DISQUS

CyndyA · 1 year ago

LMK when they get all the IM clients working with each other and then we can discuss how they'll get this mess worked out what with all the broadcasting that would have be done across multiple systems. How are you going to maintain the user relationships? Twitter can't manage to do that with one system. You can't use a centralized database across systems using a subscription architecture, so how does anyone propose this? Winer says "open source" and "federated" like it's the solution to all the world's ills, but has offered exactly zero solutions for implementation.

robdiana · 1 year ago

Cyndy
IM is a slightly different issue because all of the IM clients started completely closed. Eventually they opened up bit by bit. Now we have clients that talk to most if not all of them. There is also a standard that some follow called Jabber/XMPP for IM. Twitter actually uses part of XMPP for their realtime update API. Each user needs to have a "home" server in this type of case as well. So if I am an Identica user, that is my home and where all of my relationships are stored.

As I tried to state in the post, a federated system like email, is a very good solution. However, these things take time to build, so we still have to wait a bit.

jcarreira · 1 year ago

You think you want a distributed Twitter, but you don’t: http://blog.babelnote.com/2008/07/you-think-you...

Anthony Farrior · 1 year ago

It's only a matter of time when you think of how well the multi-IM client Trillian handles everyone.Just like you,I think a microblogging client based on XMPP will mash everybody together pretty soon. Posty has already agreed to possibly place the clients it supports in one big window...

charlieanzman · 1 year ago

Rob - I don't these this happening other than the selective importing and commenting we're seeing on Friendfeed (and any similar services that may follow). Presumably most of these services were set-up to either monetize and / or integrate commercially (or be sold). They will have the option of limiting any interaction as we've seen in the past with AIM and other chat-type services. Clients (like Twhirl) will continue to be developed, with services such as Friendfeed acting as the catch all. Any further interaction is highly unlikely. Great to see you posting to Louis' blog!

Marco · 1 year ago

Charlie -

Definitely see your point on the track record of limiting interaction but I see it taking towards the direction Rob is talking about.

There are good reasons why instant messaging services started allowing users to communicate with people using competing applications.

It was all well and good when AOL was the one dominating and innovating - they were hoarding a massive user base so the game was theirs to control. But as soon as the ways to connect to the internet proliferated and new interfaces and tools were developed that provided internet users with an incentive to jump AOL's walls they had little choice.

As we have moved through that process and the market has grown, providing users a myriad of choices for virtually each type of web activity, I believe we have all developed an expectation that the services we use play nice with others.

I don't think there is a user out there that wouldn't love to see the kind of federated approach rob is talking about - Twitter and its cousins don't have to move in that direction but that is not going to change the market's demands.

At some point somebody will figure out how to do it and it will be AOL all over again for the pioneers...

robdiana · 1 year ago

The main issue I see is that people will have friends on several services. When you get people like Scoble moving to Identica as well as using Twitter, someone will try to get things to work together. Companies want to support the Scoble's of the world, because a good word from him drives a significant amount of traffic. He may accidentally have started someone down that path already. The main reason I think it will happen is that there are a lot of early adopters already asking for it. That and the open APIs make this a possibility even in the next year.

Jim McCusker · 1 year ago

Federation is simply the beginning of the solution for a micro-blogging service. You'll also need a dynamic load-balancing system that is capable of spreading the load of requests for content acroll multiple servers. For example, a popular microblogger (eg. Scoble) might be replicated across all servers, while a mid-level microblogger may only be distributed across a few. The average microblogger may only be on two or three servers depending on the redundancy requirements.

robdiana · 1 year ago

Yeah, you are probably right about the heavy users. I was trying not to go too deep into it because there are too many infrastructure issues as well. At that point I am almost designing the architecture. Maybe that is a good topic for another blog post :)

Marco · 1 year ago

I am obviously not qualified to speak to the technical aspects but I think this kind of approach is exactly the kind of solution microblogging (macromessaging?) needs to move one step closer to fully integrating itself into the mainstream.

I looked back over Jason's post from earlier this month of Babelnote and while some of the concerns he raises seem to make sense I don't have anywhere near the technological background to know whether or not they are truly valid.

Truth is, from my perspective, the how is irrelevant - its the why that matters.
There is a fundamental difference between the kind of communication that the Twitters and FriendFeeds offer and email, instant messaging, texting and chat rooms. If there wasn't no one would be so excited about them - they would take one look at them and say "oh, that's just like (insert item from previous list)" and walk away.

Point is - the demand for the function isn't going to go away, its going to increase.

The kind of solution you are suggesting would allow the Twitter's and Indenti.ca's out there to focus on providing UIs that will meet specific needs and attract more users instead of spending all their time trying to build a structure capable of supporting everyone's everything.

Thanks to you and Winer and Jason and all of the other people out there who are taking the time to put their tech skills to use to try and figure out a solution.

So long as you guys keep thinking forward and throwing out ideas for the solution someone, somewhere will figure it out.

robdiana · 1 year ago

I read the babelnote blog entry you link to and he has some valid points. We do not have much spam because you need to explicitly follow people. Opening up the architecture with federation could have the same effect as letting spam email become a huge problem. The other reason we do not have a lot of spam is because all of this stuff is new. Email was around for years before spam became a problem. His concern is more about introducing problems into a currently "troubled" system (Twitter), and if you read to the end he also has another motive (building another system).

Jim McCusker · 1 year ago

Not really. If you hold to the subscription model of Twitter spammers will have no way in. That design decision by the folks at Twitter simply kills spammers. For that matter, spammers trying to follow thousands of users wouldn't gain any benefit since very few people would follow them. In the design that I envision the people being followed would get the main share of the infrastructure resources while average users would be peppered across fewer systems.

elliottng · 1 year ago

Seems like Summize was a virtual federated system for search and discovery of tweets. and people were using it for discovering @ replies too. Maybe all we need is for the posting services like Ping.fm to cache tweets and expose pre-submitted tweets to Summize, and then Summize can search across these caches. Totally seems possible, and why wouldn't Twitter want to do this with Summize? BTW, I still miss Summize classic branding. It seemed "smarter" with the old branding than with the new Twitter color scheme that just evokes Fail Whale to me!

charlieanzman · 1 year ago

Marco - We never really know but if any of them are looking to make money , the only one I can remember was ICQ when hey sold the company.

klecu · 1 year ago

For subscriptions across servers, there would need to be steps like 1) authenticate requesting server (is it trusted? is it blocked?), 2) transfer list of subscriptions (who on the requesting server subscribed to whom on the supplying server), 3) verify that each author on the supplying server has not blocked the requesting server, 4) transfer posts only from the authors on the subscription list. Ideally, each server would remember what other servers "subscribe" to each author, and push that data in real time or near-real time as it is posted.

For example, if I post to laconica server 1, and someone on laconica server 2 is subscribed to me, server 1 would know that server 2 needs my posts and would push a copy to server 2, which would store/cache and display to my subscribers on that server. And as bandwidth and storage marginal costs decrease, this will become increasingly feasible.

This kind of feels to me like a subscription model email system now, where you write the message without really controlling the distribution list.

robdiana · 1 year ago

Absolutely agree with your list of steps. I tried to stay at a higher level as the several services and API needed would be several blog posts and I can only write one at a time :)

I always thought that Twitter was a public IM platform. Subscription email is another decent way to look at it. I also think people are too concerned about the problems that might be added instead of the benefits. Am I right that federation is the right way to go? Maybe, let's see if someone builds it.

gregorylent · 1 year ago

ha, thought it was, can MICROBES talk to each other, and got excited, science discovering a biological basis for social media

Marco · 1 year ago

lol

svartling · 1 year ago

I think we all should go back to email. That way we could all discuss from one centralized place, and not have to bother with different accounts and services. One mail at many different places. Just joking. Federated microblogging is a great idea.

Jesse Stay · 1 year ago

Sorry - I forgot to link here when I mentioned Federation - will add the link now. Identi.ca is actually intended to be Federated. It will be interesting to see how successful they are at it. As Tim O'Reilly has been talking about recently, "Open" and "Web 2.0" can, and should go together - things like Federated micro-blogging platforms are ways to do this.

nmalaguti · 1 year ago

this sounds like you're trying to re-invent e-mail only instead you're sending small messages...

Q dub · 1 year ago

Nailed it! I'm surprised no one else has mentioned email yet, the quintessential federated technology.

And who makes a profit on email?