European regulators can set our innovators free

In 2019 I wrote an article for InterMEDIA entitled ‘Pressure to regulate’.¹ The incoming President of the European Commission, Ursula von der Leyen, had just announced her intention to bring forward legislation for a coordinated European approach to the regulation of artificial intelligence. Her stated intention was to move quickly, introducing legislation within her first 100 days in office, and this pressure to regulate set alarm bells ringing. 

I was not opposed to regulation – indeed I had spent almost twenty years working as a regulator. But that same experience meant I was familiar with the limits of regulation, in particular the risk that overly prescriptive regulation of an emerging technology will hold back innovation. I noted that the risks arising from AI depend on the use case: an autonomous vehicle presents very different risks to those of a movie recommendation engine. I concluded that regulation should only be applied where there is clear evidence of harm and that it must address that harm in a targeted manner.

This conclusion was not novel and could be accused of being obvious. But the pressure to regulate, in response to public concerns over an immediate threat, can often outweigh a more abstract desire to protect innovation. Certainly, the European Union did go on to create a comprehensive framework for the regulation of AI which was so intrusive that it is now the subject of a simplification package.²

More broadly, Mario Draghi’s recent report on European competitiveness suggested that ‘we claim to favour innovation, but we continue to add regulatory burdens onto European companies’.³ Draghi went on to conclude that this burden must be reduced, an argument which has been widely accepted, at least in principle.

I have now written a book which explores this issue in greater depth. In ‘From Beacon Fires to Fibre Broadband: A history of innovation, enterprise and regulation’, I describe how the technologies we use to communicate have evolved over the last 200 years, how entrepreneurs have exploited each new innovation and how governments have responded by regulating them. I then draw some conclusions as to what it takes for regulation to support innovation rather than harm it.

Legitimate concerns, intrusive regulation

Over the last few years, the EU has passed several pieces of legislation which deal with digital platforms (the European AI Act, Digital Markets Act and Digital Services Act) and the UK has done the same (the Online Safety Act and the Digital Markets, Competition and Consumers Act). These initiatives address legitimate concerns, and I do not intend to argue that they are wrong in principle. The question I will consider in this article is whether the pressure to regulate leads to them being too intrusive in practice. I will test this by taking one example, the ‘vertical interoperability’ rules created by the Digital Markets Act, and comparing the application of these rules with historic examples of ‘access and interoperability’ rules in telecommunications.

Since the dawn of history, communications networks have relied on access to existing civil infrastructure. Indeed, for thousands of years, postal services developed hand in hand with road networks. Then, in the 1860s, those same road networks became the subject of the first regulatory frameworks designed to promote competition in electronic communications. At the time, the cheapest way of building a telegraph network was to lay wires along railway tracks, but incumbent telegraph companies were using exclusive agreements with railway companies to deny this benefit to their competitors. Thus:

• In Britain, the 1863 Telegraph Act authorised competing telegraph companies to ‘place and maintain a telegraph under any street or public road’. Where some form of obstacle prevented them from exercising that right, the Act authorised them to lay their wires ‘under, in, upon, over, along or across any land or building, or any railway or canal’.
• Later, in America the 1866 Post Roads Act authorised competing telegraph companies to construct telegraph lines ‘over and along any of the military or post roads of the United States’ and  ‘over, under, or across the navigable streams or waters of the United States’.

Initiatives such as these, which make it easier to lay wires along public land, have been an important means of promoting competition ever since. Moreover, over the last few decades, regulators in several countries have adopted a similar approach to the sharing of certain ‘dumb’ assets controlled by incumbent networks. For example local loop unbundling allowed competing operators to take control of the copper wires belonging to the incumbent operator. Regulators around the world used it to ensure that investment in copper-based broadband was contestable.

Crucially, interventions of this kind have little downside. Where competing operators share the same physical assets, this reduces the additional cost created by the duplication of network infrastructure, but it does so without constraining the services that different operators provide. The service differentiation this makes possible is an essential ingredient of sustainable competition – if the only difference between an incumbent operator and its competitors is the scale at which they operate, then competitors are unlikely to survive – and it also ensures that competition drives innovation.

That said, shared access to ‘dumb’ physical assets is not always sufficient to ensure competition. Thus, regulators have also come under pressure to mandate shared access to the ‘intelligent’ layers of electronic communications networks, which manage the provision of services – data transmission equipment, electronic switches and routers, and so on. Of course, such an approach reduces the scope for service differentiation, but it is possible to mitigate this risk by applying an ‘essential facilities’ test to any access obligations.

Crucially, this access obligation is not constrained by any form of ‘essential facilities’ test

This can take various forms, but for present purposes I will assume it contains three components. Firstly, it only applies obligations to those assets that are essential for competition and hard for competitors to replicate; secondly, it only applies obligations to those companies that operate in several related markets, reducing their incentive to provide access commercially; and finally, it avoids applying access obligations to recent inventions, recognising that a period of exclusivity is a legitimate reward for innovation.

The overall aim is to leave scope for entrepreneurs to do the unexpected and develop services which are not simply a repackaged version of services which already exist. This is certainly possible, as can be illustrated by a couple of historic examples.

• In 1968, the Federal Communications Commission (FCC) decided that consumers should be permitted to attach their own choice of telephone to the Bell System’s telephone network. It accepted that interoperability standards would be necessary, to protect both the network and users from harm, but also argued that such standards would be sufficient to address any concerns. As it turned out, this intervention did not just give consumers a greater choice of telephone, it also allowed them to connect their own data modems to telephone lines. By 1976, the FCC was able to report that 280 different models of data modem were available from more than 50 different manufacturers.

• In 1985, the newly created British regulator, Oftel, decided its first interconnection case. The dispute was between British Telecommunications, which had just been privatised, and Mercury Communications, the first altnet. Resolution of this dispute allowed Mercury to launch its service the following year and other altnets followed its example. But the real success story of 1985 came from another direction, when a start-up called Vodafone carried the first call over its mobile network. By 2004, Ofcom was able to report that the revenue generated by mobile networks had overtaken the revenue generated by fixed networks.  

But there are also examples of regulators intervening to mandate access to assets that do not pass the essential facilities test. Such interventions have often turned out to be counter-productive. For example, in 1913, the US attorney general extracted a set of commitments from the Bell System, known as the ‘Kingsbury Commitments’. These gave independent local telephone companies access to the Bell System’s long-distance telephone network, making it easier for them to offer long-distance services to their subscribers. But competing long-distance networks were already being built and would eventually become one of the great  success stories of telecommunications liberalisation. The Kingsbury Commitments merely reinforced, for a few more decades, the Bell System’s dominance in this market.

In a more recent case, in 2000, Oftel introduced a new form of access to BT’s telephone network, called ‘flat rate internet access call origination’ (FRIACO). This made it possible for internet service providers to charge their users a fixed fee for dial-up access to the internet, rather than the more conventional pence per minute charge. However, that same year also saw the launch of the first mass-market broadband services, which provided ‘always-on’ internet access similar in character to that enabled by FRIACO. Broadband technologies provided a more efficient means of providing such services, with much greater long-term potential, and the FRIACO intervention merely created a temporary diversion. 

Access obligations under the DMA

Which brings me to the Digital Markets Act (DMA). The DMA allows the European Commission to designate digital platforms as ‘gatekeepers’ in relation to their ‘core platform services’ if they meet a specific set of thresholds. Article 6(7) of the DMA then requires these gatekeepers to implement a set of ‘vertical interoperability’ obligations, which ensure that third parties developing new applications are able to access all the hardware and software capabilities of a device.

Crucially, this access obligation is not constrained by any form of ‘essential facilities’ test. Instead, the DMA simply requires gatekeepers to provide access to ‘the same hardware and software features…as are available to services or hardware provided by the gatekeeper’. Moreover, the DMA requires this access to be provided free of charge. 

The way this works in practice is best illustrated by a specific example. Recently, the European Commission decided to mandate full interoperability between iPhones and devices that might connect to an iPhone, such as smartwatches and headphones. The goal is to ensure that connected devices made by any manufacturer work well with iPhones, and to achieve this the Commission has mandated access to nine different features of the iOS operating system.

In doing so, it has explicitly rejected the need to assess whether these forms of mandated access are actually necessary:

‘Common to Apple’s proposed legal standards is the need to assess why third parties are currently unable to provide a “competitive offering” or an “alternative solution.” Such an assessment, which would require an individual examination of the specific circumstances of each third party, is not required by [the relevant article of the DMA]. Such an assessment would introduce the requirement to investigate on a case-by-case basis the effects on competition of a gatekeeper’s given conduct, which the legislator explicitly rejected.’⁴ 

I make no comment as to whether the Commission’s position is correct as a matter of law, but it certainly reflects a much more intrusive approach than has historically been applied by regulators of telecommunications networks. It might be reasonable to mandate access to those features of a device operating system that meet some form of essential facilities test. But mandating access to every feature of such operating systems, and doing so free of charge, leaves very little incentive for other companies to develop innovative solutions of their own.

Over the period of more than 200 years described in my book, several companies once thought to have been unassailable have been wiped out by disruptive new technologies. The current generation of digital platforms may seem to be unassailable now but, taking the long view, that is unlikely to be the case. Their enormous success is due to them creating products that people want to buy and if they cease to do so, I am confident that some currently unknown entrepreneur will work out how to take their place. Speaking as a European, I would love to see more of those entrepreneurs come from Europe, but that does require us to give them space to work their magic.

From Beacon Fires to Fibre Broadband: A history of innovation, enterprise and regulation is available from Troubadour Publishing.