The Future of Integration Is Cloudy

Opinion: Microsoft's Origami doesn't push the envelope; truly portable computing doesn't tie a user to a single device.

The preannouncement hype surrounding Microsofts "Origami" reminded me of the TV ads that preceded the launch of the Infiniti brand.

Those ads never showed the cars, but instead showed pictures of surf and mountains while talking about the experience to come.

Microsofts coy video clips at were like those Infiniti messages, albeit with two new features: a soft-porn sensibility, with a floating startup button labeled "Touch Me" and background audio that sounded like seductive breathing, and a blue-hued backdrop that looked a lot like the default desktop from Apples Mac OS X.

Someone evidently told an agency, "Make this as sexy as Apples stuff!" Meanwhile, Intels agency seems to have been inspired by Targets TV ads when crafting the parallel Ultra Mobile PC campaign at

Apart from the target-market tea-leaf reading that these "lifestyle device" ads inspired, the key question raised during the buildup concerned the kind of integration that people really want in a portable device.

This drives, in turn, Web content and service design, ultimately feeding back into future devices that will range from things that we carry or wear to things that we drive and places where we live.

/zimages/4/28571.gifOrigamis future: doorstop? Click here to read more.

The two extremes of integration are what Ill call the monolith and the cloud. A monolith does many things with one object: This can be a concrete object, such as a piece of portable hardware, or an abstract object, such as a CRM suite.

A cloud can add new things at its edges: It lets anything connect to anything else, in a way that lets any element become a component of something larger—without a single-point-of-failure dependence on any one part.

Monoliths are necessary when usage models are still taking shape, when hardware is expensive and when standards have yet to emerge. Clouds become more appealing as technologies mature.

An integrated PC, for example—desktop or portable—is a monolith. If the display of an Apple iMac, or of pretty much any portable PC, should break, the device is broken. Ditto the hard drive and, on a laptop, ditto the keyboard. Yes, you can plug in an alternate display and/or keyboard to do urgent work or recover files if the hard drive is still working, but its not what the unit was built to do best.


The computing experience that I want neither ties me to a desk nor makes me carry hardware I dont want. I dont need to carry a hard drive if a Wi-Fi connection gives me access to any files that I need; if performance and disconnected operation are important, a Bluetooth-enabled flash-memory buffer in my pocket can hold whatever Ive recently created or used.

I shouldnt need to carry a display. Wireless links should discover available display devices, whether were talking about a small screen on my wrist or cell phone, a desktop display in an office or a big-screen display in a conference room.


Likewise for keyboards, speech-command systems or conference-room presentation consoles—if I want something special, Ill carry it, but, otherwise, Ill often be able to live off the digital land.

"Next-generation portable hardware" is almost an oxymoron because the next generation isnt portable hardware at all—its a cloud of network services and devices that know where I am, observe what Im doing, anticipate the resources that Ill need, secure the required connections as desired and allocate those resources to me on request.

We have the building blocks in place to do this, but there are some new points of view that we need to develop to do it well.

For example, this model depends on authenticating the user, not the device, because the notion of strong coupling between a particular user and a particular piece of gear is an artifact of the first two decades of portable computing. On the highway, we distinguish between the vehicles license plate and the drivers license; the Internet is long overdue for similarly independent identities for devices and users.

Old approaches dont become newer by making PCs smaller or making batteries last longer. Something truly new is overdue.

Peter Coffee can be reached at

/zimages/4/28571.gifCheck out eWEEK.coms for the latest news, reviews and analysis in programming environments and developer tools.