In 1987 Apple produced the Knowledge Navigator video, which presented in scenario form the kind of user interface that they thought knowledge workers would be using twenty or more years in the future. Over the last week there's been considerable interest raised by Jon Udell's revisiting of that video, and his review of how accurate its projections were.
My feeling is that Udell's assessment is sometimes a bit generous in his assessment of what progress has been achieved. Comparing Knowledge Navigator with Bruce Tognazzini's sister video, Starfire — which projected only ten years from 1994 to 2004 — shows how many of the projections of the past turn out to have been over-optimistic.
Jon Udell has found an online version of the six minute Knowledge Navigator video (14.9 MB, QuickTime required), and checks off which projections do and don't look realistic from the vantage point of 2003. Briefly, he puts ticks in the boxes for wireless and multimedia, half a tick (goal nearly achieved) for resource discovery/search, quarter of a tick for information visualisation, and a big cross for natural language communication.
I agree with 80% or more of Udell's review, but I think my perspective on the Knowledge Navigator scenario is different — at three levels. Firstly I am less sanguine about what has been, or can be, achieved in the visualisation area. Udell gives an account of how the visualisation of data shown in the video might be achieved using XML, schemas and queries. I wouldn't challenge this (I don't know enough technically to do so, even if I thought a challenge was warranted!). What I think this account misses is the intelligence required at the user interface level to determine that the animated map visualisation is the optimum one required for the user's purposes in this instance. Research into intelligent adaptive user interfaces hasn't produced anything nearly this robust (and some us have had a hunch for some time that it never will).
Visualisation practices are still in the dark ages, and gratuitous unintelligent use of technology is partly to blame for this. Think, for example, of all the annoying animations on the web, and compare them with Jakob Nielsen's 1995 guidelines for multimedia on the web, which incidentally include an homage to the use of visualisation in Knowledge Navigator. (And for an amusing parody of technology-inspired bad visualisation, see Peter Norvig's PowerPoint version of the Gettysburg Address.)
Secondly, Udell doesn't mention the way the video presents a 'virtual butler' on the user interface (except insofar as the he refers to natural language and the 'butler' agent communicates in natural language). The cause of interface agents has been set back in popular awareness by Microsoft's paperclip (here's one of many spoofs it's inspired; here's another!). And I'm not mourning it, for I wrote in 1992 that "it may not be helpful to have an agent try to 'second guess' a user". In fact it was Udell's omission of any criticism of the interface agent, that led me to resuscitate my old paper on this concept.
Finally, Udell concludes with the observation that "The obstacles that keep data and voice/video networks apart seem more political and economic than technical." And that really nails where my perspective is different from his, because I'm at least as interested in the political, economic and social factors as I am in technical factors and their product-centric focus. I'd recommend that anyone developing scenarios for future information and communications technologies invest in reading The Social Life of Information by John Seely Brown and Paul Duguid before they finish their storyboarding. Their chapter on Agents and Angels is a much more authoritative debunking of autonomous anthropomorphic user interfaces than mine!Posted by David Jennings in section(s) Human-Computer Interaction, Reviews on 2 November 02003 | TrackBack