David Gelernter: The Time of our (Streaming) Lives

David Gelernter is seriously smart.

A professor of computer science at Yale, he’s credited with everything from anchor thought in parallel computing, to helping Bill Joy advance the thinking that led to Java. His “mirror worlds” idea advances the belief that our technology can be a better partner in helping us manage information.

Key to the mirror worlds perspective is managing our “lifestream,” the river of information that flows through our days. Data in a lifestream is dynamic, continually being changed, and in turn changes other information.

In the 70’s and 80’s, Gelernter presaged a number of innovations that are only now becoming real. He posited the need for cloud computing by saying that our mirror worlds would save our information in some remote computing space. His team believed that a 3-dimensional view of data was critical. “We… built 3D displays from the beginning,” he said. “We preferred not to think of a screen as a flat space with stuff written on it, but as a viewport onto 3 dimensional cyberspace.” And he did all of this when technology was far less pervasive in our lives. “Now, you can’t get along in this country without looking at a computer every day,” said Gelernter.

(Information) Islands in the Stream

In the analog world, we “space-arrange” data: We put it in folders, folders go in file drawers, drawers go in cabinets, cabinets go into rooms. That folder metaphor has penetrated every major computing device in use today. But in a lifestream, data isn’t organized by where we want it to be (since where we receive it or want to put it usually doesn’t matter in the digital world): It’s organized by when – “Time being the skeleton key to experience,” Gelernter said. “Time is usually a good heuristic to remembering a lot of stuff… It’s the backbone of storytelling – the conversation, being in itself, a version of a diary or narrative.”

Lifestreams can be organized into substreams, and those substreams can be filtered – and even discarded. “There’s a sense of tracking, removing things that I don’t want… There’s nothing we can’t throw in the garbage, technologically,” he said. “And I hope to be making that decision early and often.”

Discussion turned to the value of privacy, and whether complete privacy creates negative results. “Anonymity is a problem,” Gelernter said, implying that anonymity reduces the incentives to be complete and truthful. “Probably, anonymity is part of a superficial culture of junk information that’s too willing not to distinguish… well-sourced from unsourced information. Do we take the time to ensure that the information we’re throwing at each other is true?”

But lifestreams aren't just theory: We already have a variety of technologies that provide lifestream-like services. Take, for example, Twitter. “Is Twitter a version of Lifestream? Absolutely,” said Gelernter. “Twitter’s a great model… [And] real-time is tremendously important. But it’s also important not to be locked into a model where peoples’ conversations are boxed into trivialities.”

Forgettaboutit

So is keeping every piece of information that’s flowed through our lives a good thing? A question from the audience mentioned Nietsche’s principle of “active forgetting,” which maintains that we deeply need the ability to forget. Gelernter’s dry response: “I’m not sure that Nietsche would be a lifestream user.”

-gB

Gary A. Bolles, CEO, Xigi Inc., gary [at] xigi.biz

Comments

[TahoeBlue]

The topic and theme presented by Mr. Gelernter resonate with other information-centric aspects of the interaction of technology and the economy with recent cultural trends. The rapid upshot of various kinds of time-streaming via the internet and mobile devices constitutes a dramatic departure from modes of interaction and communication which have heretofore been possible.

However, Gelernter's metaphor of "time-indexed" as opposed to "space-arranged" as a new paradigm for organizing data has some shortcomings, in my opinion. I would suggest that *data* is still organized by humans in a spatial, categorical fashion, even if the files, folders, and rooms are electronic and virtual. This is particularly true if the data is viewed as being something which has informational value and which may be desired to be located or randomly accessed on demand.

Although archives of conversations -- be they email, blogs, forums, or social network updates -- contain a time-ordered record of streaming experiences, information, opinions, facts, and thoughts, it seems to me that the recent advent of time-streaming is a completely new development and dimension when it comes to human interaction with each other and the web.

Streaming in and of itself is not the most natural way to organize data, if only because, if left untended or groomed, streams upon streams will defy the temporal mode of the mind to remember, locate, and recall specific points in the stream -- however much we may recall experiencing a particular event in a past stream.

There is also the issue of time itself. It takes time to simply monitor and interact with a variety of streaming interfaces -- to the extreme that a growing percentage of the younger generation are becoming glued to the hip with their realtime mobile updates of their current experience in the world as they move through it. At some point, the "cup runneth over", and, as the cursor of time moves on, previous stream interactions fade into the fuzzy fabric of historical minutiae.

I would offer up two "use cases" which might be illustrative. First, consider the (now outdated and unheard of) practice where people would record radio broadcasts on reel-to-reel tape. A simplex radio broadcast is of course an early form of streaming :) Those who would record radio programs would need to put the tape in a carton and label the carton with not just the date of the recording but also -- in order to have any chance of 'indexing the stream archive' at a later time -- the pseudo-random sequence of music titles recorded from the broadcast. This is an early example of recording metadata needed to locate information in a stream.

One of the issues with current streaming interfaces, as discussed elsewhere in these Techonomy sessions, is the *absence* of metadata which accompanies unstructured streams. Another issue is the simple fact that, to replay or access stream archives, time itself is required. The more time-based streams that we record and archive, the less likely that we will ever be able to replay or re-experience them because of the unbounded accumulation of former time and the unsuitability of "time acceleration" to revisit them given the limitations of current (and finite) time.

A second example, which is closer to the point that Gelernter makes, is the use of "labels" (keyword attributes) which users of gmail can assign to tag emails, as well as the grouping by gmail of emails sharing the same subject into "conversations" (threads). This is a good example of 'organizing' stream/conversation content for later use. Surprisingly, gmail does not allow the sorting of emails, which is a gratuitous gripe that I will throw in while on the subject.

[Iaincollins]

I think David Gelernter is likely to be more widely regarded as a greedy and self serving patent squatter trying to squeeze money out of others who have actually built commercial successful implementations of patents that he has absolutely no comparable implementation of (either existing or imminent).

As a get rich quick scheme, trying to get your mits on some "eff you" money by exploiting a software patent of dubious integrity has proven quite effective in the past (especially, but not exclusively, in the US) but it tends to leave people with the lasting impression that you're a douchebag.