New Computers | September 5, 2006
Today a home computer is a laptop sitting on your desk. Soon it may be part of your wall, your coffee table and your bedside table
They are statements that have gone down in computing history. In 1943, IBM chairman Thomas Watson predicted a “world market for maybe five computers.” in 1977, digital equipment corp.’s CEO Ken Olsen told the world “there’s no reason why anyone would want a computer in their home.”
Now, to be fair, they were talking about computers as they existed then: massive, expensive and just about as user unfriendly as it’s possible to get. But the pronouncements of these two knowledgeable men underline just how tough it is to predict the future of computing. People keep trying, nonetheless, because doing so is fun and interesting.
It was in that spirit that hp’s chief technology officer Phil McKinney took to the stage in San Francisco. The company’s mobility summit was all about how best to use today’s technology, but McKinney also took some time to gaze into the future.
His predictions were based on the implementation of existing or very near-future technology, rather than on Star Trek-like warp drives and transporters. In fact, McKinney’s vision is one of de-evolution, a “back to the future” scenario in which homes are sprinkled with what amounts to dumb terminals. As digital equipment’s Olsen knew all too well, dumb terminals were simple input/output devices — keyboards and terminals — hooked to a central computer. This idea later evolved into the thin-client concept, which employs simpler computing devices that get most of their muscle from central machines.
The future, McKinney said, is not one of individual super-charged systems with oodles of giga-whatsits and ram-ekins, but a series of simple machines that allow us to input commands and receive responses. The access devices will link wirelessly to a central computer somewhere, but the user experience will be very simple and interactive. And that makes sense, according to Toronto-based Bill Buxton, who now splits his time between Cambridge and Redmond as a principal researcher at Microsoft Research.
One of Buxton’s long-standing party tricks is to challenge an audience to draw a computer. “of course, what they draw is a screen and a keyboard,” Buxton said. “That’s not a computer at all, those are input and output devices.”
Indeed, breaking free of entrenched ideas — even the QwerTY keyboard, for example — requires rethinking everything including the word computer, said david hill, a designer at computer-maker Lenovo. hill worked on IBM’s famed butterfly keyboard a decade ago.
“I remember seeing a new processor with 25mhz speed and being told we’d never need anything faster,” hill said. “and that 8mm tape would solve all our storage problems. But what we’re doing now is far beyond spreadsheets. it is such a part of our lives, but ‘computer’ suggests a machine that calculates data, something out of math class, and it is so much more than that.”
“I don’t know what a computer or even a phone is anymore,” Buxton said. “My mobile isn’t really a phone, it’s a radio. But that realization is a good thing because it forces us to look at the larger questions such as how we use it."
Dick Tracy time
In the vision HP McKinney presented, the lynchpin of this virtual interface is a watch-like device. It acts as a mobile phone but also manages the interconnectivity with keyboards and screens distributed through a home or office environment, and links to the larger world through cellular networks and the internet. Ultimately, the watch links to a processor box which could be anywhere — in the basement or a server farm somewhere and accessed via subscription.
“The idea is to disintegrate features and functions,” mcKinney said, “to have the tech adapt to you. it’s not about integrating more features and functions into devices, but actually to disintegrate those features and functions into devices that are more focused on a single function.” The result is a series of smart devices tailored to be the right tools for specific jobs, McKinney said.
These could include a coffee table whose surface becomes a flat monitor and a wireless keyboard that can be stored in a slot when not in use; a placemat-sized tablet device; a monitor and keyboard made of flexible material that allows them to be rolled up and put away; and a wall-mounted LCD TV which can be “borrowed” by a user on demand, all products McKinney showcased during his presentation as examples of the possibilities.
It’s feasible, Buxton said, because the cost of LCD screens is falling and the ability to “borrow” screens wirelessly via a mobile computer presents a good business model. “Why not borrow the screen at a bus shelteror atamall?”hesaid.“Someone will work out the details — whether it’s subscription-based or uses advertising.”
Microsoft’s take
It sounds like wonderful science fiction, but at its June Tech Ed 2006 conference in Boston, Microsoft discussed similar concepts, especially on the enterprise side where the advantages of thin clients include better security, reduced complexity and more collaboration. Critically, in a thin-client architecture, all important data is stored on a central machine, not on the user’s system. With 600,000 laptops stolen in North America each year and some 85,000 cellphones/handhelds annually left in taxis in Chicago alone, the risk today of sensitive data falling into the wrong hands is immense. IT managers worry not so much about the hardware — which represents minimal costs — but the loss of data and the associated legislative and competitive risks.
The industry is already reacting to that reality. Microsoft’s current handheld operating system, Windows Mobile 5, supports the remote wipe of data should the device get lost. There was also much discussion at Tech Ed about having applications and related documents sit on servers where they can be shared among collaborative teams.
The goal is better security against hackers and assorted malware, better control of what users can and can’t do, and better enforcement of company IT policies across the board.
“It’s a pendulum,” Buxton said. “We’ve swung away from dumb terminal workstations to very powerful machines and back again as the network becomes more important. The computer — the box — becomes transparent. Not invisible, but transparent.”
So the idea is that people will routinely use only the essentials: an input device with its digital key to unlock information linked to a network connectivity device. This digital eco-system, Buxton said, recognizes some technology is suited to some tasks but not others, so the interface and access adjusts for “what, when, why, for whom and where we need it.”
Even the way we input is changing, Buxton said as he extracted a prototype pen designed to write on special paper and record his notes via camera for later input to his IBM ThinkPad. “If I see an article I can capture it with this,” he said, holding up a scanner that looked like a fat windshield wiper. “It’s about absorbing information and de-sorbing it.”
Rather than producing tools that do everything — the Swiss Army knife approach — why not simply use tools that do just what you need?
It’s the kind of thinking that is also behind another development in the area of supercomputing. Instead of one massive computer costing millions, why not several processors working as a team, orchestrated by a single conductor. This idea, called clustered computing systems, is still in its infancy, but has already drastically reduced the cost of supercomputing projects such as weather modeling, genome mapping or bond market risk analysis.
In 1991, a computer capable of one gigaflop (a billion floating-point operations per second) cost about $41 million. By 1997 the same power cost about $1 million. Today, four x86 processors can do this for about $5,000, which buys both hardware and software.
The advantages, then, of centrally located processors with thin clients are real in terms of both cost and control. Ironically, that means Digital’s Olsen wasn’t far off the mark when he questioned why anyone would want a computer in their home. The computer — where the real processing power resides — may soon be located off site and connected virtually, configured for each user.
