Disclaimer: This is a work in progress. Comments are welcome.

Factions in a Digital Age

-- By AndreiVoinigescu

Table of Contents

---

Introduction

In The Federalist No. 10, James Madison warned of the threat posed by factions to self determination through democratic governance. Factions are groups of individuals united by interests adverse to the rights of other citizens and the permanent interests of society as a whole. While Madison worried about how to keep factions from usurping the power of the state through the political process, the march of technology has made it evident that state coercion is neither the sole nor the most ominous mechanism of control over human behavior. Just as the American Constitution seeks to limit the power of factions by incorporating principles of federalism, separation of powers and bicameralism into the architecture of the state, so too must we now look to the architecture of the internet as a means of diffusing the power technology grants to factions. We can--and should--build a new network where ownership and control over the switches is maximally dispersed.

How Technological Change Threatens Democracy and the Rule of Law

Lawrence Lessig has identified four mechanisms though which human behavior is controlled: laws, social norms, market forces, and physical architecture. Physical architecture is a far more powerful means of regulation than law: physical architecture creates self-enforcing ex-anti constraints, while law can only threaten ex-post punishment and requires a complex bureaucracy to enforce. Laws against drunk driving are less effective then breathalysers hooked to the ignition switch.

For activities conducted over networks, the code than controls the switches defines the physical architecture of the network. It prescribes what can and cannot be done on the network, which communications can get through, which are to be modified (and how), and which are simply not to be forwarded on. Modify the code on enough switches to block the http protocol, and you could have an network where websites no longer exist, but email, VoIP? , torrents and other activities are unaffected. And that's just the crudest example of what you can do when you control the code.

Our ability to modify the physical architecture of the real world is still, fortunately, rather limited. But in a networked world, code is easy to modify. The owner of the switch has almost unlimited control over what is and isn't possible on the network. And, in the Internet as it exists now, ownership of the switches is concentrated among a relatively small number of Internet Service Providers (ISPs), most of whom are commercial entities. This creates an environment where a single ISP (or a few acting in concert) can regulate human behavior much more completely than the most determined police states of the Cold War era ever could. Comcast's unilateral decision to throttle BitTorrent? traffic across its network is a pale hint of what we can expect to see as switch hardware develops to allow real-time deep packet inspection of all network traffic and the owners of the switches become increasingly savvy about the power they control.

Network Architecture as a Check on the Power of Code

The solution to the dangerous concentration of power in the hands of the switch owners cannot be government regulation alone. First, such regulation only transfers the control from ISPs to political incumbents who would undoubtedly be tempted to employ it to consolidate their own power. Second, legal regulation it is only reactive, not preemptive; some switch owners will be willing to risk the consequences of the law if the immediate rewards are large enough. To stop factions from using the unprecedented regulatory power of code in ways that hurt democracy, we must take a page out of The Federalist #10, and create a network where ownership and control of the switches is dispersed as widely as possible.

Building a Commonly-Owned Internet

The current Internet architecture is very hierarchical. It is a network of networks that can be visualized like a pyramid with end-points at the bottom. Groups of end-points (laptops, PCs, internet-enabled cellphones, etc.) connect to a switch at the next layer up, allowing communication between all the end-points connected to that switch. Groups of switches in turn connect to one switch at a higher layer of the pyramid, and so on. Any end-point can thus communicate to any other end-point by climbing up through the levels of switches until it finds a switch that both end-points can reach, and then descending again. End-points never communicate directly with each other.

The hierarchical structure of the Internet is an part a relic of the use of wired links to connect computers together. While wireless (radio) technology and wireless routers are starting to replace Ethernet networks at the base of the pyramid, those wireless networks remain hierarchical.

A flat network would eliminate the distinction between end-points and switches. In a wireless mesh network, for example, computers communicate with each other though wireless radio links without the need for a switch to route traffic between them. Rather, A talks to B directly if B is in range; if B is not in range, then A talks to C, and C forwards A's message to B. Each computer can act as both an end-point and a switch in the network. Thus, no third party is need to construct, own and maintain the infrastructure of the network; the cost is spread out among the individual computer owners. Because the technology is cheap (and getting cheaper), the cost of owning such a switch is low. Indeed, the One Laptop Per Child Association has integrated wireless mesh networking using the IEEE 802.11s standard in sub-$200 laptops meant to be deployed in areas with limited Internet infrastructure.

In a wireless mesh network, control over the switches is maximally dispersed. Each computer user can control what code is run on the wireless switch that is part of his computer--and each user can also choose which switches to trust in forwarding his communication. Thus, if A wants to talk to B and B is far enough to require an intermediary to forward the message, A can choose if it will send the message using C as an intermediary (and thus accepting whatever code-based regulations C imposes) or opts for a different route to reach B instead--perhaps forwarding the message through D, E and F, who use switching code from a source A trusts. While most people will have little interest or aptitude for tinkering with the code that handles the switching functionality of their device, they will be nonetheless be able to meaningfully express their views on what types of code-based regulation over the network are appropriate by installing firmware created by others that embodies those rules.

February 2009, Digital Broadcasting, and Spectrum

The F.C.C and White Spaces

Broadcasting or Webcasting?

--