The current internet/web model of using many social networking sites with many different addresses and notification styles is pretty bad, and it's getting worse. We've got IM, Email, SMS, Atom/RSS, Web sites, and custom clients all competing for ways of identifying people, subscribing to messages, publishing/reading/deleting messages, replying to messages, security on messages/subscriptions/status. Any time there's a new site, each user needs to get an ID then figure out who's already there, who they are interested, invite people to join, etc.
I think there's a huge amount of re-invention of enterprise messaging that's happening on the web, and will continue to happen. Enterprise messaging regularly deals with subscribing to topics and individuals, whether subscriptions are one-way or bi-directional, broadcast and point to point (p2p) publication, security on subscribers/publishers and attributes like availability, message history, message storage during subscriber offline/not available, delivery when subscriber available, intermediaries, topic/user discovery, and much more. I think I'll write up an article on a taxonomy or classification for many of the various possibilities,
In the end, there's eventually going to be two big changes: 1) we'll have a real notification infrastructure aka real-time web; 2) we'll need intermediaries and all the related standards needed for interoperability. Jabber/XMPP can do some of this, but there's plenty more needed.
Twitter + IM
As a thought experiment, let's combine Twitter and an IM network together from a publisher and a subscriber perspective. The idea is that a publisher could use either, and a subscriber could use either.
Sending a twitter direct message is pretty analogous to an IM and plays to the traditional IM messaging. But Twitter is much more about broadcast messages to many subscribers. And that's where the rub comes in, is how do the different views map from one to the other.
Twitter Status aka Publish
How should Tweets be mapped to IM? Should they become an IM status change or an IM to each subscriber? The core of twitter style is knowing that tens, hundreds, thousands of people can casually read tweets, whereas IM core is knowing that the person you are sending messages to is online and is a 1 to 1 communication, including user x is typing status updates. The hard part is knowing whether a tweet should cause all those IM screens to pop up or not.
Follow aka Subscribe
How should twitter follow requests be mapped to IM? Should a twitter follow request be the equivalent of an Add Contact? The core of twitter style is that followers/following are separate subscriptions, whereas IM core is that the subscribe is a bi-directional subscription. The IM subscription easily maps to 2 follows in twitter, the hard part is how to map the uni-directional follow messages in twitter into the IM bi-directional subscription.
Intermediaries
Intermediaries are the obvious way to bridge twitter and IM, and in fact bridge any pub-sub networks. Twitter is already set up for intermediaries to read and write as it has an excellent API. But most of the successful IM networks are spotty. Trillian does the integration between networks, but amazingly it's the only one and regulary has problems as the IM networks attempt to preserve their proprietary APIs and client lock-in. And does Trillian provide a full-featured API?
The core of whether there can be a pub-sub web is whether the various proprietary networks will open up so that intermediaries can read/write from their network.
Standards
Even assuming that networks will open up so that publishers and subscribers have the choice of messaging network(s) for their messages, all the integrations will be one to one and very difficult until standards emerge. One of the most important standards is identifying users. They are all over the map currently, from email address on proprietary networks (x@hotmail.com) to short names (like twitter) to URIs. OpenId has the possibility of become at least the mapping point between user identification, if not quite the OpenID nirvana of becoming the "One True Nym". Then many other standards need to emerge if the cost of network bridging is to be reduced. XMPP can be a big part of this, but I think that it's history of focusing on IM makes it difficult to map to other systems, as I've shown with Twitter.
Leave a comment