtis' point about ADN currently holding all the keys. I somewhat believe the incentives story that Dalton has told. But I would be happier if it were also true that there was a real federation where nobody had all the keys.

The way to look at federation is to compare with the ISP/Internet federation in the US. The edge ISPs (cable/DSL/cell providers) make lots of money. This is mostly because they are part of regional monopolies or duopolies. The core ISPs tend to have very thin margins.

I believe a true federation would *at best* have federates with a business model with thin margins like core ISPs.

Actually, one cool thing with federations (don't know if it is in tent.io) is the possibility of partitioning trust. Each federate is a trust root and acts as an authority for names in its domains. So you can actually avoid the Diginotar-style problems which have caused problems with the SSL CAs.

Thinking about business models, if we talk about Facebook, the business model is adverts and App.net is the yearly fees. If FB and App.net agree on a protocol to interconnect, they allow better interactions where both parts of the interactions are owned by both parties. Large networks are all about how they can use user data to earn money. Thus, a connecting protocol doesn't need to depend on the kind of earning each network is doing. This about peering in ISPs. One ISP has a lot of users and the other has a lot of bandwidth, they peer to use the other's resources. Similarly, App.net connects to FB because it's users pay and demand it. FB connects because increased interactions means more data about the users' likes and dislikes.

I can see one side paying the other for interconnect. It is hard for me to see both sides allowing interconnect for free or allowing random smaller federates to interconnect for free. Gotta go for now. Interesting conversation.

Jonathon, I just read up on Diginotar. Indeed, the entire network's trustworthiness will come from the amount of spam coming from it's users. That will allow users to request blocking for those networks. The more the blocks, the more the network falls from the popularity index. That system works well with IPv4/v6 addresses and can work with social networks too.

Alternatively, I wouldn't have to worry about signing up for each network or SSO passing my data to new, yet-untrusted networks because they only get my initial details and all of my interactions. The current market value is of big data. That means they'll get exactly what they want without getting me as a user.

Also, ISP peering is more of a business transaction and slightly harder to break than the connections that social networks make nowadays solely based on trust. IFTTT-Twitter debacle shows that we need to ensure some legitimacy of interconnects. I'm sure ISPs also play each other when deciding on peering, but in the end, where business is involved, smart minds do come forward and make the right decisions.

ADN is the prime player to get in on this right now. That's where we should be applying pressure. If it get's enough traction we will see FB and Twitter buy in, but only if they have little other choice.

I don't know about that. Last time, I checked, they FB and Twitter even do OpenID, because they prefer a dedicated Log In With Facebook. They use some standards like OAuth, but only when it benefits them.

Indeed, FB and Twitter support OpenID. But the flaw with it is that they pull of the user data and then complete a signup at their end because user numbers are an important indicator of platform popularity. If we talk about interconnects, that number will go down for every network but the number of messages between them will increase. The question is whether social networks want more users or more user data. Currently, the answer is obvious. But can that trend change?

Sorry, I misspoke. I just double-checked, and they don't support OpenID as a provider (what I meant), just OAuth. I know Facebook used to support OpenID as a relying party (doesn't look like they still do), but that just illustrates their perspective on which to support.

Matt, like I said, it's just a step for them to push people towards a full signup. This doesn't help me as I'd really prefer to have One email provider (Gmail), One Cloud provider (Dropbox) and finally, One Social provider (ADN for me).

I don't mind that Facebook stores those conversations that I have with my friends on their network as long as a copy is also stored on ADN, which I trust more because I pay for it.

OpenID is another great example of an open system that just hasn't got traction. 37Signals dropped support for it because it was a hassle and most of their users were not using it.

productblogarchive.37signals.com

Yeah, I think OpenID and OAuth have both been victims of standards creep. The standard just offers too many possibilities for implementers.

The 37Signals examples shows that the correct implementation should be to allow users to quickly get to the action screen (post, new project, whatever) instead of just getting their full names and asking to signup for this service.

In that sense, do you think App.net should try to setup such a service with as many providers and social networks as it can, by faith or by cash?

It would make sense for app.net to be an OpenID provider, despite the standard's flaws (it already uses OAuth). I don't think it should be a priority, though.

Well, instead of trying to conform to older standards that people have already shunned because of flaws, I believe ADN should first implement cross posting in the Social networks that're coming up on it's own platform (Alpha, Friendle) and then move to add other social networks to the fray. PoC is important to show to the world that it works...

Cross-posting is easy. That's not the same as an identity standard. It would certainly be interesting to pioneer a new, better ID standard, but I don't think it's an app.net priority.

App.net is just working on the basics right now. Cross posting, ID management and interOp aren't high importance features for at least another year. When they do come, what matters is that App.net should have the monetary resources to make deals with other social networks to give special features to their members. Features such as cross-posting without login, spam management etc.