Metal on a Cloud

LAST EDITED APR 2, 2016 · APR 28, 2012

Apple and other companies are betting long on a new kind of network to bring the world’s computing power to your fingertips. But cloud computing isn’t about distributing power. It’s about consolidating it. The cloud is the Internet turned outside in.

Last June, Steve Jobs stepped onstage for his last keynote. You’ll never see lines around the block for what he unveiled that morning. It was a promise, an idea: “We’re going to move the centre of your digital life to the cloud.”

The cloud, my inner reductionist says, is the Internet. It brings things online. By the same trend that will before long turn your wallet and keys into smartphone apps, most of what apps do will run on distant servers. In the cloud, less and less of the state of a thing – whether the tangible form it takes to meet specified ends, or the stored potentials that choreograph its operation – resides within itself. Your devices, like you, depend on a million consorts, working in silent simultaneity, to carry actions out. Literal co-operation. That’s my reading of the metaphor.

And there’s Steve Jobs, putting the lie to my definitions.

He’s introducing iCloud. The point of iCloud is to wirelessly coordinate Apple devices and keep all their content in sync without users having to configure anything – and without any direct communication between the devices. Never again ask what’s stored where, copy files or preferences between machines, or back anything up. Better yet: never have to mess with filesystems or reprogrammable PCs at all. iCloud takes all those features behind a curtain.

While pitching the benefits of this solution, Steve touches on a deeper point: “The truth is on the cloud.”

“The truth is on the cloud.” It’s almost Platonic. A vast network of computers you’ll never see crosses the breadth of your computing experience. Apple’s “digital hub” strategy is ascended to sky, with personal computers (really panes of glass of varying dimensions) as mere spokes that touch ground on the cloud’s behalf. What you hold are now enactments of a higher design, a design that runs deep through their circuitry, not to ask for truth but to conduct it. Truth, of which your experience is only a copy.

This vision of the cloud shows the metaphor’s meaning better than any before it. Your devices, like you, are not in the cloud. You’re under it.

Apple and other cloud vendors are dreaming up a new network topology. It borrows features from the Internet, but leaves out the big one: power at the edges.

To grasp how the Internet’s topology differs from the cloud’s, it helps to go back to a 1962 Paul Baran memo. Katie Hafner and Matthew Lyon recount in Where Wizards Stay Up Late: The Origins of the Internet the thinking that would set telecommunications on its head:

Baran was working on the problem of how to build communications structures whose surviving components could continue to function as a cohesive entity after other pieces were destroyed. He had long talks with Warren McCulloch, an eminent psychiatrist at MIT’s Research Laboratory of Electronics. They discussed the brain, its neural net structures, and what happens when some portion is diseased, particularly how brain functions can sometimes recover by sidestepping a dysfunctional region. “Well, gee, you know,” Baran remembered thinking, “the brain seems to have some of the properties that one would need for real stability.” It struck him as significant that brain functions didn’t rely on a single, unique, dedicated set of cells. This is why damaged cells can be bypassed as neural nets re-create themselves over new pathways in the brain. (101)

What Paul Baran had figured as being ideal for controlled (military) communications evolved instead, almost by accident, into a sprawling digital commons. The hub-and-spoke networks of his day gave way gradually to an interlinking network of networks, designed for rapid absorption of ideas from the margins.


From RAND Corporation’s archive.

In his memo, Baran draws a diagram of three different network topologies. His favoured model is C, a distributed network. Look again at the picture behind Steve Jobs. Its lines converge on that metallic cloud. That’s model A, the centralized network from which Baran and later architects of the Internet would depart.

Not that centres of power went away. Everywhere along the contours of actual networks you will see breaks in Baran’s many-to-many pattern. Whole worlds crop up where users aren’t network constituents anymore. They’re simply users.

In some ways, these worlds bring the conveniences of proprietary networks to the open frontier. Want to make webpages without the headache of running a server? Sign up for Geocities. Want reputable strangers to bid on your unwanted stuff? Run it through eBay. Email on the go? Hotmail. Your social connections? Facebook.

Sites like these have a gravity to them. Each becomes more than a place you go to do a thing: you leave pieces of yourself in there. Pieces you can’t take back. Imagine moving your eBay reputation – all those undisputed sales, all those positive ratings – onto your personal website. It’d be like photocopying money. What having those interactions hosted on ebay.com does is broker their credibility. You can’t run your own instance of what eBay does. Likewise, Facebook makes it easy to download everything you’ve posted there and upload it elsewhere, but to what end? It’s the connections between users and mutual visibility of new content that make Facebook so habit forming, and those again you can’t download.

These pieces, these discrete elements of your connected being, can’t be divorced from either you or the services they’re part of. And they are pulling you toward the network’s centre.

Such tradeoffs of power against convenience aren’t new. What’s new is the level of accumulation. Whole categories of what the Internet is good for – storage, communications, sharing, publication, thought – are falling into vertically integrated silos. And what the App Store showed in 2008 is that even as a platform of distribution, the Internet is replaceable. When propriety networks become the frontier, you’re in new territory. The cloud, namely.

What do cloud vendors offer us in exchange for the keys to the future? The same offering Geocities, Hotmail, eBay, and Facebook were founded on. They take care of hassles.

To steal Ronald Coase’s argument for the emergence of firms in market economies, a cloud vendor will manage another online interaction if the marginal cost of doing so is below the cost of negotiating that interaction through open protocols on the Internet. Cost here includes that most scarce resource, human attention. What catches many off guard, including me, is just how many interactions the cloud might encompass. Institutions are learning ever finer ways to economize on attention, while people linked directly fall further behind.

So where does iCloud come in? Though it makes no pretense to be a parellel Internet, Steve Jobs insisted it’s no mere “hard drive in the sky” either. Very little of what people entrust to iCloud today is anything they’d want to share. But suppose iCloud evolves into a world class publishing system. This could develop any of several ways: future versions of iWorks or iBooks Author; a cohort of popular third party apps that talk over iCloud APIs; Hypercard reprised for iOS.

Whatever its form, this platform’s main features would be

  1. rich content creation with minimal hassles.
  2. access to massive audiences.
  3. iCloud at its core.

Buying into a universe where “the truth is on the cloud” means you’re committing your works, all of them, through their life cycles, to the cloud first and the Internet second, if at all.

And maybe this is where the problem sorts itself out: private networks hit a natural limit where they can’t scale or interoperate like the Internet can. The more data intermingles with proprietary code, the less hypertextual it is. A shift to the cloud for some things doesn’t have to mean an abandonment of the Internet for all things.

It’s worth watching closely, though, how this shift remaps network topology, and what consequences follow from users gravitating toward network model A. To act as switchboard for human activity is to shape it. Since cloud vendors today face none of the pressure ISPs do to act as common carriers, every bit they carry is subject to the free play of their biases.

Which brings us back to truth. There is a difference between not caring where your bits are at any given moment and having no final claims to the truth they encode.

Truth has many features of Paul Baran’s neural net model, only spread out across billions of minds. Neither the solipsism of ‘truth on local storage’ nor the piety of ‘truth on the cloud’ suffices to capture it. Truth is in between people. Authority is vested equally in each speaker, and the canon of truth is held in the balance of their endless renegotiating.

Truth outlives institutions, so the form we capture it in counts. Steve Jobs was alluding to different copies of a file, but the logic carries. The truth is never in one place at once.

The Internet, conceived of as a way to route around catastrophe, came out as an affirmation of truth’s fundamental betweenness. The cloud’s virtue is that of a single, high vantage point which mediates all lower interactions and decides the truth. What gets in the way of such top-down schemes is plurality.

Developers and users may soon confront cases where using the Internet means sidestepping the cloud. That’s a dilemma. If by consensus people’s data and code move from devices they control to servers big companies control, it’s up to hackers to find even cooler reasons for people to break out of that box.

Yet the cloud’s basic promise is right: Bring everything online. Don’t think about where your bits are. Screens are windows into your stuff.

That defers the real question one step, though. In a future in which your bits are floating in the air, who shares computing power with you, and on what conditions? The answer has more than technical implications. The answer determines who gets read-write access to the record of what is.


Thanks to jenanne ferguson and Timothy B. Lee for looking at drafts of this. Looking back at it four years later, I should have taken Tim’s suggestion to de-emphasize Apple in my argument. The symbolism was sexy, but Apple was and remains bad at the cloud.