Infrastructure and Platforms
There will be a session at the Supernova Forum on “Evolving Digital Infrastructure: Everything is a Platform.” As often happens, I wasn’t entirely sure what I had in mind when I defined the topic. I just had a sense there was a rich vein for conversation. Here are some of my current thoughts about the topic. I welcome your feedback and suggestions to help refine the session.
Fixing the topics for Supernova is a kind of magnet; it gets me to notice relevant developments. This particular topic brings together two important concepts that don’t quite overlap. One is infrastructure; the other is platforms. A number of friends such as Doc Searls and Brett Frishmann have written about the Internet as infrastructure, there is even an Infrastructure Group now at the Berkman Center. At the same time, it’s hard to follow tech industry developments without talking about platforms: Windows, the iPhone, Facebook, Java,Google, Flash, Android, Twitter… you get the idea.
Infrastructure and platforms are similar but not the same. Both involve layers: there’s something underneath (the infrastructure or platform), and some things on top. Yet I’m not sure I’d call Facebook infrastructure, and I definitely wouldn’t call a highway a platform. What’s the difference? Traditionally, when we’ve talked about the infrastructure of the Internet, it usually meant the physical network connections. The Internet service provider was the road; the hardware, software, and sites on the Internet were something else. Platforms have been all about software APIs and developer communities.
That’s changing. One aspect of “Evolving Digital Infrastructure: Everything is a Platform” is that it’s no longer easy to know what’s on top and what’s below. If I use Google services on an iPhone that’s connected through AT&T Wireless, where do the infrastructure and platforms end? And should we think about the innovation dynamics differently in different parts of that ecosystem? Steve Jobs’ explanation for excluding Flash from the iPhone and iPad captures the dynamic. Everyone sees themselves as the innovation-enhancing platform and the other guys as the innovation-killing gatekeepers. I’m not saying Verizon blocking Google is identical to Apple blocking Flash, just that they are similar problems. We have to look at the details about competition and incentives to decide what to do.
For a conference yesterday on linking engineering and policy for broadband, I was asked to talk about the layered model. This is an idea that I and several other scholars developed a few years ago. Basically, it says that we should think about regulation of the Internet the same way engineers think about its architecture: as a stack of layers connected by interfaces. This was very different from the traditional regulatory categories, which imposed completely different rules for silos such as telephone and cable TV networks. The layered model has been widely accepted in communications policy discourse, and I still think it’s a useful tool. However, like any model it oversimplifies.
The point I made in my presentation yesterday was that we need to appreciate what the layered model (and the parallel notion of “end to end” network design) imply. Layers mean you can ignore everything lower down. In computer science terms, it’s abstracted. In reality, though, that abstracted stuff may be a company or set of companies that has its own view of the network. My substrate might be your market. We need to start thinking more about the alignment of incentives across layers. It’s all well and good to ask the other layers to be simple and open, but they need to have viable business models. The alternative is to define some layers as public utilities in the classic sense, owned or funded by governments. While I don’t think we should write off that possibility, it would be better to find an equilibrium in which competitive market dynamics operate everywhere in the stack.
This is part of what the FCC is suggesting with its Third Way proposal for broadband regulation. (Disclosure: I have a consulting engagement with the FCC, although I as not directly involved in yesterday’s announcement, and these are entirely my personal views.) Finding that the transmission component of broadband access is a “telecommunications service” does not imply public utility rate regulation of all broadband providers. What does it imply? That’s the challenge the FCC faces now.
Supernova is a rare place where those who think hard about Flash vs. iPad come together with those who think hard about network neutrality and those who study the economics and science of networks. Such conversations are more important than ever.
Kevin, you’ve long been an advocate of onerous and stifling regulation of the Internet, and here we see it again: you want to break it, artificially into layers (even though, in real life, many layers are rarely implemented separately) and then regulate the layers you wish to regulate.
Will real life Internet service providers — those of us actually out in the field doing the hard practical work of hooking people up — be represented in your session? Or will you include only those who agree with your hyper-regulatory approach?
I challenge you to include some real life ISPs on the panel. Not lobbyists from big telephone companies who have never touched a Cat5 cable, but real ISPs like myself. You might then get a real world perspective on these issues.
I don’t see any call for regulation, onerous or not. What I see is a description of something that has dogged the emergence of the internet for as long as it’s been emerging. This essay is an attempt to grapple with some of the basic definitional difficulties that plague us all when talking about “the network.”
In fact I see reclassification as a way to free broadband providers not just from regulatory burdens, but also as a means to provide regulatory certainty. The ISP operators I talk and work with have stated many times that a particular regulation is not that big a deal. What drives them crazy and threatens their existence has been the constantly shifting nature of the regulation over the past 15 or so years. I mean for pete’s sake, the Telecom Act of ’96 has been to the Supremes 3 times already. It’s a heckuva way to regulate an industry of such fundamental importance to us all (to paraphrase Justice Scalia).
I hope this process affords some certainty and of a reasonably long duration to afford certainty to an industry badly in need of it.