The durability of conversation

Just come across my mate R’s writeup of our night in a pub a fortnight ago. It made me realise quite how much territory it is possible to gallop through in an evening’s chat with an enthusiastic conversation partner, especially when aided by a couple of pints of cider.

It also made me realise quite how ephemeral conversation normally is. Wilde’s quip that he put only his talent into his works, reserving his genius for his life, is a sobering reminder that if all you generate is chit-chat, then that’s all anyone will be able to remember you by. I suppose of course that Dorothy Parker provides the counterexample; the hundred of poems on which her literary career was constructed are now all but forgotten in favour of ten to twenty wisecracks (with the possible exception of One Perfect Rose, which gets regular enough outings).

This little bubble of thought reminded me that I have had in mind for some time a system for recording conversations in a more than linear fashion – because, after all, conversation is rarely as linear as we think. Instead you’d hook up some voice recognition software to a natural language parser which (being really far better than anything I’m aware of at the moment) could produce an abstract tree of the conversation and its junction points, perhaps being able to tag the junctions (the bits you’ll come back to later) by means of a search through the speaker’s web presence or indeed her other conversations… and could then project the whole thing in glorious 3D, allowing conversationalists not only to have a conversation, but to create one like a work of art, painted in front of them in real time. You could zip back frequently to things you’d meant to mention, and produce a densely branched structure… you could colour each speaker’s threads, and see which of you was more like to leap into monologue. Of course there are potential pitfalls. What if you discover you really do have the most dreadful habit of interrupting something significant with something trivial? Well, I suppose it’s better to learn it that not, although you might want to try out a few trial conversations before going public with your art.

I love the idea of being able to treat conversation as a durable art one could revisit or reopen. I love the idea of all the things you could do with a durable conversation. Cross-reference one to another. Invite people to join in later. Have time-lapsed conversations (a little like a Google Wave, but with something more like exclusive locking); have conversations with complete strangers. Have a public archive of good conversations and continue them yourselves, getting together with a few friends to tag a branch onto a chat about Nixon in China or public policy.

Then there’s all the aggregate information you could extract. What topic clusters are most likely to send you off on a tangent? How often do you lose track of your line of argument? On what topics are you most likely to underparticipate? Is there any correlation between how much you enjoyed a conversation and any other metric you can invent? What are the most common topics under discussion when conversations ABEND? How do they vary by country or geographic region?

And following on from that of course the privacy and security challenges that a sort of publicly available archive of conversations — which are, after all, some of our most private moments — would pose. I’m not suggesting that you’d have a conversation plotter running in the wee small hours to catch pillow talk and automatically uploading it all to Wikipedia; but even assuming we were limiting ourselves to willingly produced material, how do you manage the material? Do the participants have ownership? Is the addition of a new branch additive or transformative? Could branches be secured so that only certain people could follow them (which is similar to the “gevulot” concept which manages shared memory in Hannu Rajaniemi’s excellent The Quantum Thief), or so that only certain people could contribute….

I could yammer about this all night, but I’m jetlagged enough to be heading off to bed, so I shan’t. It’s worth dropping in, though, that the other obvious-ish message here is that all that stuff around privacy and managing sharing isn’t something uniquely posed by this new way of tracking conversation; it’s just that there are a set of existing rules and possibilities which we’re so used to we don’t see them (or call them basic social skills). As McLuhan had it, the environment is invisible. Good night!