When Justin Rattner takes the keynote stage and starts riffing about something slightly old hat, you know the annual Intel Developer Forum techfest in San Francisco is a wrap. Rattner is Intel’s Chief Technology Officer, and it’s his job to close IDF every September with a spot of futurology. And it’s usually rather wobbly.
A few years ago, for example, he based his talk on Linden Lab’s Second Life, when even the mainstream media had tired of the game as a shorthand for all manner of digitally driven social trends, real or perceived. Last year he went with stereoscopic 3D, just as the hype was dying down.
But what about 2010? When Rattner started talking about context-aware computing, it certainly felt like the same old back-to-the-future experience. His first demo duly involved a very familiar GPS-powered mashup from US travel guide outfit Fodor’s. The underwhelming new application does predictable things, such as recommending restaurants and other points of interest based on your location. OK, it also purports to learn your preferences in terms of cuisine and pricing, and it has a fairly funky auto-blogging feature that combines photos with auto-generated commentary on your whereabouts and activities. But, frankly, it’s nothing we haven’t seen before and even lacks a few obvious features, including user-generated content.
From there, however, things gradually became more interesting. Intel’s in-house context computing specialist Lama Nachman told us how an array of accelerometers, light sensors and GPS could be used to build up a detailed picture of behaviour, location and more. It was largely familiar stuff, and most of what wasn’t fell into the category of the theoretical, but Rattner and co did have a few examples of how you can achieve a lot from a little by combining powerful processors with cheap sensors.
One such nearly-commercialised example is a TV remote control that can identify the user by the manner in which they wield it. From there, it throws up a menu of personalised content and data – your favourite TV shows, social networking updates and perhaps your email inbox. Exactly how long or how many hand movements such a device would take to recognise the user wasn’t mooted, but it doesn’t matter, because neither restaurant recommendations nor personalised content are terribly new or profound.
Instead, the real power of contextual computing appears to emerge when you build up a more detailed picture of location, behaviour and state of mind over time. In the latter case, I’m talking about inferred state of mind. Rattner did have something to say about mind-reading computers, but more on that in moment.
Getting back on message, context aggregation is the buzz concept and personal finance is a handy analogy. For any individual, it’s easy enough to be aware of a single financial transaction. Things get a lot more complicated, however, when it comes to keeping track of spending patterns over extended time frames.
The same applies to all kinds of behaviours, and context aggregation promises to make sense of it all and then use it to enrich your life and empower it. In other words, future computing devices won’t simply react based on what you’re doing in the moment, but also what you’ve done in the past.
A simple example might be a PC that presents your favourite news website and email inbox automatically when you fire it up in the morning, or an in-car infotainment system that has your route pre-mapped when you jump in based on an entry in your calendar.
But it also means that, collectively, your phone, PC, car and TV will be building up a broader picture of how you live your life. There will be no more kidding yourself over how many hours of work you’ve done, how much exercise you do or the quality time you spend with the kids. In the future, it will all be logged in cold, hard numbers. Fascinating and frightening in equal measure.
As for mind-reading computers, Rattner duly delivered the ultimate end-game in contextual computing, courtesy of an update on research into just that by Carnegie Mellon University. Current technology doesn’t extend beyond a system that can distinguish whether a human is thinking of one of two words, but the implications are still mind-boggling.
Personally, I find the whole brain-reading thing less interesting than the more subtle possibilities for contextual computing. When the brain-readers come along, they’ll no doubt blow our minds. In the mean time, computers will be inferring all sorts of weird, wonderful and most of all unpredictable things about our lives. Maybe Rattner’s crystal ball isn’t as foggy as I thought.