Four Options
From The Web Difference
Options for a Net Neutrality Standard
Version 0.1
Background:
One of the biggest questions in the ongoing debate over net neutrality is how to best balance the sometimes conflicting interests of internet service providers (ISPs) and the public at large in the health and viability of the internet, with the public benefits derived from the fundamental operating principles of the “best efforts” networking structure of the internet[1] and the basic consumer rights previously identified by the FCC. While ISPs and other have expended considerable resources on the development of network infrastructure the pace of technological development in terms of internet content and services has in many places outpaced the growth in the underlying infrastructure in many ISPs networks, particularly in the so called “last mile” infrastructure to consumer’s homes.
The fundamental structure of the internet was not designed or built to discriminate between the packets of data being carried on the network. While the network was also not designed to guarantee that packets would be delivered to their intended recipients, it was assumed that all members of the network would do their best to pass on all data packets and that their packets would then receive the same treatment. This structure worked very well in helping to assist the rapid growth of the network and did not present a problem until there were prolonged periods of greater demand for bandwidth than was available on certain parts of the network.
Because most consumers do not have meaningful choice in the broadband service space at the present time the market is ill suited to provide solutions to these problems and thus this is an appropriate area for the FCC to regulate. The FCC has partially addressed their responsibility in this space when they adopted an Internet Policy Statement that contained four principles to govern what consumers are entitled to. Specifically they found that consumers are entitled to:
- Access the lawful Internet content of their choice;
- Run applications and use services of their choice, subject to the needs of law enforcement;
- Connect their choice of legal devices that do not harm the network; and,
- Have competition among network providers, application and service providers, and content providers.
However, these principles were all subject to the needs of “reasonable network management” in footnote 15 of that statement. Thus the key question is what “reasonable network management” is given the current state of broadband infrastructure today and how this should be properly balanced with the rights of consumers as enumerated above. This paper proposes a series of standards which could be adopted on their own or in some combination to define what reasonable network management is and attempt to give some definition to the FCC’s direction on this issue.
Assumptions:
In order to have an efficient discussion about the various options available to network managers and the FCC, the following are assumed about the present state of broadband service in the United States:
- The current infrastructure is incapable of meeting the demands of all subscribers during peak hours.
- Most users have at most two viable options for broadband service.
- Most connections to individual users are asymmetric (they provide much more download bandwidth than upload bandwidth).
- The FCC has the authority through either rulemaking or adjudication to regulate ISPs network management practices.
- The FCC believes in a general principle of neutrality as embodied in the four specific principles identified in the Internet Policy Statement (FCC 05-151, Sept 23, 2005).
None of these assumptions are in fundamental conflict with each other; however, they do present some readily identifiable tensions. Primarily this is the tension that results from a general support of neutrality in a world where the bandwidth demands on the network at some times dwarf the available supply of bandwidth.
Option 1: ISPs can only discriminate against packets from certain applications or processes if and only if the application, service, or protocol is incapable of any substantial legal use.
The FCC has long recognized that users’ right to internet access is conditioned on their legal use of the internet and this was specifically noted in the second and third principles in the Internet Policy Statement. The purpose of this standard is to give ISPs the ability to throttle or even disconnect from the network users that are placing demands on the network as a result of illegal use.
This option specifically targets only those applications that are incapable of any substantial legal use. By doing so this would allow ISPs to throttle or disconnect users whose computers might be deliberately (or inadvertently as a result of being compromised) sending out massive amounts of spam or are being used as part of a distributed denial of service attack. Thus this option would attempt to address a very real problem that exists given current vulnerabilities in computer security.
However, by protecting protocols that are capable of substantial legal uses this would prevent ISPs from targeting protocols that are being legally used even if the ISPs believe that they are putting adverse demands on the network (such as Bit Torrent). This approach attempts to strike a balance between allowing ISPs to protect their network, while still allowing users the freedom to use the legal devices and applications of their choice as embodied in the FCC’s Internet Policy Statement.
Furthermore, this approach seeks to prevent ISPs from taking actions that might raise anti-trust concerns. For example this standard on its own would prohibit an ISP from degrading a legal service that happens competes with a different technology that they offer on the basis that the competing service is detrimental to the network (even if the service in question actually is detrimental).
Option 2: ISPs can not engage in any filtering, traffic shaping, intentional delays, or other forms of packet or application discrimination on their networks.
This option represents the strictest form of net neutrality and has the strongest fidelity to the underlying architecture of the network. Under this regime some packets from users in peak times will likely be lost as the network is incapable of delivering all of the packets being sent to and from all of the users. As a result of this approach many users may have trouble getting content or using services in times of peak network usage.
The exact consequences of this sort of regime are hard to predict. Certainly many more users would be affected by the degradation of the network than if the ISPs only reduced the service of those users that had a demonstrated history of making high demands on the network. However, there are also many benefits to this approach. By requiring that ISPs maintain the basic network architecture that has served as the basis of the internet, future providers of content and services will be aware of the rules of the game as well as the inherent limitations of the network. Thus innovators will likely work to find solutions and design around the deficiencies of the network given enough time and resources.
Furthermore, this approach would put the most pressure on ISPs to continue to upgrade and build out their capacity as well as encourage current ISPs and others to further invest in alternative methods of delivering broadband to consumers (such as FiOS or WiFi broadband). While this may negatively impact the bottom line of ISPs in the short term, in the long term this would provide the strongest market incentives to those that can create and build new and/or improved network infrastructures.
Option 3: ISPs can only discriminate against packets from individuals users that are, or have been shown to be, detrimental to the viability of the network.
The main thrust of this option is to recognize that there are times when the network is incapable of meeting the demands of all users but to only penalize those users that are making the most demands on the network. The most complicated part of this option is its actual implementation, specifically in determining how to decide what users are detrimental to the network.
For example, the network could chose to discriminate against those users that are making the highest bandwidth demands during a given peak time when demand exceeds supply (thus essentially capping all of the users at a certain rate determined by the max speed of the node divided by the number of users currently attempting access). Alternatively, the network could chose to limit the bandwidth of users during peak times for only those users that have shown over a given time period (say the last month) to be those users that have in total used the most of the network’s available resources.
A secondary concern under this plan is whether or not such a restriction would violate the ISPs contracts with users wherein users were promised “all you can eat” pricing for internet access. While this is a concern, most major ISPs contracts are drafted to promise the users the most possible speed up to a certain limit, but are also conditioned on the fact that their usage does not negatively impact the experience of other users.[2]
Option 4: ISPs can only engage in network discrimination with a full disclosure to consumers and upon a publicly published finding that the traffic in questions is having a substantial negative impact on the network.
The final option is the most permissive of network regulation. This option would allow ISPs to engage in whatever network management practices they wanted so long as they first found that the traffic in question was having a substantial negative impact on the network and they publicly published a finding to that effect. While this option is very permissive, these two qualifications are not without some teeth.
By requiring ISPs to make an actual finding of negative impact this would prevent ISPs from engaging in behavior that would completely kill off any innovation on the network because when a new application or protocol is first created it is very unlikely that its early adopters would generate sufficient usage to allow the ISP to find that it is having a negative impact on the network. ISPs would not be able to restrict traffic only on hypothetical or potential harm. Rather there would need to be real data to justify any extensive regulation of a given application or protocol.
Furthermore, the act of forced disclosure allows consumer advocacy groups to be able to intelligently and effectively lobby and pressure ISPs to alter their practices if the ISPs actions are not actually in the best interest of consumers. Unfortunately, because most users are not presented with true choice in their broadband providers it is not possible for users to vote with their feet if they find certain network management practices unacceptable and it is this reality that severely limits the attractiveness of this option.
Hybrid Options:
Some of the above options could be combined in an attempt to create a more comprehensive and effective regulatory regime. For example so long as the network is not under bandwidth demands greater than its supply there is no reason not to fully embrace the strictest form of net neutrality. It is only at certain times when demand outpaces supply on the network that some form of network management of legal use might be acceptable.
Thus, it is likely that any regulatory scheme will allow ISPs to limit applications and protocols that are not capable of any legal purpose (as proposed in Option 1). However, the difficulty is in deciding when, if ever, ISPs should be permitted to regulate users’ legal use of the internet. This paper has presented three different options on that front: the first is a prohibition on the purposeful degradation of legal internet use; the second allows only for regulation that targets those specific individuals that are detrimental to the network; and the third allows for broad regulation but requires a public disclosure and actual evidence of harm to the network.
An Internet Bill of Rights?
The idea of Internet Bill of Rights has been recently proposed by Comcast.[3] Such a proposal isn't necessarily a bad idea but Comcast does want to frame it as not just as set of rights but also a set of responsibilities. Proponents of net neutrality shouldn't necessarily be against such a plan but the devil will be very much in the details of such a plan. Until more is know about what these rights and responsibilities will entail it is hard to say if they will actually represent anything more than the FCC's Internet Policy Statement.
Furthermore, there is an inherent danger in any scheme that provides a set of unenforceable paper rights. Once such rights have been granted it becomes much easier for ISPs to point to their adoption of rights as progress that doesn't actually alter reality for consumers. An Internet Bill of Rights is an interesting idea but one that should be considered carefully and deliberately.
Finally, there is a danger presented by drafting a rigid set of rights because they may not adapt well to future technological changes in the network. By trying to place a stringent set of "East Coast" Law on the network we may jeopardize our ability to more effectively change and alter the network in the future as technologies change.
Footnotes
- ↑ The seminal paper on the "End-to-End Principle" and the basics of best efforts networking can be found at http://web.mit.edu/Saltzer/www/publications/endtoend/endtoend.pdf
- ↑ The Terms of Service from ISPs are, like many simillar contracts, skewed heavily in favor of the providers. See http://ap.google.com/article/ALeqM5gzgjqaZN-Enb-TKgxeyS171dkxbgD8VR5IIG0 . Comcast specifically has already changed their ToS in light of the recent debate over network management. See http://arstechnica.com/news.ars/post/20080207-comcast-tweaks-terms-of-service-in-wake-of-throttling-uproar.html
- ↑ For more information on this proposal, for which details are very scarce see http://news.yahoo.com/s/ap/20080416/ap_on_hi_te/comcast_file_sharing
