After a year of debates and a month before the Federal Communication Commission’s (FCC’s) rulemaking on network neutrality, the GOP has finally joined the party. Through a draft bill released late last week, congressional Republicans have taken a step in the direction of supporting network neutrality. That’s a good thing, and moves them closer to the existing consensus. Roughly four million Americans submitted comments to the FCC calling for real network neutrality rules over the past year, and polls show that both Republicans and Democrats overwhelmingly support a ban on fast lanes.
But, as written, the Republican bill provides network neutrality in name only. At first glance, the bill purports to ban paid prioritization, throttling, and blocking and applies the same rules to fixed and mobile networks, echoing language used by President Obama and FCC Chairman Tom Wheeler to describe their network neutrality proposals. But on closer examination, the bill is so narrowly written that it fails to adequately protect users, innovators, and speakers against blocking, discrimination, and access fees.
A meaningful network neutrality regime requires bright-line rules prohibiting all forms of access fees, application-specific discrimination, and blocking. Unfortunately, the Republican bill is insufficient along each key dimension required to achieve real network neutrality, thereby dramatically departing from the network neutrality consensus that emerged over the past year. Thus, as currently written, the bill does not constitute an alternative to the adoption of meaningful network neutrality rules by the FCC under Title II of the Communications Act, coupled with appropriate forbearance.
Key Problems with the Republican Bill
Here are some of the problems with the bill.
1. The bill doesn’t actually ban “paid prioritization” (aka access fees).
The bill imposes a ban on paid prioritization, but then defines the term so narrowly that it captures only a small subset of practices that harm innovation and free speech online. Mirroring the definition of paid prioritization used by Internet Service Providers (ISPs) in their filings with the FCC, the bill would prohibit ISPs from charging providers of Internet applications, content, and services a fee for prioritization of their packets. It would allow ISPs to charge for any other forms of preferential treatment that do not involve prioritization.
By contrast, when network neutrality proponents call for a ban on paid prioritization, they use the term broadly—as shorthand for a ban on all forms of access fees. Network neutrality proponents are concerned about the harmful effects of allowing ISPs to charge application and content providers for any form of preferential treatment—not just about the harmful effects of charging for prioritizing packets.
The bill’s narrow definition of “paid prioritization” renders the ban on “paid prioritization” meaningless. While the bill would prohibit Comcast and other ISPs from charging Netflix for improving the performance of its online video service by prioritizing Netflix packets, ISPs could still strike deals for any other type of preferential treatment. For example, Netflix could pay to ensure that its data packets receive a guaranteed amount of bandwidth during times of congestion or to ensure that its packets do not count against subscribers’ monthly bandwidth cap—a practice called “zero-rating.” But if large, established companies can pay so that their content loads faster or doesn’t count against users’ bandwidth caps, start-ups that can’t pay will be unable to compete. Thus, these practices pose the same threat to innovation and free speech as fees in exchange for prioritization.
Worse, under the Republican bill, deals for preferential treatment other than prioritization are automatically legal and cannot be challenged. This allows ISPs to enter into the worst types of deals: exclusive or discriminatory ones. An ISP could offer a guaranteed amount of bandwidth only to Netflix, but not to Hulu, or charge different prices to Netflix and Hulu for the same service, giving ISPs a powerful tool to distort competition among applications or classes of applications. Even the rules proposed by the FCC in May 2014 would not have gone so far to insulate ISPs. Those rules would at least have allowed the FCC to review such deals to ensure they were commercially reasonable.
The bill is silent on the second kind of access fees—fees that an ISP charges to application and content providers for access to their Internet service customers. The FCC’s Open Internet Order treated this kind of access fee as a special case of blocking and explicitly prohibited it.
Any meaningful network neutrality bill needs to ban all types of access fees, not just fees paid in return for prioritization. As the FCC in its Open Internet Order and President Obama have recognized, this approach is the only way to protect Internet users’ choice and prevent companies that are able to pay from having an advantage over companies that cannot pay. This principle is essential to ensure that start-ups, small businesses, independent artists, and nonprofits can compete and be heard, which is vital to innovation, economic growth, and free speech.
2. The “no throttling” rule prohibits only a subset of ISPs’ harmful discriminatory practices.
Any meaningful network neutrality regime includes a nondiscrimination rule that constrains ISPs’ ability to engage in forms of differential treatment that fall short of blocking. Such behavior is often an attractive alternative to blocking, since it allows an ISP to make certain applications more or less attractive in a less drastic way—obtaining the same effect as outright blocking but at lower costs to the ISP. Thus, differential treatment provides another mechanism for an ISP to distort competition and user choice. Without a nondiscrimination rule, ISPs—and not the market—can pick winners and losers online.
While the Republican bill includes a “no throttling” rule, that rule has two significant flaws. First, the ban on “throttling” only addresses some forms of differential treatment: it narrowly applies only to technical discrimination, i.e., the differential handling of packets in the network. The bill does not limit economic forms of differential treatment like zero-rating or application-specific pricing, providing ISPs with ample means to distort competition and interfere with user choice. Second, even for technical discrimination, the bill allows ISPs to discriminate against classes of applications, again creating a gap in protection.
As to the first flaw, because the bill’s “no throttling” rule applies only to technical differentiation, it allows ISPs to favor certain applications over others using nontechnical means. For example, an ISP could exempt its own application (or a partner’s application) from its subscribers’ monthly bandwidth cap, while counting competing applications against the cap.
Indeed, ISPs already have engaged in this type of behavior. In 2012, Comcast introduced the Xfinity TV app for the Xbox. The application allowed people who subscribed both to Comcast’s pay TV and to its Internet service to stream on-demand video over the Internet to the Xbox. Traffic created by that application did not count towards subscribers’ monthly bandwidth cap, while traffic created by unaffiliated online video applications, such as Netflix, Hulu, or HBO GO, counted towards the cap. Like technical discrimination, these exemptions from bandwidth caps, also called zero-rating, artificially make some applications more attractive than others. And just like technical discrimination, these forms of economic discrimination allow ISPs to tilt the market in favor of specific applications and to “pick winners and losers” on the Internet.
Similarly, the bill would allow ISPs to vary charges for Internet access depending on the applications used by a subscriber. For example, in Europe, many mobile ISPs ban the use of Internet telephony applications such as Skype; people who want to use these applications on their mobile devices can buy an “Internet telephony option” that allows them to use Internet telephony for an extra fee. This allows ISPs to effectively tax certain applications and make them less attractive to users, or to extract more of the value that users derive from the use of those applications.
Second, even for technical discrimination, the protections afforded by the bill are incomplete. The “no throttling” rule only prohibits technical discrimination based on source, destination, or content. While this rule would prohibit an ISP from singling out an application by a specific provider for positive or negative treatment, it seems to allow discrimination among classes of applications. So an ISP would be prohibited from speeding up Netflix over Hulu or YouTube, or from giving preferential treatment only to its own online video application by throttling unaffiliated ones. That’s good. But ISPs would be able to steer users towards certain categories of applications and away from others. Again, this allows them to distort competition and interfere with user choice. For example, ISPs might provide low-delay service that might be used for online gaming to users for a fee, but refuse to offer low-delay service to online telephony applications like Skype or Vonage that might also benefit from this type of service, since that would make these applications more competitive with the ISP’s own telephony offering.
Any meaningful network neutrality regime needs to include a nondiscrimination rule that applies to all forms of differential treatment and bans discrimination based on application or classes of applications (what network neutrality proponents call “application-specific discrimination”). The Republican proposal falls short by just banning technical discrimination, and even in that instance, the bill only bans discrimination among applications in the same class.
3. The bill’s exception for reasonable network management does not require application-agnosticism, opening the door to discriminatory network management practices.
In the bill, the rules against blocking and throttling, as well as the device attachment rule that allows the attachment of nonharmful devices, are all subject to an exception for reasonable network management. However, unlike the reasonable network management exception from the FCC’s 2010 Open Internet Rules, the bill does not require ISPs to manage their networks in a manner that is as application-agnostic as possible. Despite copying nearly verbatim the reasonable network management exception from the 2010 Open Internet Rules, the bill fails to include the Open Internet Order’s essential qualifying language: according to the text of the Order, the FCC would have evaluated the reasonableness of network management practices based on whether they are application-agnostic and respect the principle of user choice.
Because the bill only requires network management to be “appropriate and tailored,” ISPs could still justify network management practices targeting specific applications or classes of applications as a tailored, and therefore permissible, approach to managing congestion, as long as the discrimination is limited to times of congestion.
This is a real problem. As experience from the United States, Canada, and the United Kingdom has shown, ISPs routinely manage congestion by singling out specific applications or classes of applications when they are not required to manage their networks in an application-agnostic manner. For example, an ISP might—as did the British ISP BT in 2009—throttle streaming video to 896 kilobits per second for users of its “Up to 8 Mbps Option 1” Internet service packet from 5:00 PM to midnight to manage congestion, limiting users’ ability to watch video when most users would like to do so, while allowing the use of other applications that might be equally bandwidth intensive.
Such discriminatory network management practices significantly constrain users’ ability to use the Internet as they like during peak times and make it more difficult for affected applications to reach their users. As online video company Zediva explained to the FCC in 2010,
Discriminatory network management of this type would put the affected applications at a severe disadvantage. Companies that offer these applications and services will be less able to reach their users during times of congestion, which in turn may affect their success in the market (who wants to use an application or service that is less usable during peak time, when most people actually want to use the Internet?) and their ability to get funding—thus squashing innovation before it has had a chance to prove itself in the marketplace.
By contrast, requiring network management to be tailored, appropriate, and as application-agnostic as possible gives network providers the tools they need to manage their networks and maintain a quality experience for all Internet users, while preserving the Internet’s ability to serve as a level playing field and support user choice even during times of congestion. For these reasons, the FCC and, following its example, the Canadian Radio-Television and Telecommunications Commission have long required network management to be as application-agnostic as possible. The Republican bill should do so as well.
4. The bill leaves “user choice” undefined, and this vacuum could be filled by ISPs’ problematic definition of the term.
According to the bill, nothing in it “shall be construed to limit consumers’ choice of service plans or consumers’ control over their chosen broadband Internet access service.” This sounds appealing—after all, preserving users’ ability to choose how they would like to use the Internet is one of the key goals of any network neutrality regime. However, this provision might be used to weaken the protections afforded by other parts of the bill under the guise of “user choice.” There is no explanation of what consumer choice and control mean. Depending on one’s interpretation of the concept, ISPs might be able to offer their services in ways that, while providing the illusion of user choice, still allow them to discriminate or charge access fees.
To better understand how this might work, it’s helpful to unpack the differences between what network neutrality proponents mean when they say “user choice” as compared with how ISPs use the term. The debate over “user-directed prioritization” provides a good example.
Many network neutrality proponents support giving users the ability to choose and pay for different types of Internet service (also called “Quality of Service”) when using different applications on the Internet, but only under three limited conditions. First, in addition to normal Internet service, an ISP is allowed to offer additional types of service (for example, a low-delay service that reduces the amount of time data need to travel across the network or a service that offers a guaranteed amount of bandwidth), but it cannot constrain the applications or classes of applications for which these services can be used. This means that a provider cannot control which applications or classes of applications get faster service. Second, users get to choose whether, when, and for which applications to use which class of service. For example, one person might use the low-delay service for important Internet telephony conversations, while another might use it for online gaming. Thus, people can select exactly the service quality that meets their needs. Third, the provider is allowed to charge only its own broadband Internet subscribers for use of these different services; it cannot charge application and content providers or entities that directly interconnect with the ISP’s last-mile network. This condition prevents the harms arising from access fees described above.
At times, network neutrality proponents have referred to this practice as “user-controlled Quality of Service” or “user-driven prioritization.”
Referring to network neutrality proponents’ support for “user-driven prioritization,” ISPs (most prominently AT&T) have asked the FCC to allow what it calls “user-directed prioritization,” claiming that network neutrality proponents are on board. But while the terms sound similar, they are not. AT&T’s proposal would allow a much broader range of practices than network neutrality advocates would permit. It seems that AT&T views a practice as “user-directed” if it includes some element of user direction or user choice, even if the ISP is engaging in one of the harmful practices described above.
For example, AT&T seems to think an access fee is user-directed if the user allows the ISP to charge an application provider like Netflix for preferential treatment (e.g., for extra bandwidth). According to Brian Fung, who described AT&T’s proposal in the Washington Post, “AT&T’s idea would still allow for commercial deals between companies. But they would have to be arranged as the result of one or more subscriber requests; the ISPs couldn’t offer fee-based prioritization just because they wanted to.”
But users’ consent to access fees does not resolve any of the problems that those fees pose to innovation and free speech. Under this bill, such an interpretation of the concept of “user choice” would allow ISPs to circumvent the bill’s ban on paid prioritization, enabling ISPs to charge application and content providers as long as a user agrees to it.
Similarly, a broad interpretation of “user choice” might allow ISPs to distort competition among applications by allowing users to “choose” among a constrained range of options. For example, an ISP might give users the “choice” to pay for a low-delay service for online gaming, but not for Internet telephony or another service that might also benefit from low delay. Thus, users would be able to “choose” whether they want to take advantage of the ISP’s offer for their online games, but they could not use the low-delay service for any other application that might benefit from such a service. While offering users some “choice,” such practices would still allow ISPs to distort competition by tilting the playing field in favor of certain applications or classes of applications by determining which applications can make use of a better type of service.
Though we don’t know what the bill means by “user choice,” we do know that power abhors a vacuum. Because the bill doesn’t define this term, it is left up for grabs in the midst of an ongoing debate between network neutrality advocates and ISPs. Moreover, the bill exacerbates this power vacuum by stripping the FCC of its authority to clarify the question through rulemaking. The agency would only have the power to interpret this provision after the harmful discrimination has already occurred and some adversely affected party spends the time and money to bring a complaint. All of this makes it less likely that the FCC will arrive at the targeted approach favored by network neutrality proponents and more likely that it will favor the ISPs’ approach.
5. Interconnection is left out of the bill—and can never be addressed.
In its Notice of Proposed Rulemaking in May 2014, the FCC asked for comment on whether protections for interconnection with last-mile networks should be included in the new network neutrality rules. While many network neutrality proponents and representatives of smaller ISPs (e.g., Comptel and the National Telephone Cooperative Association) supported the inclusion, and President Obama encouraged the FCC to investigate the question, the larger ISPs and their lobbying associations (i.e., AT&T, Verizon, Comcast, and the National Cable and Telecommunications Association) vigorously opposed it. The bill effectively sides with large ISPs, opting not to let the FCC oversee or even investigate practices that last-mile ISPs engage in at the point of interconnection with other networks, content delivery networks (CDNs), or large application providers.
Network neutrality rules should address interconnection. As the past few years have shown, ISPs can block, discriminate, or impose access fees either while data are traveling across the ISP’s last-mile access network or when it enters that network at the point of interconnection. For example, instead of slowing down traffic while it travels over the ISP’s access network, the ISP can just allow the point of interconnection to congest. And as the experiences of Netflix and Level 3 illustrate, instead of explicitly charging application and content providers an access fee for prioritized or otherwise enhanced access to its Internet service customers, an ISP can create a situation where the only way for an interconnecting network to receive satisfactory performance at the interconnection point is to pay the last-mile ISP for interconnection.
Although the interference happens at a different point in the network, the impact of blocking, discrimination, or access fees on users and application providers is the same, as is the harm to innovation and free speech. Users don’t care whether the eagerly awaited new season of House of Cards buffers because their video encounters congestion when entering the last-mile network at the point of interconnection or after it has entered that network. Application providers don’t care whether the fee they have to pay to get acceptable quality and remain competitive is for interconnection or for transport across the end users’ access network. Under these circumstances, regulating just one set of practices (i.e., only on the access network or only at the point of interconnection with last-mile networks) will ultimately be ineffective and irrelevant—creating a loophole that allows ISPs to engage in the otherwise banned practices at the unregulated point.
Moreover, the bill would make it impossible for the FCC to regulate the issue of last-mile interconnection at a later stage. It explicitly constrains the FCC’s rulemaking authority with respect to the network neutrality-related provisions, makes it impossible to reclassify ISPs under Title II, and even removes the FCC’s ability to adopt rules based on section 706 of the Telecommunications Act.
6. “Specialized services” are vague and largely unregulated, potentially creating a loophole in the network neutrality rules.
The bill allows ISPs to offer so-called “specialized services,” subject to few constraints. Specialized services are “services other than broadband Internet access service that are offered over the same network as, and that may share network capacity with, broadband Internet access service.” As the bill makes clear, its network neutrality rules should not be construed to limit ISPs’ ability to offer such services as long as they are not “offered or provided in ways that threaten the meaningful availability of broadband Internet access service or that have been devised or promoted in a manner designed to evade the purposes of this section.”
Apart from this expansive definition of specialized services, the bill imposes no limits on an ISP’s ability to offer them.
ISPs have long been pushing for such a broad exception to any network neutrality regime that allows them to offer additional services over a user’s Internet connection and claim that these additional services are not part of their broadband Internet access service and not subject to network neutrality rules. Although the FCC asked for additional comment on specialized services in 2010, it is not clear what kind of specialized services ISPs have in mind or whether these services could be offered over a properly regulated broadband Internet access service.
The vague “specialized services” provision could be the sort of loophole that you could drive a truck through, enabling ISPs to circumvent network neutrality rules. For example, according to Harold Feld, Senior Vice President of Public Knowledge, the bill would allow “Comcast or AT&T or any other provider [to] offer its over-the-top online streaming service as a ‘specialized service’ and give itself prioritized service. Companies could essentially sell prioritized service to specific applications or content simply by calling these fast lanes ‘specialized services.’” One could argue that this kind of service is “designed to evade the purposes of” the bill’s network neutrality rules and would therefore be prohibited by the bill, but it is not clear how this language would be applied. After all, from the perspective of ISPs, being able to offer services that are not subject to network neutrality rules is the whole point of the specialized services exception.
Additionally, exempting specialized services might give ISPs an incentive to limit (or fail to upgrade) the amount of capacity available for normal, regulated broadband Internet access service in order to have more capacity for unregulated specialized services. In the Open Internet Order, the FCC shared this concern, expressed stringent expectations for how it expected ISPs to address them, and committed itself to monitoring the issue, all steps in the right direction. While the bill says that specialized services cannot be offered in a way that “threaten[s] the meaningful availability of broadband Internet access,” this seems to be a lower bar as compared to the FCC’s approach.
What’s more, the FCC is unable to give meaning to the few, weak limits set forth in the bill and must wait on individual complaints before it can address these issues by adjudication. As with “user choice” and interconnection, the bill strips the agency of any power to affirmatively address the issues with “specialized services”—whether now or in the future.
Finally, network neutrality proponents have argued that ISPs might offer specialized services in a way that distorts competition—for example, by offering them exclusively to themselves or their partners, or by charging different prices for the same service. The Open Internet Order acknowledged these concerns, and the FCC committed to monitoring the issue, leaving its resolution to subsequent rulemakings. But under the bill, the FCC would be powerless to address any of these concerns.
7. The bill ties the FCC’s hands—in network neutrality and other emerging broadband telecommunications policies.
The bill drastically limits the FCC’s ability to implement and enforce the network neutrality-related provisions of the bill and strips the FCC of any other source of authority that might allow it to regulate ISPs at a later stage.
In implementing laws, an agency usually has the choice of whether to proceed by rulemaking or case-by-case adjudication. By contrast, the bill strips the FCC of any rulemaking ability related to the provisions of the bill. This is a huge problem. It makes it impossible for the FCC to adopt implementing regulations that give meaning to the more ambiguous provisions of the bill—of which there are quite a few—or to address future network neutrality-related problems as they arise. According to the bill, the FCC can enforce the law only case by case in reaction to complaints and “may not expand” the bill’s network neutrality-related provisions, “whether by rulemaking or otherwise” (e.g., in adjudications). By explicitly restricting enforcement to the adjudication of complaints, the bill also seems to remove the FCC’s ability to investigate violations of the bill on its own motion—an option the FCC specifically mentioned in the Open Internet Order.
Tying the FCC’s rulemaking hands and restricting it to adjudicating complaints creates several problems. First, it creates huge uncertainty, as the FCC has to wait for a complaint in order to address how ambiguous terms will be applied. Second, it tilts the playing field in favor of large, established players like the large ISPs or large applications or content providers that have the resources necessary to engage in protracted and costly FCC proceedings. Finally, case-by-case approaches are less likely to result in decisions that adequately protect the values and actors that network neutrality rules are designed to protect.
Moreover, the bill strips the FCC of any other authority it might use to regulate ISPs. It legally defines “the provision of broadband Internet access service or any other mass market retail service providing advanced telecommunications capability (as defined in section 706 of the Telecommunications Act of 1996)” as an information service, making it impossible for the FCC to reclassify these services as telecommunications services under Title II of the Communications Act.
And the bill goes even further by removing the FCC’s existing authority to regulate broadband providers. So far, the FCC has been regulating ISPs based on its ancillary authority, coupled with section 706 of the Telecommunications Act of 1996. According to the FCC’s interpretation, which was upheld by the Court of Appeals for the D.C. Circuit in Verizon v. FCC, section 706 allows the FCC to adopt measures that foster broadband deployment. While section 706 does not allow the FCC to impose common-carrier-type rules on entities that—like ISPs—have not been classified as telecommunications services under Title II of the Communications Act, it does allow the FCC to adopt other regulations as long as they foster deployment. The bill abolishes that option by inserting a provision into section 706 that prohibits the FCC and “State commission[s] with regulatory jurisdiction over telecommunications services” from relying on section 706 “as a grant of authority.”
These restrictions on the FCC’s authority have ripple effects beyond network neutrality. Most immediately, the bill would make it impossible for the FCC to rely on section 706 to preempt state laws that prohibit cities from building and running city-owned broadband networks, an approach it planned to take and that President Obama endorsed last week.
As Public Knowledge pointed out, removing the FCC’s authority to regulate based on section 706 has other unintended consequences:
This draft would undo nearly five years of work on universal service reform based on Section 706 authority, seriously disrupting efforts to provide broadband to rural areas. It would eliminate the FCC’s authority to preempt limits on community broadband. It could have serious unintended impacts on voice-over-IP services, placing the stability of the 9-1-1 system at risk and interfering with the FCC’s efforts to resolve rural call completion. Among other things, it could also limit the FCC’s authority to promote access by the disabled to communications services, protect consumer privacy, [and] promote broadband deployment by ensuring that new competitors have access to utility poles and rights of way.
Thus, while the Republican bill gives the FCC limited authority to enforce the bill’s network neutrality provisions via adjudication, it takes away any other power to promote competition, public safety, and privacy, making it impossible to protect consumers and the public interest beyond the very narrow protections afforded by the bill.
It is heartening that congressional Republicans are moving away from opposition to network neutrality and towards the consensus in favor of meaningful network neutrality rules. However, the bill as currently written does not go nearly far enough. As this piece illustrates, the bill would require a significant overhaul to ensure that it adequately protects users, innovators, and speakers against blocking, harmful discrimination, and access fees.
The good news is that we don’t need to go back to the drawing board. We can instead direct our attention to the FCC. Next month, the Commission could adopt a proposal that has unprecedented support, including from President Obama: a ban on blocking, application-specific discrimination, and access fees under Title II of the Communications Act. We should urge the FCC to adopt this proposal, the most clear and direct way to establish meaningful and lasting network neutrality protections.
Barbara van Schewick is a Professor of Law and Helen L. Crocker Faculty Scholar at Stanford Law School, the Director of Stanford Law School’s Center for Internet and Society, and a professor (by courtesy) in Stanford University’s Department of Electrical Engineering.
Morgan N. Weiland is a J.D. Candidate at Stanford Law School, Class of 2015, and a Ph.D. candidate in Stanford University's Department of Communication. She is also a Student Fellow at the Center for Internet and Society, a Bradley Student Fellow at the Stanford Constitutional Law Center, and a Stanford Graduate Fellow at Stanford University.