“The Netflix debate tells us there is a yawning gap between the reality of current network architecture and the outdated theories supporting our regulatory policies. This gap is the single biggest threat to the virtuous cycle of invention, investment, and growth that have characterized the Internet over the last decade.”
I’m having my own case of Cassandrafreude after reading the responses to my posts on Netflix’s decision to block consumer access to its new Super HD service. One commenter says it is a “great thing” that Netflix is relieving Internet congestion (a tacit admission that Internet congestion actually exists) by deploying computing power inside ISP networks. Another commenter suggests Netflix is attempting to “vertically integrate (from just content provider to content provider + CDN)” because “existing CDNs may not be equipped to handle the new traffic Netflix wants to push over them.”
I don’t disagree with these comments (though I expect existing CDNs could handle the Super HD service as well as Netflix). My Cassandrafreude stems from watching these Netflix defenders “flip the other guy’s argument” by discrediting the “end-to-end” and “infinite bandwidth” theories that originated the net neutrality debate. This “flip” reflects transformations in the architecture and economics of the Internet that are not reflected by our current regulatory framework or recent policy debates.
The Netflix debate tells us there is a yawning gap between the reality of current network architecture and the outdated theories supporting our regulatory policies. This gap is the single biggest threat to the virtuous cycle of invention, investment, and growth that characterized the Internet over the last decade. Though the technological transition from dial-up to broadband network architecture is largely complete, “the regulatory transition is still struggling to get started.” Ironically, net neutrality advocates who once complained that plain old telephone service was inhibiting investment in the Internet are now insisting that we continue maintaining the telephone network.
We should be abandoning regulations intended to preserve what worked in the past so we can discover what will work in the future. If we don’t modernize our outdated regulatory policies, they will set unnecessary boundaries on innovation and investment in our communications networks. Now more than ever, we need a 21st Century approach to Internet policy that promotes boundless innovation.
The architecture and economics of the Internet are continuously evolving
Remember the Internet at the turn of the century? Dial-up Internet access using plain old telephone service was the norm, and the era of broadband Internet access was just beginning. Some predicted that broadband would yield “infinite bandwidth” at zero cost, which would result in an end-to-end (or “dumb”) network that could “let all messages careen around on their own” and “let the end-user machines take responsibility for them.” The end-to-end principle states that the application-specific functions (or “intelligence”) of the network should be limited to end-nodes (nodes on the “edge” of the network) rather than intermediary nodes (i.e., nodes in the “core” of the network), if the “edge” nodes can function “completely and correctly.” This principle was a reaction to the limitations of the old circuit-switched telephone network that had dominated Internet access in the 20th Century.
In the year 2000, law professors Mark Lemley and Lawrence Lessig published a paper equating the “end-to-end” principle with the “social values” of innovation and competition. They argued that a consequence of the end-to-end principle is “non-discrimination among applications.” In their view, resources in the network core should not be optimized for any single application “even if a more efficient design for at least some applications is thereby sacrificed.” They believed this limitation kept the “cost of innovation low” and maximized competition by eliminating the potential for “strategic network behavior.”
The theory posited by Lemley and Lessig in 2000, which ultimately became “net neutrality,” dominated communications policy discussions until the FCC adopted net neutrality rules in 2010. In the meantime, the engineers responsible for making sure the Internet actually works realized that “infinite bandwidth” was one among several fallacies of distributed computing. Engineers also realized that the “if” in the end-to-end principle – which Lemley and Lessig discarded – is a big “if” for streaming video.
While lawyers, economists, and policy makers debated net neutrality, network engineers started implementing the “more efficient design” Lemley and Lessig had been willing to “sacrifice” for their version of “non-discrimination.” Engineers started adding “intelligence” to the network to optimize it for particular applications, and companies that once operated only at the “edge” began to vertically integrate with the network’s “core.” By the time the FCC adopted net neutrality rules, the network architecture envisioned by Lemley and Lessig in 2000 – where all “intelligence” resides at the “edge” with no vertical integration – no longer existed. The network engineers who inspired the social theory of net neutrality had already routed around it.
The engineering approach to streaming video proved the fallacy of the titular assumption in the Lemley and Lessig paper – that “preserving the architecture of the [dial-up] Internet in the broadband era” would maximize innovation. If the engineers had “preserved” the Internet’s dial-up architecture, it’s unlikely that Netflix would be able to offer its Super HD streaming video service at all. The engineering accommodation for streaming video – the distribution of “intelligence” throughout the network – demonstrates the critical role that experimentation plays in producing innovation.
The Internet is killing “cable” and “telephone” networks while 20th Century regulations are attempting to “preserve” them
Today’s broadband Internet bears little similarity to the switched telephone network that inspired the Lemley and Lessig vision of net neutrality. Yet many believe that “Title II” regulation designed for the telephone network in 1934 should nevertheless be imposed on the broadband Internet.
The problem: Title II wasn’t intended to address the intermodal competition and two-sided market issues presented by today’s Internet. For example, former FCC Chairman Reed Hundt recently noted that a 1980s-era government policy that provided dial-up ISPs with “free use of the telephone network without paying” (the “ISP exemption”) accelerated the growth of the Internet. He neglected to mention, however, that telephone subscribers paid to maintain the telephone network the dial-up ISPs used for “free.” Today there are not enough telephone subscribers to support the plain old telephone network, let alone subsidize broadband Internet access.
I’m not suggesting that broadband has eliminated the potential for strategic behavior on the Internet. My point is that the context in which we evaluate that potential is radically different than it was in the 20th Century era of “cable” and “telephone” networks, and this new context (the broadband Internet era) requires a different approach to communications regulation.
The business model Netflix is proposing for the distribution of its Super HD content emphasizes this point. When Harold Feld viewed the Netflix strategy in “cable” terms, he deemed it a “retransmission” fight. In my view, however, the Netflix strategy is something more.
I expect Harold Feld was referring to Time Warner Cable’s recent testimony that “broadcasters’ demands for massive fee increases by blackout threats” are “harming consumers.” In the “cable” context, “retransmission consent” refers to the right of a cable operator to retransmit programming distributed by over-the-air broadcasters. Cable operators cannot retransmit broadcast programming without the consent of the broadcaster. Broadcasters with access to valuable programming typically require payment from a cable operator in exchange for retransmission consent and often withhold their consent (i.e., cause a “blackout”) if the cable operator refuses to pay. But, no matter how much a cable operator must pay, once it obtains retransmission consent, all cable subscribers have access to the broadcaster’s programming.
Though Netflix’s blocking of its Super HD service resembles a “blackout,” it also involves a novel form of tying, a competition issue that isn’t involved in traditional “cable” programming disputes. A tying arrangement is an agreement by a party (Netflix) to sell one thing (Super HD service) only if the buyer (consumers and ISPs) also purchases a different product (Netflix’s “Open Connect” CDN). Though there is controversy regarding the extent to which tying is anticompetitive, the novel tying arrangement presented by Netflix implicates both the per se prohibition on tying (due to the potential competitive harm to other over-the-top video distributors and CDNs) and the rule of reason applicable to non-foreclosing ties (because consumers who don’t subscribe to Netflix will absorb the cost of the tied product (Netflix’s “Open Connect” CDN), but won’t have access to the tying product (Super HD service)).
To be clear, I’m not asserting that Netflix has violated the antitrust laws. Tying is a complex issue and a complete analysis isn’t necessary to make my point: that our current regulations aren’t designed to serve the needs of consumers or our global competitiveness in the broadband Internet era.
The architecture and economics of the Internet are continuously evolving, but our regulations are stuck in the past. To maximize innovation and investment, all participants in the Internet economy should be allowed to evolve their technologies and business models to meet the demands of these dynamic changes. FCC regulations intended to preserve outdated infrastructure and economic relationships are currently inhibiting their evolution – which harms consumers, the economy, and our global competitiveness.