A phrase that seems to be cropping up more and more to describe Web 2.0 and the evolution towards Web 3.0 is ”software above the level of a single device.”

Not exactly a mellifluous phrase that rolls off the tongue, but the concept is key. It’s also locked into the concept of ubiquitous computing, although they aren’t the same thing.

The way I understand it, software above the level of a single device has the following characteristics:

1. Platform agnostic
2. Integrates computing power of different scales and different purposes

The classic example widely used is iTunes. It seamlessly integrates (1) the user’s computer (which runs either Mac OS X and Windows—not quite platform agnostic, but still useful as an example) (2) the user’s iPod and (3) Apple’s iTunes Music Store servers.

But I also think that Web 2.0 encompasses things like RSS aggregators, webmail clients, and SMTP-to-SMS gateways. You could probably throw in IM into the mix, too. Online RSS aggregators like Google Reader are definitely platform-agnostic and integrate computers of different scale and purpose, tying together blogs, online incarnations of dead tree newspapers, and community forums like Slashdot and digg, and you may very well be able to get your mobile phone to browse Google Reader directly (although I haven’t tried it myself.) Neither webmail clients nor SMTP-to-SMS gateways really have the sheen and polish that is typical of Web 2.0, mostly because they have existed long before AJAX became such a widely-deployed program technique. But their existence allows you to easily keep a real-time online presence, assuming you always carry your phone with you. With most mobile phone software suites including some form of online instant messaging, the evolution continues, and now that Gmail integrates AOL Instant Messaging, the cycle of recursion is complete.

To elucidate it more concretely, what I mean by integrating computing power of different scales and different purposes is that there is an asymmetry between the computers (used in a relatively loose sense) involved in the integration. For example, your mobile phone or digital music player may very well sport an ARM XScale CPU, running at something like 624 MHz, with at least a few hundred megs of RAM, plus-or-minus a hard drive. Meanwhile, your desktop or notebook has some sort of Intel or AMD processor running somewhere in the 2 GHz or more range, with possibly several gigs of RAM and anywhere between 100 GB to a terabyte or more of hard drive space. And the average server farm usually uses several boxes with several Intel, AMD, or even PPC CPUs, each with some enormous amount of RAM, and some exponential amount of hard drive space.

And generally, the purposes of each level of computing power differ as well. Your mobile phone is generally most useful for, you know, calling people, and maybe sending text messages as well. Your digital music player pretty much plays digital music. Your notebook or desktop, while still a much more general purpose machine, is, for the average person, often relegated to workaday tasks like composing documents. Meanwhile, these server farms are usually dedicated to a single purpose: hosting content. In the case of iTMS, this means digital music and video. In the case of Google, this means the index of the entire web. In the case of Wikipedia, it means a compendium of everything and anything.

I think this may be where the client-server concept missed the mark. In the 1990’s, people really envisioned thin-clients in a really limited role. They were just more sophisticated dumb terminals connected to some type of server that didn’t necessarily have to be a mainframe. The so-called Network PC never really took off because all we were doing was shuffling around the hardware. In the end, you would use a thin-client in pretty much the same way you would use a full-fledged PC.

Handheld computing also sort of missed the mark as well. Palms and Pocket PCs really just try to cram desktop computing applications into a smaller space with a weirder UI—they’re essentially a substitute for situations for when you don’t have access to a desktop, or can’t really carry a notebook.

What needed to happen was specialization.

The all-in-one capabilities of a PC are starting to become less critical. Computer appliances are finally really starting to take off. Mobile phones are essentially ubiquitous. Digital music players are as prevalent as portable cassette players and portable CD players were in their respective eras. Digital video recorders are practically standard with cable packages nowadays. Gaming has been widely dominated by consoles and their handheld counterparts. The applications on dedicated (but smaller scale) computers (again, loose usage) are typically far superior in terms of convenience when compared to their notebook/desktop counterparts.

On the other end of the spectrum, any content that hopes to be significant must sit within cold iron that is tethered to the Internet. While you can certainly serve content from your desktop PC, this is neither reliable nor generally supported or even allowed by many ISPs. Any serious web app has to be available 24 hours/365 days, and a DSL connection ain’t gonna cut it.

So what is the role of the PC? If you follow O’Reilly’s thought process, it is the conduit. The configurer. The deployer. While you can certainly configure your mobile phone from within its sometimes inadequate UI, and even browse the web on it, things are much easier and cheaper when you can just push content from your desktop to your phone. The reigning paradigm is, and will probably continue to be, browsing the web from a large screen. This is likely to continue even when desktops and notebooks as we know them are largely obsolete. Ubiquitous computing as envisioned by the intersection of popular culture and science fiction in films such as “The Matrix” and “Minority Report”, while abolishing the notion of a huge non-descript ugly beige box with a noisy fan, still favor visual representation of information on a large viewport or at least on multiple mid-sized viewports. Even with the advent of the iPhone, I am skeptical that the general public will favor manipulation of data on small viewports over large tactile viewports that will initially be displayed on large plasma screens, then perhaps on digital paper, then, eventually, when the ubicomp revolution is complete, in thin air using holographic technology.

When ubicomp becomes mainstream, I envision an era where the PC as an actual device may not even exist. When hardware is cheap and truly ubiquitous, in the sense of a thorough blanketing of all the meatspace commonly traversed by people, the function of PCs can be distributed. If input devices become essentially virtual (like the tactile GUI featured in “Minority Report” or even existing holographic keyboards) and the display is also purely holographic, with ubiquitous wireless communications, the actual computer hardware need not be in the same room, or even in the same building, and if the wireless network is truly ubiquitous, you can easily roam from place to place.

But seeing the awkward phrase “software above the level of a single device” often has got my gears turning. There’s got to be a more elegant way to encapsulate this concept.

The move from the all-in-one PC to a world more dependent on smaller, cheaper, dedicated smart peripherals (meaning peripherals that have their own CPUs, RAM, and even hard drive space) as well as on larger server clusters and mainframes reminds me a lot of the evolution of single-celled organisms to multi-celled organisms. You can think of all-in-one PCs as single cells, and thin-client networks as colonies of single cells. Mainframes are huge single-celled organisms. Server farms and clusters are syncitiae. The real evolutionary leap is when single cells began to specialize and perform unique but limited functions.

I am tempted to refer to the use of software above the level of a single device as multicellular computing, but this is unlikely to resonate with people who don’t have backgrounds in biology. Still, there’s got to be a better name for it.