Thanks to Tim for the neologism. Once you start looking, there are lots of possibilities for systems which make information amenable to a passing glance: everywhere can be mediated, if you have the right tools. Most work on ubiquitous and intimite comptuing has focussed on new technology to deliver pervasive networking and sensor-rich environments. I’m more interested in what we can do here and now, with a bit of thought, using cheapish, commoditised componemts and the occasional bit of hacking. Of which, Tim’s Homeland Security Alert Status glanceware desktop is a fine example. What could be more important than being able to determine the current likelihood of having a dirty nuke parked just around the corner, without having to waste precious seconds mousing though to a government website for confirmation. Glanceware gives you more time to run.
On a more practical level, I’ve been using the fine web-scraping service at myRSS to build glanceware for market intelligence. RSS is fine, but the majority of corporate news sites (at least here in the UK) haven’t woken up to it yet, and are unlikely to for some time. To solve the problem without having to write my own web-scraping system, I’m using myRSS, which enables anyone create, for free, publically available RSS feeds from ‘any website’. Well, from any which uses HTML in a predictable manner, anyway. For a fee, you can increase the frequency of update from daily to hourly, or get a hand-customised filter built for any of your sites. Fee-paying users also get credited as ‘sponsors’ of the channels they create. Assuming myRSS catches on, early adopters will do well out of the visibility of their sponsorship. Ku24 is sponsoring hourly RSS feeds from the most important UK media news sites (currently including mad.co.uk, Brand Republic and BroadcastNews). I’m recouping the money by offering newsfeeds plus a newsticker (powered by wticker) as a Ku24 product. The feeds I’m sponsoring are available to anyone, but I can afford to cover the cost from the value-added ticker product, which feels like a reasonable way to make money from open-sourced information and technology.
a while ago a client approached me with the need to do some analysis of a “pay per click” directory portal (ie that site charges its listed clients every time a web surfur clicks through the search portal) … and for whatever reason the portal site had all the figures of “cost per click” up there for the world to read. through some judicious coding i managed to dump the entire site a number of times and pull their entire database of “cost per click”. it was a lovely thing. a multi-GB database of names and costs with cost changes over time.
anyway, what i gleaned from that experience is that the need for things like myRSS will always be there … standards take a while to propagate and by the time they do something better has come along. meanwhile, you need translators. in fact, in many cases, translators may be BETTER. there are reasons why, lets say, a web-scourer-to-rss-feeder is more flexible than the rss standard itself. sites have their own proprietary systems than may need bots to navigate databases. if and when they create an rss feed, chances are it may not be what *you* want anyway. there are definitely reasons to create custom rss feeds with this kind of tool … or, god forbid, move to something a little more flexible at both ends! give me some push and pull.
i suspect the core of glanceware is really some form of filtering intelligence for all the aggregated feeds coming its way. Homeland Alert goes RED and meanwhile your net-connection-lost-alert triggers. what the hell is your desktop going to do? flash?
“i suspect the core of glanceware is really some form of filtering intelligence for all the aggregated feeds coming its way. Homeland Alert goes RED and meanwhile your net-connection-lost-alert triggers. what the hell is your desktop going to do? flash?”
i expect in that event the question is academic. its all probably melting into the floor at that point 😉
its actually a kind of ‘degrees of freedom’ issue. how many parameters do you want to represent? how multivalued are they? what senses do you have available? how sensitive are they…rememebr that research from i think the 70s about representation of multivariate data as faces showing different expressions? didn’t they find that you can represent at least 7 tristate variables in a glancable iconic face?
Hang on, my google sense is tingling:
Chernoff, H. (1973). Using faces to represent points in k-dimensional space graphically. Journal of American Statistical Association, 68, 361-368.
Chernoff, H., & Rizvi, M. H. (1975). Effect on classification error or random permutations of features in representing multivariate data by faces. Journal of American Statistical Association, 70, 548-554.
There’s some basic stuff here, and a lovely diagram i can’t include here coz i’m in comment mode. sigh. maybe worth a proper posting. anyway, I gues the trick is finding glanceable forms which we’re good at parsing very very quickly (which argues for them being primarily visual/iconographic or musical…maybe)
heading for a Cronenbugian pulsing GM mutant mouse thing that starts to quiver when the aggregate feeds frighten it. that “personal amanuensis” is following us again.
just remember, it all has to work on your phone as well.
or at least something nearby needs to know it’s nearby and morph into a twitching mousething when relevant.