Network Neutrality. Can Regulation Save the Internet?

1. Who Would Win? The Internet or...

In December 2017, the net neutrality regulation of the US was repealed (Early and Bustillos 2018). One year later, evaluations of the consequences show an immense bandwidth. Some find that the (American) internet did not change at all or even improve because of the legal change, citing spurious relationships such as increased internet speed rates (Moran 2018). Others see very concrete negative consequences, for example that the promises of ISPs about more network investment, lower prices and no negative traffic discrimination were not kept (Sohn 2018). And research conducted in 2018 using the app Wehe concluded that in their tests, “nearly every” US cellular ISP (internet service provider) throttles traffic, meaning setting a limit on the available bandwidth for at least one streaming video provider - before and after the repeal of the law took effect (Choffne 2019). So what can be inferred from the ‘natural experiment’ with net neutrality regulation in the US by first introducing and then negating it again?

On the one hand, the fact that the debates sparked primarily in the US at different points of time triggered similar actions in other parts of the world shows the importance of the question. As average citizens are usually not too well informed about or involved in the shaping of internet governance, the massive mobilization of civil society that led for example in the US (2014 and 2017), Brazil, the EU and India to thousands of contributions on public policy consultations is even more striking (Marsden 2017). With an ever increasing demand for more bandwidth and rising numbers of internet users especially in developing countries, not only the topic of net neutrality will become even more salient but users will come to realise this salience. On the other hand, the content of the debates and the number of countries without net neutrality measures in place (Dynamic Coalition on Net Neutrality 2018) shows relatively small progress since the beginning of the century. Reasons for that are the lack of empirical evidence for common claims or about the impact of introduced regulations as well as the missing link between economic, social, technological, political and human rights based arguments in the discussions (Marsden 2017). But one of the main reasons is probably lacking awareness among policymakers and national stakeholders how pressing the issue is. In the following I try to address those challenges by asking if and when there should be regulations on net neutrality.

2. Neutrality Regulations For Every Netizen

The term net neutrality, or network neutrality, was coined by Tim Wu in 2003 (Wu 2003). It means “the principle according to which Internet traffic shall be treated equally, without discrimination, restriction or interference regardless of its sender, recipient, type or content, so that Internet users’ freedom of choice is not restricted by favouring or disfavouring the transmission of Internet traffic associated with particular content, services, applications, or devices” (Dynamic Coalition on Net Neutrality 2013). This encompasses the ISPs obligation to transmit data with the best possible effort without blocking of content, applications, services, or devices, throttling of internet traffic or paid prioritization of traffic, no matter if the traffic comes from affiliates or other CAPs (content and application provider). There is quite broad consensus that this very narrow definition can be opened up to allowing the blocking of harmful traffic and devices, the prioritization of emergency services and enforcing court orders (e.g. Dynamic Coalition on Net Neutrality 2013; Bauer and Obar 2014). Apart from this concept, different models have been used and discussed. They mainly are (1) prohibiting blocking of content, applications and devices, (2) prohibiting a termination charge of ISPs towards CAPs i.e. zero price constraint, (3) quality of service (QoS) tiering without discrimination between CAPs, (4) QoS tiering with an obligation to provide a minimum quality service tier, and (5) QoS tiering with discrimination i.e. no net neutrality specific regulation in place (Olmos 2013; Bauer and Obar 2014). As Olmos (2013) and several others point out, there are different means of trying to enforce net neutrality such as legal measures and active monitoring of enforcement or ex post complaint systems, non-binding ones such as a code of conduct, or relying on antitrust law and a competitive market. But how can be determined which option is the best?

I argue that every actual or potential user of the internet should be covered by regulations ensuring some kind of net neutrality. One should not wait until policymakers in charge identify a problem but rather save time and costs by making regulation or giving respective authority to agencies to can be matched to potential innovations in technology (Marsden 2017). The common pitfall of internet governance of slow legislation reacting to technological advances that cannot be applied to potential innovations just around the corner could be avoided like that. Also, how does someone know there is “a problem”, that net neutrality is a policy issue that calls for regulation? This is one of the questions I try to address in the following.

2.1 Net Neutrality as a Safeguard for Human Rights in the Digital Age

Net neutrality is a norm of human rights discourse. Therefore it places legal obligations on states to safeguard net neutrality under the regime of internationally accepted (human rights) law (Carrillo 2015). Human rights are the normative framework to account for arguments of the social, economic/technical and political discourse, as those often rather talk past each other than engage. Net neutrality affects the right to freedom of expression, information and communication. Freedom of expression in turn enables other rights, such as the right to hold opinions and religious beliefs without interference, the right to education, the right to freedom of association and assembly, the right to full participation in social, cultural and political life, as well as the right to social and economic development. Net neutrality enables and protects free speech by prohibiting discrimination and blocking which can be seen as an act of censorship, guaranteeing the right to transfer or access information without restrictions (Carrillo 2015) and therefore safeguarding the right to free communication and the free choice of content. Carrillo (2015) showed that among others, different United Nations bodies made clear that in current times states have the obligation to promote universal access to the internet, connectivity, in order to realise the right to freedom of expression. This means that the legal discourse not only makes clear that human rights extend online but that connectivity and further expansion of broadband networks are (policy) goals that should not be neglected in the debate about net neutrality.

In concrete terms, net neutrality can be seen as in place if other ‘rights’ are ensured: the right to choose ISPs, the right to access the internet, and the right to use data and content (Cheruvalath 2018). Empirically especially the first point is very problematic, I will come back to that. If citizens who do not have access to the internet at all should be included in the principle of net neutrality and how, for example by zero rating, remains a hotly debated topic I cannot address here due to space constraints.

2.2 Net Neutrality as Facilitating Force for the Online Public Sphere

Many have written about the role of the ‘liberation technology’ (Diamond 2010) internet in democratization and its general implications for democracy. But all these findings and the whole notion of the internet as many-to-many communication or Web 2.0 had to be questioned if in the long run, the ‘worst case scenario’ of net neutrality becomes true and everyone who does not pay for prioritization has to deal with blocking or throttling. With practices like blocking, throttling and prioritization in place, it would become easier to ‘buy influence’ and pursue more or less hidden political agendas in the online media environment and connected to that influence on voters. For example, foreign actors can seek to influence elections by getting political content prioritized while ISPs throttle or block access to content of national competitors. As the internet is a key information source for millions of citizens all over the world, net neutrality links directly with media pluralism and cultural diversity online, which are especially vital for democracies (Olmos 2013). Musiani and Löblich (2016) support the argument that in order to work, an (online) public sphere in the sense of Habermas (1989) has to represent the diversity of information, ideas, and opinions of citizens. Without net neutrality, this becomes more difficult (Bauer, Obar 2014).

In turn, if the online public sphere becomes constricted, that hurts the resilience of societies and the quality of democracy for example by discouraging civic engagement when at the same time realising collective action potential with the help of the internet becomes more salient. Activist usually do not have aligned interests with ISPs or the money to alleviate blocking/throttling but they are dependent on the internet (Hussain and Howard 2013, Ruijgrok 2017). If large ISPs and CAPs determine what end-users can access online at what price with blocking and throttling, citizens lack an open and cheap opportunity to educate themselves without discrimination. Net Neutrality is as cornerstone for media literacy as it is based on equality and diversity, especially in the context of zero rating. Media literacy is an important prerequisite to fully participate in political life and in democratic processes and becomes more salient with the emergence of new technologies like bots, the spread of misinformation and increasing mediatisation. Therefore it should be in the interest of every democracy to enforce net neutrality principles.

2.3 Net Neutrality as an Economic Factor

It is widely known that the internet has a great economic impact. As the stakes are high, economic arguments are quite prominent on both sides of the debate about net neutrality. As Lee and Shin (2016) summarize correctly, the economic debate centers around the questions if the vertical integration of ISPs harms fair competition, if ISPs have an incentive to discriminate against CAPs and if network neutrality regulations increase the incentive to invest in the broadband network. The challenge is that no economic theory provides clear support for any position and respective analyses based on empirical evidence are rare. Literature reviews mostly report mixed evidence and theoretical claims (e.g. Easley et al. 2017; Bauer and Obar 2014).

In a qualitative case study of the UK, Cooper and Brown (2015) provided evidence that discrimination can hamper application innovation by creating costs for developers and, in contrary to frequent theoretical arguments, cannot be deterred by competition. A common observation is that the internet has low access/entry barriers for end-users, CAPs and ISPs. An open and interoperable ecosystem guarantees that new ideas can be added, systems can be complemented and products and services can be enhanced by other market players constantly and ‘without permission’. This environment which is friendly to innovation should be preserved (Olmos 2013). Without net neutrality, smaller online competitors are put at an disadvantage because they are not in the position to pay for prioritization of their content or applications, not well-known enough (yet) or simply because they are competing with services of ISPs, for example providers of VoIP (voice over Internet Protocol) services. In the long run that could mean that on the internet, no longer the best innovations thrive but ISPs act as gatekeepers by making deals with CAPs (Wu 2006, Giles 2017). This would not only threaten the innovative character of the internet but also decrease the values of an internet connection for users. E.g. if Sprint blocks Skype because it is competing with Sprint’s telephony (Choffnes 2019) Skype is effectively censored for all Sprint subscribers.

A report commissioned by the Dutch Ministry of Economic Affairs also finds that net neutrality stimulates innovation, arguing that payments for prioritization and the blocking of competing services and applications reduces competition between ISPs and increases the entry barriers for CAPs, particularly for small CAPs (Kocsis and Weda 2013, Olmos 2013). There are two plausible reasons for that: First, as illustrated in the example, paid prioritization and blocking of competing services mean costs that discourage innovation by new and established CAPs. Second, because ISPs want to avoid the ‘commodity trap’ – the internet as a homogenous product – and be able to increase fees for end-users, they have an incentive to reduce competition by differentiating themselves from other ISPs. For that purpose, in absence of net neutrality they offer paid prioritization only for certain large/well-known CAPs with which other ISPs do not contract. But if competition is reduced, also the incentive of ISPs to innovate are reduced because they do not need to compete for end-users as much as before. That means that net neutrality does not only increase competition between ISPs but also fosters innovation at the core and at the edges, i.e. of ISPs and CAPs, by reducing transaction and adaptation costs for innovators. These reduced costs can also widen the scope of feasible innovation projects at the edges (Kocsis and Weda 2013, Bauer and Obar 2014). Olmos (2013) relies on past violations of net neutrality, for example citing a BEREC (Body of European Regulators for Electronic Communications) study from 2012 where 28 out of 155 European operators on mobile networks reported restrictions to the transportation of VoIP traffic affecting at least 20% of subscribers. She concludes that real-world experience shows that as soon as ISPs have the possibility to apply negative discrimination against OTTs (over the top services), meaning in the absence of net neutrality, they will use that option (Olmos 2013).

Others, for example Josh Steimle prominently promoted the view that the absence of net neutrality regulation would entail more competition among ISPs, benefitting the end-users (Steimle 2014). Baake and Sudaric (2018) conclude that if there were no regulations on the prioritization of traffic, ISPs had an incentive to lower their subscription prices in order to increase the number of connected households. Their argument is that as CAPs can self-select if they want to pay to prioritize their data packets, prioritization becomes more profitable for them as more households are connected to a network. But empirically the mechanism that lower subscription prices directly lead to more connected households is to be questioned, as factors like for example physical infrastructure also play a role. The authors note themselves that competition is a condition that is not always met, therefore I conclude that this incentive is actually really weak, as cases like the one of Netflix paying ComCast shows (Gustin 2014, Greenstein et al. 2016). Still there is a difference between noting this fact and not considering cases which actually happened (Gustin 2014, Greenstein et al. 2016, Coldewey 2017, Marsden 2017, Karr 2018, Choffnes 2019) where ISPs used their gatekeeper function to let CAPs pay them for not blocking their traffic without considering changes about their own policies towards end-users, be it lower subscription prices or investment in the network. It also illustrates that the assumption of Baake and Sudaric “[s]ince only offers with sufficiently high added quality would be prioritized, indicating a higher willingness to pay for said prioritization, the selection process would be efficient” (p.234) is questionable. Out of all the OTTs or time-sensitive services that the internet has to offer, from VoIP and live streaming to high quality online gaming, Netflix does not seem like the first service to be on the list but simply one of the biggest OTTs. Netflix was more or less forced to pay because their data was throttled before, not because their service is particularly time-sensitive. Out of that, it is rather unlikely that ISPs would prioritize data after a deliberative debate with all content providers, but rather if they get paid by them. So, can one still speak about an “efficient” process if, say Skype pays the ISP to get its content prioritized but a company with the equally time-sensitive kind of data does not have the financial resources to do the same? Or if an ISP which also functions as a content provider prioritizes own content even if it is not time-sensitive?

Also, when speaking about competition for end-users, one has to note two important facts. One, in many countries, the ISP market is not especially competitive so that many end-users only have few or even only one option for an ISP. So arguments about competition do not apply to all places in the world evenly. Two, end-users ability to ‘vote with their feet’ depends on transparency and said competition: if end-users are informed of their ISPs policies and there is real diversity of choices, the demand could drive the market towards offers without negative discrimination by ISPs. But empirically, the mentioned BEREC study from 2012 finds that even the highly competitive EU telecommunications market has been unable to avoid restrictions (Olmos 2013). In reality, end-users are not well informed enough to shape the market (Cooper and Brown 2015, Olmos 2013). Additionally, an indicator that the incentive for ISPs to deter competition by for example discriminating against VoIP services is bigger than the negative effects of transparency and a competitive market on this incentive is the case of the Netherlands. Before net neutrality regulation was introduced there, all mobile communication providers were blocking VoIP (Olmos 2013).

The example with Netflix addresses the question of traffic management that becomes increasingly prevalent with the spread of content and applications demanding a lot of bandwidth: If end-users consume big volumes of data, e.g. stream videos, that does not influence their fees to the ISP. So, end-users do not (have to) chose their content or applications based on the price. But that also means that ISPs do not get recompense for the bandwidth they provide (Antonopoulos et al. 2017). As numerous violations of net neutrality show, ISPs rather charge OTTs extra than invest in infrastructure for the purpose to offer prioritization (Gustin 2014, Greenstein et al. 2016, Coldewey 2017, Marsden 2017, Karr 2018). If net neutrality is in place, ISPs are not allowed to do that. Does that mean that under net neutrality, there is no incentive to invest in broadband networks?

ISPs would be able to get additional revenues from selling prioritization. Some present this as the only option for them to be able to invest in broadband capacity (e.g. Easley et al. 2017). But many voices base their arguments on numbers about capital expenditure of ISPs in the US. Capital expenditure cannot be used as a proxy for the influence of net neutrality regulation on investment because capital expenditure is influenced by more factors than just this one. Apart from that, arguments about investment in developed countries with high internet adoption rates often do not apply to developing countries or least developed countries.

Other actors and scholars argue that ISPs do have an incentive to invest under net neutrality. Antonopoulos and colleagues (2017) for example agree that OTTs free ride the network deployed by the ISPs and therefore distort incentives for investment. But after examining financial data, they conclude that the interests of OTTs and ISPs are not necessarily in conflict here, since the economic gains of the OTTs were positively correlated with the ISP revenues and vice versa. Also, they can not find a clear motivation for the OTT providers to contribute financially to the network infrastructure, as capital expenses seemed to stimulate the economic growth of ISPs, but were detrimental for OTT profits (Antonopoulos et al. 2017). Another argumentation is that if CAPs are allowed to pay for prioritization, ISPs have an incentive to keep the capacities of their network low so that the CAPs have an incentive to pay for prioritization. That means that if paid prioritization is not an option, ISPs have an incentive to invest in their network in order to make profit. But one has to note that here the assumption is that the ISPs do not ‘artificially create’ that incentive for the CAPs by throttling their traffic (Verbraucherzentrale Bundesverband (vzbv) 2015; Easley et al. 2017). Ford (2017) showed with a counterfactual analysis of the US that net neutrality had no negative consequences on investment.

Net neutrality also stimulates connectivity between end-users and CAPs. With more innovations, the value of an internet connection rises, making it more attractive for more end-users to connect to the internet as it offers more and better content and applications. At the same time, larger connectivity makes providing that innovative content and applications more attractive for CAPs as they convert the network externalities (the benefits from connectivity) into financial benefits. Also, with greater connectivity CAPs can benefit indirectly through more advertising revenues (Kocsis and Weda 2013). The Dutch report mentioned before argues that as the value of the internet connection itself increases for end-users, they are willing to pay more for an internet connection which would profit ISPs, making them also benefit from larger connectivity, which fosters innovation. If one follows this argument, one could conclude that prioritization shifts the focus of investment from innovation to the status quo (Kocsis and Weda 2013).

2.2 Don’t Fix Something That Is Not Broken?

A common argument of net neutrality opponents is that the internet has ‘worked’ with only little regulation so far and new legislation is therefore not needed, or even harmful. But how does one know that the internet ‘works’ in regard to net neutrality, that there is no need for action?

As participants of the debate note (Olmos 2013; Cooper and Brown 2015; Musiani and Löblich 2016; Marsden 2017), the vast majority of end-users is not well informed enough to assess if they are victims of violations of net neutrality principles. That policymakers do not hear about complaints or rampant violations does not mean that net neutrality is in place. Complaint systems like in the USA or Chile have proven to be not sufficient (Allen et al. 2017, Marsden 2017). Still, many cases have become public (Gustin 2014, Greenstein et al. 2016, Coldewey 2017, Marsden 2017, Karr 2018). Qualitative self-reports of traffic management are a paradoxon - to ask ISPs if they discriminate traffic and expect this helps with the enforcement of net neutrality. The only option left that is implemented by governmental actors is active monitoring of the QoS (quality of service). Measuring net neutrality as a concept remains a challenge (Marsden 2017; Allen et al. 2017; Internet Governance Forum 2018). This is also one of the reasons for the lack of meaningful objective empirical evidence. National regulatory agencies in Brazil, Canada, the US and the EU (including the UK) monitor traffic using SamKnows (Marsden 2017, 2018) while researchers and organizations also use applications like Wehe (Marsden 2017, 2018; Choffnes 2019). BEREC published some useful technical guidelines, among them the Net Neutrality Regulatory Assessment Methodology in 2017 (Allen et al. 2017; Nieminen 2018). BEREC is also in the process of developing an open source measurement tool to help measure QoS and discover potential illegal traffic management practices. But even if QoS measurement can detect prioritization and throttling, Marsden (2017) notes that currently, the only method to detect net neutrality violations is if ISPs advertise those practices, for instance in the case of zero rating.

As there are no standards about measuring net neutrality, many countries do not assess the state of net neutrality or components of it at all and evidence based policymaking is rather an exception, it is not an option to wait until policymakers around the world recognize net neutrality to be in such a bad condition that they have to work on legislation. Practices like zero rating can shape not only the market but also citizens in ways that is hard to reverse (Mirani 2015; Verbraucherzentrale Bundesverband (vzbv) 2015; Carrillo 2015; Russell 2015; Greenstein et al. 2016; Musiani and Löblich 2016; Marsden 2017; Solon 2017; Cheruvalath 2018).

3. Evidence Based Policy is Needed

The internet evolution and success have mainly relied on net neutrality. With advancing technology, policymakers can no longer take net neutrality for granted as long as they do not know about any discrimination. In general, traffic management can enhance the quality of service, for example when using VoIP (e.g. Olmos 2013; Welch 2015) but as soon as prioritization is paid, the outlined threats overweight. In the worst case, ISPs hardly invest in their networks but put prioritized traffic at the current usual speeds and everyone else at a lower speed (Marsden 2017; Baake, Sudaric 2018). That would dramatically decrease the user experience as with technological development humans get used to certain speeds and dissatisfied if they do not receive the speed they are accustomed to. This makes them change their behavior (Krishnan, Sitaraman 2012) such as change the content provider, the application or the ISP, leading to ‘guided customers’ replacing ‘content is king’.

Net neutrality suffers from the common internet governance problem that there is a lack of policing and enforcement of extraterritorial jurisdiction. The calls for a binding international treaty or clear rules for the application of human rights legal norms are met by deaf ears (Marsden 2017; Cheruvalath 2018). Because technologies are fast-moving, typically have international impact and require expert design choice to implement policy, enforcement is difficult and uneven on the Internet. This is why also ‘soft law’ types are required (Marsden 2017).

As I showed, especially arguments about competition and connectivity can not be generalized across countries and regions. Out of that, net neutrality regulation should be flexible and not too detailed (van Eijk 2014) as policy often only catches up with business and technology (Marsden 2017; Allen et al. 2017). Whatever version of net neutrality policymakers may strive for, it is most important that they base their considerations on empirical evidence and ensure active quantifiable monitoring of enforcement. So far, the medium-term effects of net neutrality regulation are not sufficiently examined. The ‘natural experiment’ of legislation in the US could not provide insights here yet, as it would not be plausible for ISPs to immediately radically change all their policies towards content providers and customers when public awareness for the topic is still high. In this US centric debate, further independent objective research providing empirical evidence is needed, particularly about the impact of net neutrality regulation and its enforcement in different contexts (Marsden 2017, Internet Governance Forum 2018, Choffne 2019).

[...]