I've addressed some of your comments about the broadband market and differentiating web traffic based on type. Now I'll turn to another big subject of your comments on net neutrality.
Doug, Keith, Drywall, asokoloski, psmith, and others engaged in a lively debate about whether and how Google and other Web companies compensate the telecom infrastructure providers for our use of their network facilities. As many well know, the Internet’s longstanding charging arrangements allow each party to pay for its own connection to the Internet. That party then is free to utilize that connection in whatever lawful ways are desired. Google believes that consumers should be able to acquire higher speed or performance capacity from the broadband providers, and then use this capability to reach any service they wish on the Internet. In particular, consumers should be able to purchase tiered pricing arrangements, based on the use of bandwidth, latency requirements, or other objective measures. Such arrangements would constitute an appropriate, cost-based practice that fully compensates the broadband provider for the additional capabilities provided.
On the other end of the “pipe,” Internet-based companies spend billions of dollars annually on R&D to create and deploy compelling content, applications, and services for American consumers. This massive amount of material typically is deployed on millions of Web servers located around the country. In order for the content and applications to be delivered into the Internet, so it then can be made available to consumers, Web companies must arrange with network operators to: carry the data traffic from company facilities to their Web servers over local telecom lines (the “last mile”); carry the data traffic from the Web servers into the Internet over high-speed, high-capacity data lines (“special access”); and carry the data traffic over the numerous interconnected networks that make up the Internet (the “Internet backbone”). To accomplish these important connectivity and transport functions in a fast and effective manner, Internet companies collectively pay many billions of dollars per year to network operators, which fully compensates them for their network investment.
We believe that broadband providers should be precluded from charging content providers for terminating traffic to a particular end user. Allowing broadband providers to leverage their “situational monopoly” over terminating traffic would allow them to choose which content providers receive preferential treatment over others, thereby distorting the marketplace. The institution of terminating charges also could lead to the balkanization of the Internet, in which each of the hundreds of local telephone and cable operators around the country – and, perhaps even more importantly, around the world -- would assess its own set of fees for terminating traffic on its network.
I hope these clarifications have been helpful, and that you'll keep sharing your thoughts.
P.S.: Be sure to check out Robert Cannon’s outstanding blog, Cybertelecom, which should be required reading for anyone interested in the Internet and broadband policymaking discussions in D.C.
Yesterday I addressed some of the comments on my net neutrality post dealing with the broadband market. Today I'll delve a little deeper on another issue you asked about: type-based traffic differentiation.
Several users commented on Google’s position that reasonable type-based differentiation of Internet traffic can be an acceptable business practice. As we explained in our FCC comments, we do not dispute that broadband providers should have the ability to manage their networks, as well as engage in a broad array of business practices. To us, the real question comes down to what kinds of business models and network management techniques rely on unilateral control over last-mile broadband facilities (the proverbial “on-ramps” to the Internet), in the service of anticompetitive or discriminatory intent.
Most known network management techniques will create few if any marketplace harms. So, for example, we believe that a broadband provider should have the leeway to utilize legitimate application and content-neutral network management practices that seek to neutralize objective network harms. These practices would include halting harmful denial of service (DOS) attacks, or blocking certain traffic containing viruses or worms.
We also stated that it may be a reasonable business practice to prioritize all packets of a certain application type. Our rationale for that position is that there may well be tangible end user benefits from giving preferential treatment to certain Internet packets, such as those in a streaming video transmission, in order to enhance the end user experience. As long as the categories of “type” are identified and designed with objective criteria in mind (such as sensitivity to latency or jitter), and prioritization is apply in an even-handed manner to all packets in that category, the practice can be a fair one. If, on the other hand, type-based prioritization is used to promulgate discriminatory practices – such as degrading or prioritizing certain applications based on an intention to impair the offerings of competitors – such practices should be prohibited as unreasonable.
I will be the first to say that allowing type-based prioritization is a close call, and reasonable minds certainly can differ. Many in the Internet community lack trust that the broadband provider will employ packet prioritization over last-mile networks in a manner that still preserves an open Internet environment and does not facilitate the introduction of anticompetitive practices. Moreover, prioritization generally creates a host of practical, economic, and technical problems, not least of which is that the broadband carrier has fewer incentives to build out its network capacity where it can make more money simply by charging for differentiated service.
On balance, though, we believe that the possible end user benefits from differentiating between certain broad categories of Internet traffic outweigh the potential competitive and discriminatory threat. That doesn’t mean that we cannot subsequently criticize, and seek to halt, any such practices that take an anticompetitive turn. Nor does it mean that Google somehow is going “soft” on network neutrality. We have merely drawn the line in a slightly different place than others in the pro-net neutrality camp.
Tomorrow, I'll address your comments about another net neutrality topic: paying for bandwidth.
Thanks to all who read my initial posting on network neutrality, and especially to those folks who took the time to leave comments. While I don’t have the personal bandwidth (ouch) to respond to each and every posting while also taking care of my “day job” here at Google, I will check back periodically and offer follow-up reactions.
I believe it is important for companies like Google to establish a place of meaningful dialogue with the general public, and to open our policy advocacy role to outside analysis -- and yes, criticism. I also welcome your thoughts on other telecommunications and media policy issues of interest to you (my own current favorite topic is the FCC’s ongoing consideration of rules governing the upcoming 700 MHz auction). And I urge folks to take their views to the places where they ultimately count: the well-trod halls of the FCC and the U.S. Congress.Today, I'll offer some thoughts on one of the key issues raised in some of the comments on my net neutrality post: the broadband market. Later this week I'll address two other issues you asked questions about: type-based traffic differentiation, and payment for bandwidth.Market analysis
Scott Cleland asked whether the search market is as highly concentrated as the broadband market, and thus deserves network neutrality regulation as well. Scott asked me the same question at an EDUCAUSE policy conference last month, but I’m happy to repeat my response and elaborate here.
I’m certainly no economist, but I do try to keep up on the latest thinking about how markets function. The available evidence demonstrates that the U.S. consumer broadband market is highly concentrated, with extensive barriers to entry, high consumer switching costs, and no near-term competition. By stark contrast, the search market is robustly competitive, with numerous major players, new near-term competition, no significant barriers to entry, and zero user switching costs.
Together, these salient factors -- excessive market concentration, no viable competitors, considerable consumer switching costs, and substantial barriers to entry -- should lead policymakers to conclude that there is a major competition problem in the broadband market. No such problems exist in the search market.
I'll have more to say later this week about some of the other issues you've raised. In the meantime, what do you think?
Local and state officials can layer and tailor secure information about their jurisdictions and feed it into a broader database that will give state and federal decision makers valuable and timely information. With existing state GIS (Geographic Information System) and orthophotographic data, we are able to transform massive amounts of useful information into a common operational picture. Examples of real-time applications include emergency evacuation routing, vehicle and asset tracking, critical infrastructure mapping, plume modeling, real-time sensor feeds, real-time streaming video, risk visualization, and post-event imagery placed alongside pre-event imagery.