NEW MODELS AND CONFLICTS IN THE INTERCONNECTION AND DELIVERY OF INTERNET-MEDIATED CONTENT ABSTRACT

Transcription

1 NEW MODELS AND CONFLICTS IN THE INTERCONNECTION AND DELIVERY OF INTERNET-MEDIATED CONTENT ABSTRACT As the Internet has evolved and diversified, interconnection terms and conditions have changed between Internet Service Providers ( ISPs ). These carriers experiment with alternatives to conventional models that classify interconnection as either peering or transiting. The former typically involves interconnection between high capacity carriers whose transoceanic traffic volumes generally match thereby eliminating the need for a transfer of funds. Historically smaller carriers have paid transit fees to larger Tier-1 ISPs for the opportunity to secure upstream links throughout the Internet cloud. With the growing availability of bandwidth intensive, full motion video content carried via the Internet, traffic volume disparities have increased between ISPs. A new category of ISP has targeted the downstream video content delivery market, all but guaranteeing that these Content Delivery Networks ( CDNs ) will have more traffic for which they need to secure delivery to end users than what retail ISPs can or will hand off to them for upstream delivery. Such asymmetry in traffic flows traditionally has forced CDNs to become transit payers even if previously they qualified for zero cost peering. The migration from CDN peer to transit payer represents one of many adjustments in interconnection compensation arrangements triggered by changes in traffic flows. Heretofore commercially driven negotiations have managed the transition without resulting in many service disruptions. However it appears increasingly likely that interconnection negotiations will become more contentious and protracted, particularly when retail ISPs demand compensation from sources of high volume, bandwidth intensive video content even though the ventures do not interconnect directly. Traditional peering and transiting occurs between directly interconnecting carriers, but some retail ISPs believe they should receive compensation from content sources

2 2 like Netflix and YouTube, because of the downstream torrent of traffic these ventures generate for final delivery. Content providers have balked at making such payments, because they already pay CDNs and other carriers with which they directly interconnect. Additionally retail ISPs charge their customers often hefty broadband subscription rates that appear sufficiently compensatory for their costs incurred in providing access throughout the Internet cloud. This paper will examine existing and likely future interconnection disputes with an eye toward identifying where conflicts will arise and how they can get resolved. The paper supports commercially driven negotiations, but suggests that National Regulatory Authorities may need to arbitrate and promote settlements when now essential Internet access becomes blocked or congested. The paper concludes that ISPs should have the opportunity to provide both end users and content sources alternatives to best efforts content delivery, but that they should not create artificial congestion as justification for additional compensation. Keywords: interconnection, Internet, peering, transiting, convergence I. INTRODUCTION Forecasts of media convergence have become a reality as the Internet increasingly provides a single medium for the delivery of information, communication and entertainment ( ICE ). 1 As media technologies merge, previously stand alone channels of distribution begin to combine making it questionable whether existing business models and regulatory regimes will persist. 2 Already trend setting consumers have begun to abandon pay television subscriptions as they cut the cord 3 and access content on their own terms. These early adopters of new technologies and self-help strategies consider anathema the traditional concept of appointment television 4 where viewers access content at a specific time, on a particular channel and via a

3 3 single display technology. One can anticipate a future when most content reaches consumers via on-demand options instead of the traditional live, broadcast model. Early adopters of new media expect to have access to content anytime, anywhere, via any device and in any presentation format. Many consumers have no patience with business models that ration content access through a series of display windows that match availability with willingness to pay. For example, movie access traditionally has run a time sequence starting with theatrical presentation and followed by pay per view, DVD sale, premium cable and satellite channel access, DVD rental, broadband download, etc. Many consumers now resort to self-help strategies that achieve access to pirated premium content via multiple screens soon after initial release. Television sets, computer monitors, smartphone screens and tablets offer much of the content previously made available exclusively via the movie screen, or to only one of the many other mediated screen options. New media options and convergence have the potential to disrupt the business plans of incumbents, but also established legal, regulatory and policy assumptions. However if finding alternative ways to access desired content results in abandonment of existing content distribution options by many consumers, content providers and distributors may slow down or end their experimentation with new distribution technologies 5 even as they embrace new and effective digital rights management ( DRM ) techniques to prevent piracy and even fair use. 6 In any event, the proliferation of new media access options will require much more coordination between and among the providers of the broadband pathways, commonly referred to as Internet Service Providers ( ISPs ) and the providers and distributors of content. As the Internet has evolved and diversified, interconnection terms and conditions have changed between ISPs as they explore alternatives to conventional models that classify interconnection as either peering 7 or transiting. 8 The former typically involves interconnection between high capacity carriers whose transoceanic and transcontinental traffic volumes

4 4 generally match thereby eliminating the need for a transfer of funds. Historically smaller carriers have paid transit fees to larger Tier-1 ISPs for the opportunity to secure upstream links throughout the Internet cloud. 9 With the growing availability of bandwidth intensive, full motion video content carried via the Internet, traffic volume disparities have increased between ISPs. A new category of ISP has targeted the downstream video content delivery market, all but guaranteeing that these Content Delivery Networks ( CDNs ) will have more traffic for which they need to secure delivery to end users than what retail ISPs can or will hand off to them for upstream delivery. Such asymmetry in traffic flows can generate interconnection compensation disputes as occurred between the major CDN for Netflix content, Level3, and a major ISP, Comcast, which provides last mile, retail delivery of Internet content. 10 CDNs typically become transit payers even if previously they qualified for zero cost peering, but questions remain whether retail ISPs, such as Comcast, have an affirmative duty to try offsetting traffic imbalances. Likewise consumers wonder what service commitments a retail ISP should make in exchange for sizeable monthly Internet access subscription revenues. The migration from CDN peer to transit payer represents one of many adjustments in interconnection compensation arrangements triggered by changes in traffic flows. 11 Heretofore commercially driven negotiations have managed the transition without resulting in many service disruptions. However it appears increasingly likely that interconnection negotiations will become more contentious and protracted, 12 particularly when retail ISPs demand compensation from sources of high volume, bandwidth intensive video content even though such ventures do not interconnect directly. Traditional peering and transiting occurs between directly interconnecting carriers, but some retail ISPs believe they should receive compensation from content sources like Google, Netflix and YouTube, 13 because of the downstream torrent of traffic these ventures generate for final delivery. 14

5 5 Content providers have balked at making such payments, because they already pay CDNs and other carriers with which they directly interconnect. Additionally retail ISPs charge their customers broadband subscription rates that appear sufficiently compensatory for their costs incurred in providing content access throughout the Internet cloud. This paper will examine existing and likely future interconnection disputes with an eye toward identifying where conflicts will arise and whether commercial negotiations can reach closure on a timely basis. Additionally the paper will examine Internet interconnection models with an eye toward assessing whether National Regulatory Authorities ( NRAs ) will need to arbitrate and promote settlements when access via the Internet and other media, such as cable television, becomes blocked or congested. The paper will consider whether ISPs should have the opportunity to provide both end users and content sources with alternatives to traditional best efforts content delivery, 15 including better than best effort premium service. Additionally the paper will assess whether ISPs can engage in quality of service discrimination without creating artificial congestion as justification for additional compensation. This research expands on the considerable discussion about whether governments need to safeguard neutral and open access to the Internet. Much of the existing research 16 on network neutrality 17 lacks empirical data and fails to examine the types of available ISP network interconnection options. The proposed research will integrate the mechanics of Internet interconnection with an examination of peering and transiting contracts not subject to nondisclosure agreements. 18 Research on the feasibility of Internet connectivity and blockage will provide empirical evidence that can help answer questions whether and how governments need to enforce neutrality requirements. II. DECONSTRUCTING THE INTERNET CLOUD Explanations about the interworking of the Internet often use cloud analogies to emphasize the apparent ease with which networks interconnect to form a complete end-to-end

6 6 link 19 However, when one examines the actual means by which traffic arrives at its final destination, the Internet constitutes a complex array of facilities operated by different carriers using many types of equipment manufactured by a variety of companies over several generations of innovation. The Internet operates as a network of networks 20 thanks to the ability of shared operating standards and protocols to make compatible the networks of many carriers using equipment of many manufacturers. The network of networks and cloud analogies also ignore the complex and potentially contentious matter of financial compensation when a cooperative carrier agrees to route traffic of another carrier onward to its final destination, or to another carrier. Internet carriers initially could ignore questions about traffic flow and financial responsibility, because governments subsidized network rollouts. During phases 21 of government incubation and anchor tenancy, Internet carriers did not need to meter traffic flows and determine whether compensation should flow from one carrier to another in light of traffic imbalances. The network of networks cooperative model started with interconnection based on a rough justice barter system called peering where carriers agreed to eschew cash settlements on the assumption of a general balance of traffic flows, or the view that metering traffic would prove too costly. Interconnection based on assumed parity of traffic volume and the absence of a need to transfer funds sharply contrasts the model used by telephone companies. These carriers have never ignored the matter of financial compensation arrangements even when they received taxpayer subsidies to promote universal service and infrastructure expansion into the hinterland. From the onset of service financial compensation models for telephone service carrier interconnection have relied on traffic metering and a negotiated financial settlement. 22 Over time, as governments have reduced or eliminate subsidies, Internet carriers have recognized the importance of measuring traffic flows and using financial settlements when traffic volumes lack symmetry. However the migration from peering to payment based transiting has

7 7 not always occurred smoothly, particularly when commercially driven terms impose new financial obligations on some carriers that previously used the zero payment barter process. In turn these carriers have sought to recoup these costs from end users, particularly the highest volume subscribers. Internet carriers typically offered unmetered, all you can eat subscription model in the early phases of development and promotion. Now they consider or already have migrated to service tiers that place caps on the volume of traffic a subscriber can consume, or slows down transmission delivery speed ( throttling ) after a downloading threshold has occurred within one month. 23 While arguably more efficient and fair, new metered retail subscription models have triggered much consumer opposition and even assertions that tiering discriminates and reduces the value of a subscription. 24 The need for Internet carriers to pay attention to traffic flows and the cost of providing peering and transit services to other carriers evidences the importance of network interconnection and perhaps as well its vulnerability. A carrier dissatisfied with the status quo will seek new and more favorable commercial terms to which other carriers may not readily agree. If negotiations reach an impasse the carriers at least temporarily will no longer interconnect and accept traffic from each other. Such de-peering typically can occur without service disruption, because alternative routing arrangements exists with other carriers. 25 However the viability of the alternative carrier option depends on where in the cloud the network disconnection occurs. The Internet ecosystem operates with highly varying degrees of competition and alternative routing options. Content providers and distributors generally have many options for securing the long haul carriage of traffic. So-called Tier-1 ISPs offer redundant, duplicative and low cost options for transcontinental and transoceanic carriage. So even if a major Tier-1 ISP decided not to carry the traffic of another ISP, on financial or other grounds, the disconnected content provider/distributor could readily find alternative routes and carriers.

8 8 First and last mile retail access presents a different picture. End users may have a limited number of ISP service options for content uploading and downloading. Typically the incumbent telephone company provides a Digital Subscriber Line ( DSL ) option, the cable television company provides a faster and more expensive broadband alternative and one or two satellite carriers provide a comparatively more expensive and slower speed delivery option possibly most attractive to users lacking other choices. Terrestrial wireless carriers have begun to offer a competitive option, albeit one typically already imposing content downloading caps and raising questions about their ability to maintain broadband speeds during peak demand conditions. 26 Regardless of which technological option end users choose in terms of actual service most consumers select one and only one carrier to handle all of their Internet traffic requirements. Should a service disruption occur upstream almost all ISPs can secure interconnection options almost immediately. But at the retail sector, even consumers with competitive options will encounter some delay and expense in migrating from one carrier to another. In light of the possibly limited competitive options available to retail Internet access subscribers and their sole reliance on one carrier, the chosen ISP has significant negotiating power with both end users and upstream ISPs. End users may balk at the inconvenience of changing carriers and upstream ISPs will have no migration option at all if they want to secure access to all end users. Put another way, if a single ISP enjoys a dominant market share of the retail market, which occurs in many localities a substantial portion of the market exclusively relies on that single ISP making it absolutely necessary for upstream ISPs to secure an agreement with that ISP for its delivery of content. A single ISP has the potential to exert exclusive control over access to a majority of the end user market in many places. Content providers and distributors are captive to that ISP in the sense that they must secure delivery to

9 9 the televisions, computer monitors and smartphones and tablets that access the Internet solely via a single ISP. 27 III. THE IMPACT OF FULL MOTION VIDEO ACCESS The consequences of retail ISP market dominance have become more important as consumers increasingly rely on the Internet as a one stop shop for access to all forms of ICE. With the wider availability of broadband network access, consumers now can use the Internet as a medium for the delivery of full motion video like that previously available only from broadcast, cable and satellite television. Some consumers have cut the cord by abandoning legacy services and opting for access to video content exclusively via the Internet. An ever increasing inventory of choices has become available to alternative video display screens. The terms Internet Protocol Television ( IPTV ) 28 and Over-the-Top Television ( OTT ) 29 refer to the ability of content creators and new or existing content distributors to provide consumers with access to video content via broadband links, in lieu of, or in addition to traditional media. IPTV options, which include such popular sites as Netflix, Hulu and YouTube, offer consumers greater flexibility in accessing content. Viewers no longer need to tolerate appointment television 30 access to content at a time prescribed by content creators or distributors and available only on a single broadcast, satellite, or cable channel. Access is becoming a matter of using one of several software-configured interface options capable of delivering live and recorded content anytime, anywhere, to any device and via many different transmission and presentation formats. But as empowering as these technological options appeal, they rely on a very small number of options for the final delivery to consumers. Cable modem service providers, DSL carriers, and wireless operators provide the essential last mile delivery of content and the ever increasing volume of downstream traffic generated by IPTV demand provides these carriers with increasing leverage in interconnection negotiations.

10 10 One cannot yet conclude that retail ISPs regularly abuse their control of a last mile bottleneck, but the growing number of interconnection disputes between ISPs as well as other ICE ventures should trigger concern. Because more traffic arrives at retail ISP switching facilities, these carriers enjoy opportunities to demand compensation even from other ISPs with which they previously had a zero cost peering arrangement. The possibility exists that retail ISPs will negotiate fairly and not act on their incentives and ability to meddle with downstream traffic flows to secure greater leverage in interconnection negotiations. However one also can anticipate instances where a retail ISP does not refrain from exploiting the opportunity to demand more than a fair share. Should this scenario play out retail ISPs all but invite scrutiny by NRAs, including boards with authority to implement competition policy and sanction unfair trade practices. In nations where ISPs have qualified for little regulation, abuse of their retail bottleneck may trigger legislatures to reclassify Internet access so that some public utility, essential facility or common carrier responsibilities apply. 3.1 TREATMENT OF CONTENT DELIVERY NETWORKS The most contentious recent interconnection disputes among ISPs has involved retail ISPs and Content Delivery Networks ( CDNs ), a new category identifying ventures that concentrate on providing downstream delivery of full motion video. 31 By emphasizing downstream services, CDNs all but guarantee traffic imbalances with interconnecting carriers, especially retail ISPs. In late 2010 Comcast, a major retail ISP in the United States, sought to impose a surcharge on traffic volumes generated by Level 3, a Tier-1 ISP and CDN in light of a significant increase in downstream traffic generated by Level 3 after it secured the opportunity to serve as the primary carrier for delivering Netflix movies and television programs content to subscribers. Level 3 sought regulatory relief at the Federal Communications Commission ( FCC ) and also launched a public relations campaign to frame the dispute in terms of Comcast imposing a

11 11 toll booth on the Internet and singling out Level 3 and Netflix traffic for a surcharge to raise the cost of a major alternative to Comcast s pay per view movie services. 32 Comcast responded with an equally forceful campaign to explain that the dispute simply addressed a commercial peering matter. 33 Comcast claimed that Level 3 s increased traffic triggered the right to demand more compensation in light of the higher volume of traffic Comcast delivered to its subscribers. 34 This dispute provides a high profile example of a how ISPs now have keen interest in the financial terms and conditions under which they interconnect with other carriers and sources of content. Comcast correctly stated that Level 3 and it had executed a peering agreement for reciprocal and zero cost treatment of traffic, provided the flows remain nearly symmetrical. Because Level 3 began to generate far more downstream traffic for Comcast to deliver than Comcast hands off to Level 3 for upstream transmission, the typical peering agreement would require Level 3 to compensate Comcast if the traffic flows cannot return to near parity. Unless the parties could find a way for Level 3 to receive more traffic from Comcast, Level 3 contractually bore a legal obligation to compensate Comcast. On the other hand, Level 3 correctly stated that the peering agreement it has negotiated with Comcast covered both downstream traffic from Level 3 to Comcast as well as upstream traffic from Comcast to Level 3. It remains unclear whether Comcast could have and should have taken affirmative steps to offset the traffic imbalance, perhaps by using Level 3 s Tier-1 ISP network facilities for more long haul traffic routing. Additionally Comcast risked inconveniencing its subscribers with the prospect of congestion, degraded quality of service and even dropped packets resulting in outages. Bear in mind that during the dispute Comcast continued to receive subscription payments even though its dispute with Level 3 could have resulted in congestion that would have caused Netflix picture quality to deteriorate.

12 12 Comcast generates significant profits from its retail cable modem service subscriptions 35 the terms of which authorize the company to manage its network in ways it see fit, but arguably does not reserve the right to block, degrade or conditionally deliver traffic. Likewise broadband subscription agreements do not specify that service is contingent on the retail ISP receiving payments from content sources and upstream carrier who agree to pay a surcharge when an interconnection dispute arises. Most Comcast subscribers had no understanding about the potential for degraded Netflix service. Had this outcome occurred consumers most probably would not have blamed Comcast, or understood that this carrier could have resolved the problem, but did not do so to generate greater negotiating leverage with Level 3. The incentive for Comcast to inconvenience its subscribers to discipline a peering partner parallels what happens when television broadcasters cannot reach closure with cable television operators on the terms and compensation for cable delivery of local broadcast stations. Such retransmission consent negotiations sometimes fail to reach closure before the cable company has to stop carriage. Consumer anger at denied access to must see television, such as live sports, typically forces cable operators to capitulate, but content providers, such as CBS and Fox have identified and used new Internet access denial strategies to secure even greater negotiating leverage. The companies used techniques to identify the Internet Protocol addressed used by broadband subscribers of the cable companies with which they had a retransmission dispute. By identifying these subscribers identities, the companies succeeded in blocking cable broadband subscribers access to video content available at the CBS web site and at the Hulu content aggregator web site. These content creators had a perverse incentive to deny access to eager viewers despite the reduction in audience ratings and its commensurate impact on advertising revenues. The companies understood that they had more to gain from higher cable

13 13 television operator retransmission fees and willingly used Internet access blocking techniques to secure even more negotiating leverage. The ability of CBS and Fox to block access to content far away from retail ISP facilities identifies a new location where carrier interconnection can dispute can arise and frustrate consumers. Much of the debate about network neutrality has focused on the incentive and ability of retail ISPs to operate in discriminatory ways that could favor corporate affiliates and other content providers and distributors willing to pay a surcharge for preferential delivery services. By blocking access to content far upstream at the source, or between the source and a content aggregator, such as Hulu, CBS and Fox have shown how selective blocking of another type of network interconnection in the Internet cloud can occur much to the dismay and displeasure of consumers who find themselves denied access to content based on who provides their broadband access to the Internet cloud. 4. LIKELY CHANGES IN CLOUD INTERCONNECTION ARRANGEMENTS ISPs already have used commercial negotiations to structure alternatives and adjustments to the traditional dichotomy of barter (peering) or payment (transiting). 36 Such diversification in interconnection compensation largely results from the proliferation of ventures requiring access as well as broadening differences in the number of subscribers, points of interconnection, available transmission capacity, portion of the total traffic carried representing video and the volume of traffic received from and handed off to particular ISPs. The structure and bargaining power of Internet carriers has become somewhat less hierarchical, particularly for ISPs that do not qualify for Tier-1 status. These ISPs may handle significant traffic, but the volume may be limited to retail services in specific localities, or concentrated in specific geographical regions. Many carriers that do not qualify as Tier-1 ISPs have agreed to interconnect with other similarly situated operators, often at mutually convenient locations where many carriers and even content providers locate equipment. Such

14 14 interconnection supports the creation of a new or increased set of Tier-2 carriers and the use of Internet Exchange Points 37 serving as a mutually convenient and centralized interconnection location. 38 ISPs also have increased the number of carriers with which they interconnect. This process, commonly referred to as multihoming, can increase the speed of content delivery and achieve more reliable service. ISPs, which do not qualify for peering now can secure most of the benefits of direct interconnection with major carriers, by paying them. This paid peering process 39 differs from transiting, because the paying ISP does not simply select, interconnect with and pay one Tier-1 ISP for complete access to the Internet cloud. Instead an ISP might select several carriers, not limited to Tier-1 ISPs, to handle a portion of the total access requirement. ISPs that opt for paid peering may operate a significant network of their own, but find it necessary to secure more transmission and switching capacity at locations where they do not operate. For example, it appears that the onslaught of Netflix downstream traffic has forced Level 3 into a paid peering relationship with retail ISPs, such as Comcast, because Level 3 lacks network capacity to reach endusers and existing zero payment peering arrangements required near parity of traffic flows. 40 CDNs by nature of their business plan have become paid peering customers. With large volumes of traffic that must reach very many end users, CDNs cannot qualify as zero cost peers unless the retail ISP uses an affiliate of the CDN for long haul, upstream delivery services. The CDN-retail ISP commercial relationship typically involves asymmetrical traffic volumes, because end users typically download more content than they upload and CDNs, operating upstream from the retail ISP have growing volumes of traffic needing downstream delivery. The CDN and its upstream sources of content may object to a payment obligation in addition to the sizeable Internet access charges paid by the retail ISPs subscribers. Arguably a retail ISP subscription includes a representation that the carrier will provide Internet access

15 15 service at a particular delivery speed. If subscribers make increased use of a resource the carrier would prefer they not actually use, then the remedy lies in download caps and/or higher service fees, rather than demands for compensation from upstream carriers. The alternative view favoring retail ISPs narrows scrutiny to the commercial relationship between the upstream venture and applies the peering/transit dichotomy, or a recent alternative such as paid peering. CDNs and other upstream carriers also will dispute economic rationales that support double payments to retail ISPs based simply on the premise that they operate in a two-sided market: 1) the retail, Internet access service provided to end users and 2) the wholesale, downstream delivery service provided upstream ISPs and their content produce/distributor clients. 41 Regardless whether ISPs can invoke economic rationales, bottleneck control of access to very many eyeballs puts them in a superior negotiating posture with upstream ventures. Recently some retail ISPs have proposed to content providers direct payment models in exchange for superior access to end users, or an agreement not to debit their traffic from end user monthly downloading caps CONTENT PROVIDER/DISTRIBUTOR SELF-HELP Facing the growing likelihood that retail ISPs will demand compensation above that provided by Internet access subscriptions, content providers and distributors have pursued alternatives to using CDNs and paid peering. Ventures like Netflix understand that they need to ensure high quality video reception. They strive to reduce delivery costs, but also reduce the distance between the origination point for content delivery and its consumption by subscribers. For example, Netflix has proposed that ISPs partner with it in an Open Connect Network. 43 In application this program appears to convert Netflix into an ISP with one downstream customer, itself, but also with the ability to provide upstream service exceeding 2 Gigabits per second. Netflix also proposes to install equipment on retail ISP premises that will store the most popular

16 16 content thereby eliminating the need to use a cascade of many ISPs providing a portion of the route used to download the content. 44 Netflix promotes new interconnection arrangements as a way to reduce burdens on retail ISP carriers, including fewer subscriber service complaints and the need to increase bandwidth to handle the downstream torrent of full motion video content. However Netflix does not appear willing to pay for peering, despite the fact that direct interconnection with retail ISPs probably will reduce Netflix s total content delivery costs and reduce paid peering revenues generated by Netflix s CDNs. As the Open Connect Network grows, Netflix s need to rely on CDN services declines as the company installs or leases transmission capacity for direct interconnection with retail ISPs. 5. THE SHORT AND LONG TERM In the short term ISPs will continue to negotiate customized interconnection arrangements when traditional and conventional models do not suffice. Such flexibility can accommodate specialized requirements of content creators, distributors and consumers. However they also can impose anticompetitive and unfair trade practices that both consumers and regulators may not readily undercover. 45 Examples of the former include accommodation of consumers desire for better than best efforts traffic routing of mission critical bits. Video gamers and viewers of high definition video programming want heightened quality of service and may willingly pay for guaranteed premium service. Such performance enhancements should not trigger network neutrality concerns or regulations provided ISPs offer end users options on a voluntary basis without intentionally degrading standard service. Prioritized delivery options for content creators and distributors also may satisfy specialized requirements in much the same ways as CDNs offer higher quality of service and delivery guarantees. However the distinct possibility exists that premium or specialized service

17 17 options can cross the line and constitute unreasonable discrimination. Examples of such anticompetitive and unreasonable trade practices include deliberately dropping packets and other forms of standard service degradation with an eye toward forcing migration to more expensive premium service options. ISPs offering content access to their subscribers necessitate close scrutiny, because they have both the ability and incentive to favor their content options. For example, an ISP could impose a severe cap on downloading volume, but offer not to meter traffic that that routes content to a specific gaming platform, e.g., Xbox consoles. Comcast and other ISPs have begun offering such options, based on the questionable claim that such traffic routing traverses a specialized network and not the general Internet cloud. Such rationale works to exploit a loop hole created by the FCC that allows discriminatory exceptions to the general requirement of open and neutral access when an ISP engineers a specialized network unlike its conventional Internet routing. 46 Another potential anticompetitive practice lies in a deliberate strategy of generating congestion for the traffic generated by a single venture. Retail ISPs can identify and target a particular carrier or source of content for deliberately degraded service. For example, if a retail ISP wanted to force an upstream carrier or content source to pay a surcharge on grounds that they have increased their carriage demands, the retail ISP could reduce or refuse to expand the inventory of available capacity available for the delivery of all or some types of traffic generated by the targeted company. A number of content creators and distributors have claimed that retail ISPs have generated artificial congestion, because the retail ISP has ample downstream capacity, but has assigned specific channels and ports of capacity in a manner that creates congestion for a subset of all downstream traffic. In a nutshell upstream carriers and content creators/distributors have claimed that they encounter degraded service based on the saturation

18 18 of the capacity made available to them even as other ventures, including corporate affiliates, partners and surcharge payers, encounter no such problem. 47 In both the short and long run, parties will dispute whether a retail ISP truly has incurred network congestion as a result of increased downstream volume, or as a result of clever network management designed to create congestion for some, but not all users. More broadly parties will dispute what constitutes reasonable and necessary network management. While a retail ISP surely owns its property and can manage it in any way it sees fit, the carrier can use the necessity of network management as justification for anticompetitive and unreasonable trade practices. The history of telecommunications carrier interconnection contains very many instances where parties deliberately refused to cooperate, and instead pursued many strategies to handicap competitors through outright refusals to deal, inferior and more expensive arrangements, delays and time consuming litigation. 48 Bear in mind that carriers could frustrate competition, despite having a legal requirement to deal and interconnect as common carriers, subject to unquestionable regulatory oversight. In the Internet ecosystem, most nations do not regulate ISPs, or apply a far less burdensome regulatory regime than that applied to telephone companies. If clever telecommunications service providers could avoid and evade regulatory mandates to interconnect fully and fairly, unregulated or lightly regulated ventures can do so as well. Marketplace driven self-discipline may not guarantee that every interconnection negotiation will reach closure on a timely basis with reasonable terms and conditions, particularly where one venture enjoys disproportionately greater negotiating leverage. In both the short and long run ISPs with market power need to curb their incentives to gouge and discriminate excessively. Even through discriminatory tactics can be masked and not easily identified, over time consumers may understand the cause of service degradation and

19 19 punish the culprit. If retail ISPs push too far to maximize revenues, they risk upsetting the balance needed when pricing in a two-sided market as well as losing their unregulated status. Despite a preference for unregulated, commercial negotiations, carriers long abusing market power risk punishment in the court of public opinion and possibly in the marketplace. The possibility exists that in the longer term every leg in the Internet cloud will become sufficiently competitive, or at least one or more ISPs will emphasize neutrality and fair dealing as a market differentiating feature. Governments should defer to commercially driven interconnection arrangements, but stand ready to resolve disputes that become intractable and harmful to consumers. ENDNOTES 1 [O]ne of the most consistent drivers of profound change in the area of cyberspace law has been the convergence of cloud-based products and services with the distribution and consumption of entertainment content. Specifically, the public s relentless desire for access to and consumption of any content, anywhere, at any time, coupled with the widespread proliferation of Internet-connected consumer devices--from the mobile phone, to the tablet, to the Smart TV--has played a substantial role in the marked growth of novel cloud-based products and services. Daniel Schnapp, Legal Implications of Cloud-Based Distribution and Consumption of Entertainment Content, Aspatore, 2013 WL (Aug. 1, 2013). 2 The [Internet] cloud s appeal is its flexibility. Using the government as a source of norms in the early stage of the technology s development will not only moot this flexibility, but will also have the effect of locking the technology in at its current state. A government-centric regulatory model in cloud computing is inadvisable, as government is not sufficiently responsive to keep pace with the rapid manner in which technology evolves. Consumers will be the ones to shoulder the burdens of the cost of regulations made in ignorance of the technology s full potential. Carol M. Celestine, Cloudy Skies, Bright Futures? In Defense of a Private Regulatory Scheme for Policing Cloud Computing, 2013 U. Ill. J.L. Tech. & Pol y 141, 158. With the convergence of these various technologies--for instance, Voice Over Internet Protocol competing with circuit-switched telephony or Internet Protocol Television competing with broadcast and cable [medium specific]... silo approach to regulation makes less sense today. Hon. Maureen K. Ohlhausena, Net Neutrality vs. Net Reality: Why an Evidence-Based Approach to Enforcement, and Not More Regulation, Could Protect Innovation on the Web, 14 Engage: J. Federalist Soc'y Prac. Groups 81 (Feb., 2013) (unpaginated). 3 Incumbent multichannel video program distributors ( MVPD ) benefit from the current concentrated market larding them with high profits while they fail to compete or innovate. They have long feared some users would cut the cord on pay-television by canceling their MVPD subscription, deciding, instead to choose a potentially competitive virtual MVPD providing video content through the Internet. Marvin Ammori, Copyright s Latest Communications Policy: Content-Lock-Out and Compulsory Licensing For Internet Television, 18 Commlaw Conspectus 375, 378 (2010)[hereinafter cited as Ammori]. 4 A secular trend toward narrowcasting has intensified on the web, as more individuals forsake appointment television for the long tail of online content. Frank Pasquale, Beyond Innovation and Competition: The Need for Qualified Transparency in Internet Intermediaries, 104 Nw. U. L. Rev. 105, 110 (Winter, 2010).

20 20 5 Many programmers--including both broadcast programmers and non-broadcast programmers--have increasingly begun to circle the wagons with incumbent MVPDs, concluding that they too are better off with a cut of the MVPDs supra-competitive profits than with the potential wild-west competition enabled by the Internet. Ammori at Fair use refers to the lawful, but unapproved use of copyrighted content. In the balance between rewarding innovation and promoting access to content, fair use provides opportunities for limited and conditional access to content without securing and paying for a license when such use serves the public interest and causes little or no financial harm to the creator. See Anjanette H. Raymond, Heavyweight Bots in the Clouds: The Wrong Incentives and Poorly Crafted Balances that Lead to the Blocking of Information Online, 11 Nw. J. Tech. & Intell. Prop. 473(Aug. 2013); Phil Weiser and Gideon Parchomovsky, Beyond Fair Use, 96 Cornell L. Rev. 91 (2011). 7 Peering refers to a barter arrangement for traffic exchange where two Internet Service Providers agree to accept traffic from the other on without the transfer of funds. The carriers agree to a settlement-free arrangement, because traffic volumes generally match. 8 Transiting refers to an exchange of traffic that triggers a financial settlement and transfer of funds. This arrangement typically results when a small carrier needs the services of a larger carrier to reach all Internet carriers and end users. 9 The Internet cloud refers to the vast array of interconnected networks that make up the Internet and provide users with seamless connectivity to these networks and the content available via these networks. The increasing functionality of the Internet is decreasing the role of the personal computer. This shift is being led by the growth of cloud computing --the ability to run applications and store data on a service provider s computers over the Internet, rather than on a person's desktop computer. William Jeremy Robison, Free at What Cost?: Cloud Computing Privacy Under The Stored Communications Act, 98 Geo. L.J. 1195, 1199 (April, 2010). 10 See. e.g., Daniel L. Brenner and Winston Maxwell, The Network Neutrality and the Netflix Dispute: Upcoming Challenges for Content Providers in Europe and the United States, 23 Intell. Prop. & Tech. L.J. 3 (March 2011); Rob Frieden, Rationales For and Against Regulatory Involvement in Resolving Internet Interconnection Disputes, 14 Yale J. L & Tech. 266 (2012). 11 For background on peering, transit and new interconnection arrangements, see Dennis Weller and Bill Woodcock, Internet Traffic Exchange, OECD Digital Economy Papers No. 207 (Jan. 29, 2013); available at: Ana-Maria Kovacs, Internet Peering and Transit (April 4, 2012); available at: Dr. Peering International; available at: 12 By regulating the terms upon which content providers use their networks to reach consumers, broadband providers could manipulate the flow of information in society. For example, Comcast could conceivably block consumer access to websites like that criticize the company. Perhaps more realistically, Comcast could block or degrade content and applications like Netflix that compete against its other revenuegenerating services. Unlike America Online and other first-generation dial-up Internet access providers, most broadband providers do not specialize in providing Internet access alone. Rather, the largest broadband providers are cable and telephone companies, which have incentives to prevent customers from using their broadband connections in ways that threaten their other revenue streams. For example, consumer groups have expressed concerns that broadband Internet providers that also offer on-demand movie rentals via cable might discriminate against other services (such as Netflix or BitTorrent) that make movies available over a broadband connection. Daniel A. Lyons, Net Neutrality and Nondiscrimination Norms in Telecommunications, 54 Ariz. L. Rev. 1029, 1034 (Winter, 2012). 13 Several cable executives, including Liberty Media Corp....Chairman John Malone, have expressed interest in charging content providers like Netflix Inc.... for the capacity they consume when sending traffic over their networks to subscribers homes. Mr. Malone, who led Liberty to take a big stake in cable operator Charter Communications Inc. earlier this year, said in June that, Reed has to bear in his economic model some of the cost of the capacity that he's burning.

21 21 He was referring to Reed Hastings, chief executive of Netflix, whose streaming video service has accounted for around a third of total North American Internet traffic every night. Ryan Knutson and Shalini Ramachandran, Verizon Fights FCC on Web Rule, The Wall Street Journal (Sep. 8, 2013); available at: 14 Heretofore content providers have refused to provide compensation, but exceptions have begun to arise. AFP, Orange forces Google to pay for mobile traffic (Jan. 16, 2013): available at: 15 The Internet developed initially as an academic curiosity, based on a commitment to the end-to-end principle. This principle requires that all Internet traffic, whether an , a Voice over Internet Protocol (VoIP) call or a video stream, be treated equally and managed through best efforts connections. In such a network, data packets pass from one router to another without the prioritization of any particular packets. In practice, this means that Internet traffic reaches its destination at varying times, depending on the traffic levels of the relevant Internet communications links. Philip J. Weiser, The Next Frontier for Network Neutrality, 60 Admin. L. Rev. 273, (2008). 16 See, e.g., Amanda Leese, Net Transparency: Post-Comcast FCC Authority to Enforce Disclosure Requirements Critical to Preserving The Open Internet, 11 Nw. J. Tech. & Intell. Prop. 81(Jan. 2013); Adam Candeub and Daniel McCartney, Law and the Open Internet, 64 Fed. Comm. L.J. 493 (May, 2012); Dirk Grunwald, The Internet Ecosystem: The Potential for Discrimination, 63 Fed. Comm. L.J. 411 (March, 2011); Rob Frieden, Assessing the Merits of Network Neutrality Obligations at Low, Medium and High Network Layers, 115 Pa. St. L. Rev., No. 1, (Summer, 2010); Marvin Ammori, Beyond Content Neutrality: Understanding Content-Based Promotion of Democratic Speech, 61 Fed. Comm. L.J. 273 (March 2009); Sascha D. Meinrath & Victor W. Pickard, Transcending Net Neutrality: Ten Steps Toward an Open Internet, 12 J. Internet L., No. 6, 1 (Dec. 2008); Tim Wu and Christopher S. Yoo, Keeping the Internet Neutral? Tim Wu and Christopher Yoo Debate, 59 Fed. Comm. L.J. 575 (June 2007); Tim Wu, Network Neutrality, Broadband Discrimination, 2 J. Telecomm. and High Tech L. 141 (2005); Christopher S. Yoo, Would Mandating Broadband Network Neutrality Help or Hurt Competition? A Comment on the End-to-End Debate, 3 J. Telecomm. and High Tech L. 23 (2004); Mark A. Lemley & Lawrence Lessig, The End of End-to-End: Preserving the Architecture of the Internet in the Broadband Era, 48 UCLA L. Rev. 925 (2001). For background on network neutrality initiatives outside the United States, see Catherine Jasseranda, Critical Views on the French Approach to Net Neutrality, 16 No. 9 J. Internet L. 18 (March, 2013); European Parliament, Directorate General for Internal Policies, Policy Department, Network Neutrality: Challenges and Responses in the E.U. and the U.S., IP/A/IMCO/ST/ (May, 2011); available at: Toshiya Jitsuzumi, Discussion on network neutrality: Japan s perspective,3 Comms & Convergence Rev., No. 1, (2011). 17 Network neutrality refers to government mandated nondiscrimination, transparency and other requirements on ISPs designed to foster a level competitive playing field among content providers and to establish consumer safeguards so that Internet users have unrestricted access limited only by legitimate concerns such as ISP network management and national security. 18 Some ISPs publicly disclose generic peering agreement models. See, e.g., LINX, Model Peering Agreement Template; available at: VIX Vienna University Computer Center, available at: twtelecom, Public Peering Policy available at: 19 The Internet cloud refers to the vast array of interconnected networks that make up the Internet and provide users with seamless connectivity to these networks and the content available via these networks. The increasing functionality of the Internet is decreasing the role of the personal computer. This shift is being led by the growth of cloud computing --the ability to run applications and store data on a service provider's computers over the Internet, rather than on a person's desktop computer. William Jeremy Robison, Free at What Cost?: Cloud Computing Privacy Under The Stored Communications Act, 98 Geo. L.J. 1195, 1199 (April, 2010). 20 The Internet is a global network of networks that has been the platform for revolutionary innovation. The role of the Internet in enabling innovation is not accidental; rather it flows from the Internet's architecture. The key innovation-enabling feature of Internet architecture is comprised of layers, narrowly understood as defined by code or