Net neutrality is the regulatory principle that defines the relationship – or rather the need for no commercial relationship – between social media and streaming providers and the telecommunications networks over the top of which (OTT) that content is provided to end users.
The UK regulator, Ofcom, recently released a discussion paper on net neutrality in which it proposes a post-Brexit version of the EU net neutrality regulation that will be somewhat more liberal for the telcos, but also side steps ‘bigger asks’ of the content providers demanded by the telcos.
As the Ofcom discussion paper is likely to add to the building debate on net neutrality (and the paper itself is well over a 100 pages), this week’s article is a little longer than usual.
What is net neutrality?
"‘Net neutrality’, sometimes referred to as the ‘open internet’, is the principle that users of the internet (both consumers and those making and distributing content) should be in control of what they see and do online – not the broadband or mobile providers that connect people and businesses to the internet (otherwise known as internet service providers or ISPs)."
Net neutrality addresses the concern that the provider of ‘the connectivity pipe’ into the home can advantage its own competing content businesses over those of the OTT providers. The net neutrality rules impose ‘must-carry’ and ‘non-discrimination’ obligations on ISPs, which is achieved by prohibiting them from blocking, throttling, or applying differential treatment of on-net vs off-net traffic.
The more things change, the more they remain the same
Net neutrality is an export from the USA, developed well over 20 years ago. While Ofcom acknowledged that the internet value chain has changed significantly since the net neutrality rules were introduced, it considers that the basic rationale for net neutrality remains valid.
First, while today there is more competition between fixed local broadband networks, Ofcom considers the risk from vertically integrated ISPs remains:
“ISPs compete against each other for consumers to subscribe to their connectivity services. Consumers typically subscribe to a single ISP over a period of time. Every ISP therefore provides a unique route for content application providers (CAPs) to reach an ISP’s customers over that period of time and CAPs require access to multiple ISPs to reach end users widely.”
Second, Ofcom acknowledged that mobile and fixed networks increasingly compete against each other and most customers have both fixed and mobile subscriptions. While in theory this gives CAPs alterative pathways to the same customer, “in practice.. distributing content via every fixed and mobile ISP is likely to be essential for CAPs because fixed and mobile connectivity tend to serve different consumer needs.”
Third, Ofcom acknowledged that are now other powerful providers in the internet value chain that also hold ‘gatekeeper’ positions and control the content accessed by consumers, such as Apple and Google. While these larger CAPs “can potentially use their strong bargaining power to protect themselves from being exploited by ISPs”, the smaller CAPs remain vulnerable.
Fourth, while the competitive landscape is no longer dominated by incumbent local carriers like BT, there would still be reason to apply net neutrality to ISPs regardless of whether they have market power, because of the ‘friction’ that would result if CAPs had to negotiate access with ISPs. The practical outcome of net neutrality is that CAPs can provide their services on an OTT basis without needing a commercial agreement with the underlying network provider:
"All CAPs should be able to access the widest possible market, and any barriers for them doing so could be problematic. This could be especially true for smaller CAPs who have limited scale and less resources…. Such barriers could exist irrespective of whether ISPs may have market power over CAPs. For example, if CAPs were required to agree with ISPs to deliver online services to their customers through some form of negotiation, contract or registration, this could deter a long tail of CAPs from accessing end-users…If market forces were left to their own devices, such barriers could undermine smaller CAPs’ ability to distribute their content to consumers and citizens widely.”
It has to be said that this is a somewhat surprising conclusion – in today’s dynamic connectivity markets, connectivity is ‘glued together’ across multiple domestic and international networks, between network layers and between big and small operators by a web of commercially negotiated arrangements.
So why re-think net neutrality?
The central problem is that the CAPs are driving an explosion in traffic which the ISPs have to scramble to keep up with by making massive, continuing investment in network capacity:
- the average year-on-year traffic growth in the UK over the period 2013 to 2021 was 42% for the average fixed broadband consumer, while average growth was 37% for the average mobile data consumer;
- the busiest traffic period (around 8pm), to which ISPs have to build network capacity, is becoming “peakier”;
- Yet about 50% of UK busy period traffic is driven by just 5 CAPs, Amazon, Facebook, Google, Netflix and Sky.
But Ofcom considered that this exponential growth in traffic was not one-sided dilemma for the ISPs. Because the CAPs’ business case is all about the ‘eyeball experience’ of consumers, CAPs have their own incentives to manage the traffic burden on the underlying networks to ensure a higher quality of service. CAPs do this by investing in their own caches (servers) or relying on third party Content Delivery Networks to improve service quality by hosting popular content closer to the customer, so that less of the shared public internet is used to deliver the content.
Ofcom noted the success of these CAP strategies in reducing the load on the ISP networks:
“for the major ISPs, just over 50% of traffic is delivered via caches deployed in ISPs’ datacentres (sometimes referred to as on-net caches) and just under 40% via private peering [ed. note: as these are commercial arrangements where traffic is broadly balanced between networks, sometimes with out-of-balance payments, Ofcom seems to be implying the ISPs are commercially covered for this traffic]. Private peering traffic is likely to mostly be made up of traffic delivered via caches in datacentres. Of the remaining traffic, approximately 4% is delivered via public peering and 7% via IP transit.”
Ofcom also observed that CAPs are offering novel and innovative services, including new delay- and congestion-sensitive applications, with services delivered over the general internet that provide a mixture of both VOD and live content.
In short, Ofcom seems to consider that the CAPs are pulling their weigh on managing traffic demands, and that, despite the vociferous arguments put by the ISPs, this alone would not be a basis for reviewing net neutrality.
Ofcom’s real concern appears to be that net neutrality may be constraining the ability of ISPs to innovate in building new retail products which exploit recent technological advances which allow a range of service qualities, including on parameters such as jitter and latency. While Ofcom does not expressly say so, this seems to reflect industry concerns that, as basic connectivity is commoditising, ISPs need to find new, more profitable revenue streams to fund their networks, including to invest in the massive capacity expansions required to support CAP services.
So the focus of the Ofcom paper is how to tinker with net neutrality to loosen restrictions on innovations by ISPs, without disturbing too much the benefits of net neutrality for CAPs.
Ofcom tinkering No: 1: Inclusions in data caps
The EU net neutrality rules prohibit an ISP entering into agreements with end users based on commercial or technical conditions which limit end users’ rights to access and distribute the information of their choosing.
As in Australia, a common feature of retail broadband plans, especially on mobile, is to have a data cap beyond which additional ‘penalty’ charges apply for usage. A feature of these data caps can be a zero-rating offer: an ISP gives its customers more favourable access to certain content over other types of content by not subtracting usage of zero-rated content from the user’s data allowance.
In September 2021, the European Court of Justice ruled that certain zero-rating offers breached the EU net neutrality rules.
As a post-Brexit Ofcom is no longer bound by the ECJ decisions, Ofcom was free to take a different view:
“zero-rating offers give ISPs another way to tailor retail products to meet customers’ different preferences. This can allow ISPs more flexibility to innovate with differentiated retail products that provide consumers with greater choice and access to the content that they value the most.”
But even so, Ofcom is proposing to put some big caveats around ISPs’ freedom to zero rate. Zero-rating will be considered harmful to competition:
- If CAPs are effectively excluded from participation in the zero-rating if they wish. Zero rating will not be ‘genuinely open’ to CAPs if they are required to pay a fee to be zero-rated because that would exclude smaller CAPs with limited resources, even if the fee was fairly modest. There should not be stringent technical, legal or other financial requirements which could also deter relevant CAPs from participating in an offer. CAPs must be treated equally within the zero-rated plan (including with the ISP’s own content).
- If it is important for a CAP to be zero-rated to compete effectively. This might be because so many customers have signed up to a zero-rated plan. Or, an ISP with market power enters into an exclusive agreement with a CAP (to exclusion of other ISPs) to zero rate its content regarded as ‘must have’ by consumers. Or, a CAP with market power participates in zero rating, sheltering its content from competition.
- If the offer is likely to influence consumer behaviour to prefer zero rated content over other content. The tighter or lower the data cap, the more risk that whether a CAP is zero-rated or not will influence consumers’ buying decisions between CAPs. Ofcom noted that as the size of data caps are increasing and many consumers do not fully utilise their data allowances, this is less likely to be a concern. In 2022 more than 80% of UK consumers used less than half their monthly data allowance.
Ofcom tinkering No.2: more scope for differential service quality
The EU net neutrality rule requires ISPs “to treat all traffic equally, when providing internet access services, without discrimination, restriction or interference, and irrespective of the sender and receiver, the content accessed or distributed, the applications or services used or provided, or the terminal equipment used.” There are limited exceptions for ‘reasonable traffic management’ measures.
Ofcom considered that to date the strictness of this equal treatment requirement has worked well to date because “as ISPs’ investment has generally met the increasing demands on networks, there has been a limited need for ISPs to manage traffic to address congestion or otherwise ensure robustness of their networks.”
But going forward, Ofcom was concerned that innovation in retail products could be constricted:
“Retail competition on quality parameters, and the prospect of attracting a particular group of customers who value quality and are willing to pay a premium for it, is a key driver for ISPs to invest in their networks to improve quality of experience.”
Ofcom thought that, in turn, take-up of innovative ISP services catering to different customer needs might also have a positive impact on innovation in content and services markets. For example, it could encourage CAPs to develop innovative solutions designed around different broadband quality levels.
Different quality tiers at different price points also might help with managing congestion by encouraging customers to explicitly consider what quality of experience they want, particularly when networks become congested, and what price they are willing to pay for it.
The current EU net neutrality rules already may allow some (uncertain) level of retail product differentiation on service quality, but Ofcom proposes to clarify that the following are acceptable in the UK:
- retail offers which provide different levels of quality of service for different ISP subscriptions: the same quality must apply to all the content and services accessed by a given subscriber.
- retail offers which provide multiple quality of service levels within a single subscription if the level of quality of service is independent of the content and services accessed. These include, for example, offers in which a customer can subscribe to an add-on to (temporarily) boost their quality of service or vary the contracted quality of service across the day.
However, the current EU net neutrality rules clearly prohibit retail packages which provide a particular level of quality for specific content and services: e.g. lower jitter for a particular video-conferencing application.
Instead, Ofcom proposes to allow these kinds of ISP offers on a case-by-case basis, foreshadowing that it may step in “where an ISP and / or a CAP enjoy a degree of market power and use it to undermine the open internet and open internet-based innovation (e.g. either by directly restricting choice or by distorting competition)”. Or, a service plan may be more suspect “where an ISP requires CAPs to sponsor the higher quality scheme i.e. pay a fee to ISPs if they want to be included in higher quality retail packages, although harm would not necessarily occur in all circumstances”.
The quid pro quo for an ISP taking up this somewhat more liberal approach on differential service quality will be extensive record keeping on actual vs. promised quality levels, ensuring that the general internet service quality is not unduly sacrificed to support the higher quality services, and clarity in explaining to customers what differential service qualities mean and their rights to exit plans for poor performance.
Ofcom also proposes to clarify that the ‘reasonable traffic measures’ an ISP can apply to address congestion can include throttling all traffic to a similar degree and prioritising retail traffic for which has given contracted service levels.
Side-stepping the issue of ISPs charging CAPs
The most controversial issue between ISPs and CAPs is whether CAPs should be required to make a contribution to the capital costs ISPs incur in expanding network capacity to meet the peak traffic levels generated by the CAP services. Ofcom neatly side steps this issue by saying it is a policy matter for the Government. However, Ofcom clearly comes down on the CAPs’ side in this debate.
Ofcom has a stab at estimating the traffic sensitive costs ISPs face, which ISPs probably would consider to low ball the impact of CAP traffic:
“the information we have suggests that 50-75% of both backhaul and core fixed network costs and mobile network costs are likely to be dependent on the level of traffic at peak times. Using this estimate, these peak traffic-dependent costs would represent approximately £30-45 per fixed customer per year and approximately £20-30 per mobile customer per year.”
Ofcom notes that the case made by the ISPs for a CAP contribution to network costs is the standard economic efficiency rationale that “charges can ensure that the costs are recovered from those whose actions are causing the costs to be incurred.”
However, Ofcom considers that the ISPs’ argument falls down because “the extent to which CAPs determine the timing of traffic, and hence cause the peaks in traffic and the associated ISPs’ network costs, is limited.” It is the CAP’s customer who decides when they want to watch content, and of course, live streaming events are driven by their own scheduling. There is some limited amount of traffic which a CAP can time shift, such as updates to software etc, but Ofcom considered that the CAPs were already doing so.
Ofcom acknowledged that CAPs have an indirect impact on ISP capacity requirements through how the CAPs chose to deliver their services:
“there are various externalities present in the use of ISPs’ networks. For example, one retail customer’s use of the network at peak can increase the risk of reduced network performance and worsen quality of experience for other customers of the network. The way in which the CAP chooses to deliver that traffic at peak can have the same impact. The implication of these externalities is that they can lead to inefficiencies if CAPs and/or ISPs’ retail customers take insufficient account of these adverse impacts. Such inefficiencies can therefore provide a potential justification for a charging regime, if such a regime can provide greater incentives for CAPs to minimise the impact they have on ISPs’ network costs.”
But as noted above, Ofcom thought that CAPs had their own independent incentives to engage in strategies such as caching without needing the added discipline of network charging by the ISPs.
Ofcom acknowledged that there are significant uncertainties around the returns of future investment in fixed and mobile networks. But Ofcom had two answers to that:
- historically, unit cost decreases have enabled ISP networks to keep broadly consistent levels of expenditure over time despite significant increases in overall traffic demand. There is no reason to believe this will not continue, even if a little more uncertain (i.e. the regulatory equivalent of ‘it will all come out in the wash’); and
- ISPs can increase overall revenues by providing a higher quality or a more diverse range of services: selling premium services like high quality home broadband or 5G enterprise services. Ofcom notes that the very loosening of the net neutrality rules proposed by Ofcom will reinforce this.
Ofcom also thought that a charging regime would be too hard to implement and regulate:
“Currently, traffic is not consistently marked in a way that allows its origin or type (e.g. whether it is delay-sensitive video content) to be identified as it is routed through networks. Achieving a consistent and trusted mechanism for marking traffic and ensuring it is used correctly would require multi-lateral and international agreements between CAPs, ISPs, and transit providers.”
Finally, the consumer outcomes are unclear. In theory, ISPs should be able to reduce their retail charges to reflect the new revenue from CAPs, but it is unclear whether ISPs, especially those with market power, would pass through the benefit to customers. Conversely, as CAP costs will go up, their retail charges also may go up. But the powerful CAPs may decide to ‘eat’ the ISP charges, placing the smaller CAPs at a competitive disadvantage.
Ofcom concludes that “[w]e have not seen evidence from ISPs that persuades us that the current net neutrality regime leads to a reduced ability to recover future investment costs required to meet expected traffic growth, as they claim.”
Ofcom's position on net neutrality
Ofcom is no longer required to “take the utmost account” of the European guidelines on net neutrality. But for all its post Brexit flourishes, in the end Ofcom does not stray too far from the EU position because of its view that “[i]n general, net neutrality has worked well and supported consumer choice as well as enabled content providers to deliver their content and services to consumers.”
Read more: Ofcom proposes new net neutrality guidance