A short while ago, in a post entitled Wanting a Multidimensional Lifestream, I pondered the topic of how to add more context to my RSS lifestream so that it might move beyond a mere data stream and start to become more like an actual human, nuanced identity… a digital me. The example I gave in that post dealt with ABBA, Last.fm and irony… namely the inability of the service to parse the context of the data for something like irony.

In imagining this further, and in the comments to the original post, concepts such as weighting and tagging offered new possibilities, but still didn’t quite get to where I wanted to be. There’s no doubt that the spirit of what I’m seeking is in those solutions, but they still reek of compromise. I wanted Last.fm (or any service that I permit to gather data from me) to really know me… to understand when I’m being ironic… and to consider that when it formulates its representation of me.

And then today, while listening to The Who’s Tommy and reading Clifford Pickover’s Sex, Drugs, Einstein, and Elves, it hit me… perhaps that is just too much to ask of any one service. Not only that, but would I even want that one service to have that kind of hold control over the “me” that lives online? Probably not. No, what I think I really want is something, someway, to assimilate the data (my data) that exists in the variety of services I use… Twitter, Flickr, Delicious, Netflix, Amazon, whatever. It seems to me that the data pooled in these services might provide a sense of “me” in a way that any one of them couldn’t on it’s own.

Could this data, if synthesized in a certain way, provide fertile ground for the development of accurate representation by an engineered system? Could there be some sort of Kalmanesque process that would in fact take this data and distill it into an output approaching my digital essence? Might it be an evolution of Feedburner? Or a highly intelligent and self-organizing descendent of feed splicing? Could there ever be a run of numbers that take a set of data and from it approximate the experience of “knowing mike” online? And if this app were built, what might it look like? Wouldn’t I just be putting all my eggs in one basket… exactly what I didn’t want to do in the first place?

Perhaps… but what if it wasn’t an application, or a service, or a branded site that did the work, but rather an open process? A freely distributed and openly licensed library of code or an algorithm (an engine) that could be used by mad developers the world over so that anyone with the brains and the motivation could elicit real, live, virtual, digital “strange loops” from the informational patterns of predefined data feeds. Sort of like Brad Fitzpatrick’s Social Graph, but instead of representing who I know, it represents who I am.

I have no idea what the output of the representation engine might look like… I just know what it would feel like. It would feel like meeting me in person… like shaking my hand and looking into my eyes… roughly. It would offer interaction, sometimes live with me, sometimes with a digital “me” fabricated from data. It would have a URI and the ability to answer and ask questions on my behalf. It would be a feedback loop continuously evolving and changing as I do.

Granted this all sounds a bit like science fiction and there are some disturbing possibilities embedded in the very idea, but it is still something to ponder. While I’ll probably unpack the concept in future posts and follow it down all the disturbing rabbit holes, I’ll leave you today with a taste of the first possible side effect that came to me during the writing of this post (which, by the way, started with a germ of an idea and then unfolded all by itself into what you’re reading right now).

Could this imagined representation engine (if it ever was developed) not only be a way to create better digital representations of ourselves, could it also be used as a way to create entirely new digi-people? Imagine the output resulting from feeding my Last.fm feed and your Netflix queue and somebody else’s Amazon wishlist into the (fictional) representation engine. Imagine a hybrid of all three of us that could live on the web at a specific URI and grow and evolve as its representational parts grow. Imagine asking it questions and getting answers. Imagine it asking questions and understanding answers and rolling that information into itself. Imagine it living, looping… digitally.

Imagine all of that and then ask yourself this question: Could all of this web 2.0 data just floating around out there in RSS and Atom feeds be the seeds of a viable artificial intelligence?