Gordon Moore, co-founder of Intel whose silicon has been at the heart of the last forty plus years of computing, came up with Moore’s Law in an article in Electronics Magazine in April 1965.
Moore predicted, with what turned out to be uncanny accuracy, that the number of ever shrinking electronic transistors you could place on an integrated circuit would roughly double every two years. And, upholding Moore’s prediction, the CPUs at the heart of the personal computer revolution went on to become progressively smaller – and therefore cheaper – and more powerful right in line with that expectation. Indeed, they say Moore’s Law is good for at least another ten years.
Earlier this month Intel Labs shared their prediction that the size and power requirements of meaningful compute would approach zero in around seven years from now. What they meant is that by the end of the decade, processor production methods will have advanced sufficiently that they’ll be capable of fabricating chips powerful enough to deliver a meaningful, everyday amount of processing power delivered from an integrated circuit that will be so tiny that it might as well be invisible, or at least be capable of launching a new generation of ubiquitous, out-of-sight computing in everyday objects.
That’s an amazing notion and conjures incredible images of what the server farms of 2025 might look like compared with today’s cavernous, football field sized server farm installations the likes of which project services like Google, Amazon and even Xero onto our screens. While you might imagine that the server farms of today could shrink in size to something akin to a 1960’s mainframe that could fit inside a regular sized room – which is certainly a possibility – I suspect what is more likely to happen is that server farms will remain enormous (and might actually get bigger) and will just get more much more powerful as increasing numbers of CPU cores are crammed into them. Just as desktop PCs didn’t shrink all that much but got progressively more powerful since the first IBM PC appeared in 1981.
Regardless of how large or small server farms end up getting, the point is largely moot unless you have the misfortune of living next-door to one since they are in all practical senses invisible to us, the users.
Hold that train of thought.
Lose It and Use It
A few weeks ago I ordered up a Raspberry Pi, an ultra low-cost and almost matchbox-sized single board computer that was designed to be as affordable as possible in order to inspire a new generation of children to re-discover the wonders of computing. It was also designed to re-introduce the practice of programming back into UK schools where there has been a progressive dumbing down of the school computing curriculum over the last twenty years.
Computing in UK schools wasn’t always so dumb. As a kid in the 1980’s I was lucky to have my own imagination ensnared by the Pi’s spiritual ancestor, the BBC Computer. I spent endless hours with my own BBC in my parents attic and with those in the computing lab at my high school, teaching myself BASIC programming and discovering the fundamental wonder of computing. This innocuous piece of tech sits at the epicentre of my career in technology and the passion for computing it ignited in me as a kid still burns some thirty years later.
Actually as an aside, as I’m writing this I can see that the benefit of hindsight is funny sometimes. As the smart-ass in my high-school computing class, my teacher set me a much tougher assignment than the rest of the class for my O-Level end of year project, and one that would see me writing a complex set of code that would fully replicate the functionality of a banking system, complete with multiple bank accounts, basic security, debits, credits, transfers, an ATM interface, back-end reporting; the lot. I also gave it an awesome UI. That sounds so familiar. If only she’d asked me to replicate NASA’s rocket guidance system…
So, the Raspberry Pi has some pretty big shoes to fill and while it’s very small, it’s not quite invisible. However, the other day I was chatting with a fellow Xero about what I planned to do with my Pi when it arrived, and without thinking I concluded that for convenience – rather than going to the bother of connecting it to its own screen, keyboard and mouse – I’ll attach it to my home WiFi network and just access it virtually with some simple VNC software through my Mac, or any other device I choose.
I won’t even need to remember where I put it – I’ll just snatch time with it whenever and wherever I choose; as a device it’ll just be floating around in the ether of my home. In other words as practically invisible as server farms are today – albeit in a domestic setting.
Ubiquitous computing isn’t just a fancy prediction from Intel nor is it just a convenient way to access cute little devices around the home. We already drive around in car shaped computers and fly long distances in long tubular shaped computers both of which are invisible as computers, at least in an abstract sense.
We’re entering an era – in fact we’re probably already well over the threshold – where the way we engage with computing and carry out tasks is changing from being a deliberate, sit-at-your-desk-and-press-buttons affair to a passive or even ambient experience where tasks and jobs waft momentarily in front of us, and remote agents do more leg work.
We already have the first of these invisible agents today. Consider Xero silently automating repeat invoice production and delivery at 12am on the first day of every month, or any concoction of rules and triggers and tasks you can imagine with cool new web services like IFTTT.com. * Be sure to check the update at the end of this post for changes to made to IFTTT subsequent to this post.
IFTTT (as in ‘lift’, and an acronym for If That Then This) caught my attention recently and in spite of its somewhat enigmatic frontage is very interesting. You can use it to build simple agents to carry out trivial automated tasks like sending you a personalized SMS weather report every morning, or anytime your local weather forecast changes beyond preset parameters. But it got really interesting last week when IFTTT added the capability to automatically create and populate Google Spreadsheets with information scraped from services like Twitter or your Google Mail inbox.
It’s Like Fishing for Nerds
So, as an example; with Xerocon landing in London this month I created an agent to monitor and grab every mention of #xerocon on Twitter and to automatically append the details of each mention into a Google Spreadsheet document. Honestly, my heart actually fluttered the first time I checked my spreadsheet and found that it had already captured its first few mentions – it’s like fishing for the nerd generation.
Once Xerocon London is over, I’ll pull together all the collected data my agent has fished up for me and post up a blog with up the resulting word cloud, or something. Although what I use this data for isn’t really the point, because what’s actually cooler is having an invisible agent building spreadsheet databases for me containing whatever I choose from the vast expanse of the web.
Say hello to do-it-yourself Big Data.
Whether it’s invisible server farms, hobbyist computing devices you can freely misplace, invisible accounting processes with a predilection for the color turquoise or invisible agents filling out invisible spreadsheets, the most interesting thing about this new era is as much about the new kinds of ‘stuff’ we’ll be able to do, as it is about the way our expectations for software and devices are changing and evolving. And as easy as it is for Intel’s scientists to predict when computing will become effectively invisible, it’s actually inversely as difficult to predict what we’ll do with invisible or ubiquitous computing when it appears.
Or, rather, doesn’t.
You only need to look to the launch of the iPhone 5 this month to see how difficult it is to predict this stuff over the medium term. Throughout the course of an hour-long launch presentation, the telephonic characteristics of the iPhone garnered about fifteen seconds of airtime. Yet we still call it a phone.
We think of Xero an accounting system but I can’t wait to see what we’re doing with it in ten years time, invisible or not.
UPDATE : Twenty-four hours after this post was first published it was announced that IFTTT was closing down the ability to build Twitter triggers on the their platform following what one can only assume was an unpleasant conversation with Twitter, who lately seem to have adopted the opposite of Google’s old ‘Don’t be evil.’ mantra.
I’m jumping to huge, fanciful conclusions about the fact that the Twitter triggers disappear from IFTTT on September 27, the day of Xerocon in London, thus permitting me to at least track #xerocon Twitter mentions on the day as I described above. It seems the Xero blog is more influential than we realised…