A few years back Steve Yost, Ben Hammersley, David Weinberger, Danny Ayers,  Shelly Powers,  Jason Shellen, Jay F, Mark Carey, Jon Lebkowsky and myself (and others) participated in a mail list/conversation about the notion of a standard for representing conversations.

We called it ThreadsML and the web site and archive of the mail list are still around.

I seem to remember that a message board that David Weinberger used throughout the nineties had “gone away” and he was quite pissed about losing all those archived ‘conversations‘.

So I found myself on this list, and we knocked around various principles, requirements, usage scenarios and specifications - which when built - would enable all sorts of new kinds of interaction and ‘fluid conversations‘ via the web.

Well look what happened!

It happened WITHOUT ThreadsML!

But like any good new idea, it’s ownership and control will be tussled and fought over by various early stage vendors who are first to the game.  Sound like Twitter, FriendFeed, CoComment and Disqus to you?

ThreadsML is exactly what is needed now.  And lots of inter-connected servers running ThreadsML, with a DNS-like backbone striping and copying these conversations as they happen in real-time.

But before I get too far - let me stand back and explain what ThreadsML is - in an abstract and get some historical perspective as well.  I’ll start off with some background:

- Way back when conversations would start as mail list threads.  Endless going back and forth with email messages, conversations were difficult to follow - but spirited and lively anyway!  One would create a ‘group’ in Outlook to keep track of all the folks on the thread and we’d ‘manually’ maintain the conversation.

- Then blog comments came along and conversation morphed, just as they were morphing with IM, chat rooms and IRC back channels.  We also got a feature called ‘trackback - and blog search engines could also inform us if anyone had picked up on a ‘meme’ you posted and started an unofficial conversation.  A blog conversation.

- One could also follow a specific meme or keyword or phrase - and watch it birth itself throughout the blogosphere.  Our decentralized world was evolving.

-  Several years passed in this state until Twitter came along and track.  Now we could Twit what we were doing and one could follow specific people’s Twits which naturally led to all sorts of conversations, both long and short.

- FriendFeed put an interesting twist on all this - enabling us to comment on ANY conversation that flowed through FriendFeed.  So we’re now watching Robert Scoble and others perfect this new way of conversing - in a decentralized manner.

- Now the point comes up “who owns these conversations?“  Call them fluid - call them liquid - they’re happening completely free-for-all in the open marketplace - just like standards are SUPPOSED to evolve.

- Steve Gillmor has pointed out that the TinyURL facility has also given us independence of vendors and the ability to cannoize our conversations.  His ecosystem and nervous system metaphors may confuse some - but this is a complex chess game we’re playing here - and one which needs all sorts of ’savants’ preaching to their interested choirs.

- So in that spirit I’d like to add to this discussion the notion that a standard that can represent a conversation - a persistent re-entrant conversation - can help unite all of these memes flying around the decentralized meshed web.  A standard is possible and in fact is staring us in the face.

ThreadsML exists as a web site that David Weinberger has been paying for - for almost 5 years now.  We created a spec for ThreadsML (back then) and we’ve been waiting for this day to arrive - convinced that it’s need would become apparent when - well what is happening today happens.  And it DID happen!

The need for an open standard to unite disparate kinds of conversations, threads, IM sessions, comments, tracks, mail lists, etc.  These conversations get started on all sorts of service, in many different ways.

The following is my interpretation of what ThreadsML is - apologies to anyone if I get it wrong.

Technically a ThreadsML server would be a threaded object store - which would enable anyone to contribute into any thread node and attach media to any node.  A DNS-like sub-system would unite LOTS of these servers together and all sorts of glue code would convert proprietary threaded data structures and protocols into ThreadsML and back.

So ThreadsML is more than just some schema.  Its a system that unites conversations and stores them as ‘memes‘.

David Recordon and the DiSO project seem to be working on something similar to ThreadsML.

The original ideas are still there - waiting for anyone to build them.

We clearly need a neutral place:

- which is owned and controlled by no one party (NEA)

- which can house these conversations

- convert them into other forms

- enable re-entrant contributions - later

- stop spam

- enable one to attach media to any node of the conversation

- and keep it all interactive and two-way!

That’s what the dream of ThreadsML was all about.

Here’s a historical artifact I created - back in 2003 - about ThreadsML:

Little did I realize that ‘near term’ meant “wait five years”.

This entry was posted on Sunday, June 22nd, 2008 at 8:45 am and is filed under . You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.