final logo red (RGB)


Select Committee on Communications and Digital

Corrected oral evidence: Freedom of expression online

Tuesday 13 April 2021

3 pm


Watch the meeting

Members present: Lord Gilbert of Panteg; Baroness Bull; Baroness Buscombe; Viscount Colville of Culross; Baroness Featherstone; Baroness Grender; Lord Griffiths of Burry Port; Lord Lipsey; Baroness Rebuck; Lord Stevenson of Balmacara; Lord Vaizey of Didcot; The Lord Bishop of Worcester.

Evidence Session No. 21              Virtual Proceeding              Questions 174 - 181



I: Daniel Gordon, Senior Director of Markets, Competition and Markets Authority; Simeon Thornton, Director, Competition and Markets Authority.



This is a corrected transcript of evidence taken in public and webcast on



Examination of witnesses

Daniel Gordon and Simeon Thornton.

Q174         The Chair: Welcome to our first witnesses today, Daniel Gordon and Simeon Thornton from the CMA, both of whom are reasonably regular witnesses before this committee. Thank you both very much indeed for joining us again. Today’s session will be broadcast online and a transcript will be produced.

As you will know, the committee has been looking at a whole range of issues in relation to freedom of expression online, thinking at the same time about the forthcoming Online Safety Bill. We have quite a few questions for you in a number of areas.

Before we start questioning, do either of you have any introductory remarks that you would like to make, or further explanation of your relative roles at the Competition and Markets Authority?

Daniel Gordon: I have no introductory remarks, but thank you for inviting us back again. It is a pleasure to be here.

I am senior director of markets. In my area we carried out the market study on digital advertising, which I am sure we will talk about a bit later. I also looked after the proposals that went to the Government in the context of the digital task force for the new regulation of internet platforms.

Simeon Thornton: I am a director at the CMA. I led the study Daniel referred to on online platforms for digital advertising. Currently, I am leading a project looking at Google’s privacy sandbox proposals to remove third-party cookies from Chrome, which I think speaks to some of the issues you may be interested in in this inquiry.

The Chair: Indeed. We certainly looked at that in previous inquiries and again in this one. Lord Stevenson has the first question.

Q175         Lord Stevenson of Balmacara: I add my welcome to you both. In some sense we have been building up to you. We have so many questions that we are anxious to get going, because so much of what we heard from other witnesses has been about how and in what form regulation might be applied. Of course, you are the living testament that you can actually have it, even though statutorily you are not yet in existence. Is that right? I think it would be helpful to get a sense of that activity. You have hinted that you are already moving in this area and it is very exciting to hear that. What can you do, given you have not yet been set up statutorily? Is there anything you can be getting on with other than simply research at this stage?

Daniel Gordon: We have funding to start work with the new Digital Markets Unit, but we still await the legislation. There is no shortage of things for us to be getting on with in the nearer term before the legislation appears.

I put those in two broad buckets. The first bucket would be just getting things ready for the new regulatory framework. There is a lot that needs to be done. We do not want to be in the position of just waiting for the new powers to appear and then start the work. We want to do what we can using our existing powers to build a view of the platforms and do as much as we can in the exercise of designating them—whether they have strategic market status and hence require regulation—and start to work on what the codes of conduct that would apply to them would look like, so that when those powers come in we can hit the ground running.

To do that we can use our existing powers, much in the way of the market study that we carried out last year. That is one very important area that is all building up to the new DMU.

We also recognise that in the interim period we need to do things that take action and have an effect. We have a very active programme in that regard. We have launched two very significant enforcement cases. The first one, which Simeon has already mentioned, is about Google’s use of cookies and sharing them with third parties. As Simeon mentioned, that case is potentially of particular interest to your Lordships, and it might be worth Simeon saying a bit more about it at the end of my comments.

Another case that we have launched is about Apple’s App Store. A number of app developers have brought complaints to us. It is an area that we have been looking at ourselves internally. The concerns are that app developers are reliant on the App Store if they want to be able to sell their wares on an iPhone or iPad. They also need to agree with Apple’s terms and conditions, including up to 30% commissions, and use Apple Pay. That is the nature of that second case.

Those are two big cases that we have launched in the past couple of months. We are also exploring a very active pipeline, which obviously I cannot go into here.

In addition, our mergers work carries on. That involves digital platforms to an increasing amount. Recently, we have had mergers involving Amazon, Facebook and Uber. We have also revised our merger guidelines so that they are better targeted at addressing the harms that we find are becoming more prevalent with online mergers involving the big platforms. We have recently finalised the changes on those guidelines.

Lastly, on our consumer enforcement side we have a number of cases in those priority areas of consumer concern. To pick just a couple, one is on the posting and trading of fake online reviews; another is on the use of incentivised endorsements—hidden advertising in the form of celebrity endorsements on platforms such as Instagram.

That is our programme and it is pretty active in the space in those two areas. I suggested earlier that Simeon might say something about the Google Chrome case, if you would like.

Lord Stevenson of Balmacara: Before he does that, may I press you a little? You say that you are up and running. You do not have statutory powers, but you have funding. Will you give us a broad sense of what that funding comprises and how many staff you have?

Daniel Gordon: As for staff, we have in the order of 55 to 60 FTEs. That is the scale of the shadow DMU at its interim stage. I cannot remember the equivalent pound figure, but I can get back to you with that number, if it helps.

Lord Stevenson of Balmacara: Clearly, it is never enough, but you have sufficient to get yourself up and running. Without wishing to be quoted on that, is that right?

Daniel Gordon: Yes. An exercise is needed to get things up and running. I would not extrapolate straight from that to enforcing and running the new regime. Ideally, we would be looking ideally at more of a step change in resourcing to deliver the programme itself.

Lord Stevenson of Balmacara: Thank you for sharing that. Simeon, do you want to add to it?

Simeon Thornton: We thought it would be of interest to the committee to hear about our Competition Act investigation into Google’s privacy sandbox proposals, which were introduced in the name of protecting users’ privacy and involve important data protection considerations.

It is also worth noting that third-party cookies, which Google is proposing to remove, play a fundamental role in the digital advertising market. It is quite possible that their removal by Google will have a big impact on digital advertising, not just in the UK but globally, reducing the effectiveness of digital advertising expenditure and potentially entrenching Google’s market power.

There are two reasons for wanting to bring this to your attention. First, it is a vivid demonstration of our willingness to use our existing powers to the fullest possible extent pending introduction of the new regulatory regime. In many respects, this is not a typical enforcement case because the proposals have yet to be implemented, but we were clear that, given the importance of these changes and the considerable range of stakeholders who expressed concerns about them, we needed to act, so I think that it is a good demonstration of us pushing the boundaries pending the introduction of the new regime.

Secondly, I think this is a good example of the close relationship between competition considerations and broader social concernsin this case, data protection and privacy. These two agendas are strongly linked, and as a competition authority we cannot afford to take a blinkered view of what is a legitimate competition concern. Indeed, we have worked closely with the ICO on this and we both recognise the importance of considering Google’s practices in a holistic way, acknowledging the impact that any action that we take may have on a wider set of policy objectives.

It seemed to us that this was a relevant case when you are considering the interaction between competition and broader social objectives such as freedom of speech.

Lord Stevenson of Balmacara: Indeed. That is exactly where much of our thinking is going. We will certainly want to explore that further.

As I said at the beginning, this is very exciting. It is good to hear that there are people with real skills, knowledge and power who are able to take a very close look at some of these companies and their activities, but is there not a frustration that you do not yet have statutory powers? Can you shed any light on what the Government’s thinking is on that, because time is moving on?

Daniel Gordon: As the committee will know, the Government have committed to consulting in the first half of this year. We have every reason to expect that to happen; we believe that they will keep to that timeframe. The Government’s commitment on legislation is that that will happen as soon as parliamentary time allows. That is ultimately a decision for the Government, but we have no reason to worry at this point.

We do not want the legislation to be rushed through, but the sooner the power is in place to implement the regime, the better. We are doing what we can in the form of the shadow DMU to minimise any gap between the time those legislative powers land with us and the regime is up and running and implemented.

Lord Stevenson of Balmacara: It is uncomfortable, is it not? From what you say—you do not need to respond if you feel it will be difficult for you to do so—I assume that this will not be in the Online Safety Bill on the basis that that will be the subject of pre-legislative scrutiny after the Queen’s Speech, and that therefore we are talking about probably three years down the track before it gets Royal Assent. Do you agree that your need is more urgent and that therefore it might need to be dealt with in other ways?

Daniel Gordon: I will take your offer of ducking your question. I think we should have priority in the queue from our perspective, but this is tricky stuff. While we have made some quite detailed proposals, it is important that the legislators have got their heads around it and come out with a coherent package. While I think we would be keen that it happens sooner rather than later—in the timeframes the Government have been indicating—we would also be nervous of rushing through legislation and having to pick up the pieces subsequently.

Lord Stevenson of Balmacara: That is a good point.

Q176         Lord Lipsey: I wonder whether we might explore the role of competition. We have all been brought up in the modern world to think that, naturally, more competition is a great thing, but there are worries about what competition could do in this field. Donald Trump, once thrown off Twitter, could get somebody else to carry on his abusive and insulting remarks. Will you briefly set out your case for thinking that competition is a form of making progress in this field?

Daniel Gordon: It is a very open question. I would refer back to our primary duty, which is to promote competition for the benefit of consumers. If it is not benefiting consumers, that is not in our duty. Competition is subordinate to the benefits of consumers, so in any assessment that we did we would not be promoting rivalry or the number of players just for the sake of it; we would be looking for a framework that led ultimately to benefits for consumers. We have been willing to depart from what might conventionally be described as promoting competition when that is not the right way to go—for example, by putting price caps in place in certain markets where it is not being effective.

Having said that, it is probably worth exploring what is within the consumer benefit label. We tend to take a very broad view of that, and I am happy to expand on it if you like.

Lord Lipsey: It is not clear that freedom of expression, which is the subject of this inquiry, is of benefit to consumers as a whole. Freedom of expression may lead to some consumers being totally abused, or taken to pieces by floods of people using social media to attack them. Therefore, does freedom of expression come within the definition of consumer benefit, which you say is your objective to advance?

Daniel Gordon: There is a range of issues that all of us are facing in the regulatory sphere, and they are coming simultaneously because of the rapid progress of these platforms. We are all finding out new things and different types of harm. They are very new forms. Simeon has alluded to the important issue of privacy, and the one that you are referring to concerned with online harms is also a parallel sphere. As you know, the Government are currently consulting on legislation in that area—we have mentioned it already—which they think Ofcom will implement.

There are potential overlaps and we are very mindful of those interactions, but it is equally important to be clear about the objectives. As regulators we are independent organisations answerable to our boards and driven by the powers that we get from Parliament, so we need to have those quite clearly demarcated. The important thing is that they work together, and we are doing a lot with other regulators to make that happen so that they link together.

I think that some of those issues would be for a competition regulator, but not all of them.

Q177         Viscount Colville of Culross: I was also going to ask about freedom of expression, but I feel that my question has already been asked by my colleague.

The CMA and other competition regulators have well-tried theories of harm, but freedom of expression is not a criterion that can be used when looking at harm. How difficult would it be to change that and to include it as a criterion for competition regulators?

Daniel Gordon: To refer back to the answer I have just given, one of the important differences to note between the plurality and public good elements of these harms is the question of what consumers want versus what they should get. The bits that they should get are more in the plurality and media sphere; the bits that people want lie more in the competition sphere.

That takes us neatly into the points that you raise. Our interpretation of consumer benefit, which will have begun to become clear from Simeon’s earlier comment about the Google case, is wider than you would see in many other jurisdictions. I think we are willing to go wider. It is definitely not just prices and quantity, which some jurisdictions do keep quite slavishly to. We spend a lot of time focusing on quality and innovation, which is so important in these markets.

Having seen some of the previous testimony that you have received, I am aware that one of the concerns has been whether we can deal with privacy issues fairly, and that aspect of freedom of expression. We can say that not only can we do so but that we have done so. A critical argument here is in relation to social media platforms. In the first instance, to take Facebook, that might be seen as a must-have item. They also, however, have a take-it-or-leave-it offer when it comes to what data you share with them and the terms that you sign up to. That is a potential example of a lack of competition hampering aspects of freedom of expression. That exact fear of harm and the analysis of it was one we explored and concluded on in the market study. I think that indicates that we feel able to get within those sorts of issues in our current powers.

I would add a rider. This is all new territory for competition policy. These are not issues that you see when dealing with bricks-and-mortar competition. A lot of these issues have been untested in courts. Therefore, when making our recommendations in relation to the Digital Markets Unit’s powers, we felt that it was really important to have clarity about what could be considered within the consumer benefit area. We recommended in the digital task force that it is the duty to promote the interests not just of consumers but of citizens. Whatever the position there, it is really important that there is clarity of expectation of the issues that the Digital Markets Unit can get into so that we can deliver on them.

Viscount Colville of Culross: Simeon, do you have anything to add?

Simeon Thornton: I agree with Daniel’s answer. In the framework within which we operate, we try to consider a broad range of impacts, one of which Daniel alluded to: privacy.

In addition, in the work that we did on digital advertising, front and centre of our concerns was the impact that a well-functioning market had on sustainable news media and a variety of high-quality news media. I think we can take those considerations into account, but, as Daniel says, in the context of a new regime where the DMU will, we hope, be taking decisions over a shorter period, we think that it is worth clarifying the extent to which it can consider those broader issues.

Q178         Baroness Grender: I agree with Lord Stevenson that it is great to have you up and running, so to speak. Let us hope that the legislation comes sooner rather than later, although you will have heard in previous sessions our scepticism about the pace of it.

I want to talk a little bit about walking in the shoes of the platforms and their design issues. We all understand that this is a design and business model that in a way has got away from itself, because the measure of your success is how many people you manage to capture and hold and continue to stimulate. That was made very clear to us by a previous witness, Ross LaJeunesse, formerly of Google, who is highly critical of that business model.

Another very interesting witness was Jimmy Wales of Wikipedia, which does not drive that kind of business model and is a very different operation.

Can you talk to us about what you think the challenges are for the platforms in their design and how they can mitigate consumer harm?

Daniel Gordon: Hearing you relay the comments of the first witness, in some sense those properties are common to many markets. Companies want a brand and to keep loyalty; they invest in their brands to get consumers and then keep them. We are not in principle averse to that. We are averse to it being done under false pretences, or in ways that the consumer is not very clear about and is being caught out.

What is also most important in these markets is the position of the companies that are carrying out those sorts of activities. In particular, what is most striking about the biggest platforms is their entrenched market position. This was the key finding in the market study that looked at digital advertising and hence at Google and Facebook. What matters there is that, if you have a platform that has a lot of consumers and is keeping hold of them because it is doing very well, but another platform could come along and offer a better service and replace that platform, that is potentially a competitive model that we could live with.

You could argue that that characterised these markets in the early days: Google replaced Yahoo; Facebook replaced Myspace. Those examples are looking rather long in the tooth now; they are almost prehistoric in the timeframes we look at in these markets. What has happened in the interim is that the biggest platforms have entrenched positions. When they have entrenched positions it is due to their position with data, their scale and network benefits. It is to different degrees with different platforms, but they tend to reinforce each other.

When the platforms have that position, that dynamic does not work. You expect them effectively to be predisposed to exercising market power. That means the markets tend not to correct automatically; they are not self-correcting. A typical approach with competition policy is that you would find a harm—an exclusive arrangement or a cartel, say—and you would address it and expect the market to go back to a broadly competitive state once you have done that.

When the companies start from such an entrenched position you cannot expect that; you expect the harms to show themselves in different ways. That is why it is often described as a game of whack-a-mole with these platforms. You have the same forces driving the harms and you are just looking at different ways to address that. That is why an enforcement route has benefits, but it is a bit like the hammer on whack-a-mole. The regime that we are proposing is about recognising that and addressing the harms before they show themselves in the form of a code of conduct, unlocking the source of that market power through having the powers to put in place pro-competitive interventions and things such as data interoperability, addressing defaults and those sorts of things.

I would hope that when we have those measures in place we will begin to get a very different landscape, or at least the current landscape will begin to show itself quite differently in how it serves consumers.

Baroness Grender: Simeon, there has been a lot of concentration so far on the business model issue. I just wonder whether you could direct your comments particularly to the platform design issue.

Simeon Thornton: The distinction that you are drawing between business model and platform design is what precisely?

Baroness Grender: It is about designing out harms, but that may militate against your business model. You have some of the stuff that we have rehearsed well about driving people towards YouTube and getting people deeper and deeper into various extremes of opinion, for instance. What are the designs that we are expecting platforms to introduce to try to mitigate those harms?

Simeon Thornton: How about if I start with a bit of a deep dive on one particular business model that is probably front and centre of your thinking, which is Facebook? In relation to your original question, the incentives to keep people on the platform so that you can monetise their attention play out much more clearly for Facebook than, for example, for Google in relation to Google Search. With Google, you want to get someone putting in an entry. If it is a monetisable entry, fine; they will click and that is it. It does not have an incentive for people coming back to Google Search except when they need it.

For Facebook, the incentives are different. In understanding those incentives, I would agree with Daniel’s point that we need to understand that they are in a position of market power and how that impinges on the ability of users to move away if they are not happy with what Facebook is doing. Facebook has a share of about three-quarters of the total time spent on social media in the UK. It is able to monetise that time much more effectively than any of its rivals by selling digital advertising. We found that it earns about £50 per user per year, which is about 10 times more than its nearest rival.

The reason for this is that it has two big advantages. One is what we call network effects—a somewhat technocratic term given to the fairly simple intuition that a platform is more attractive to a user if that user’s friends and family are already on that platform, self-evidently.

The other is that Facebook has access to vast amounts of users’ data, which it can use to target that digital advertising, meaning that its digital advertising inventory, to use the jargon, is more valuable than that of its rivals.

That is its business model. Almost 100% of its revenues is digital advertising, and that drives the incentives that you alluded to about keeping people on the platform and targeting the advertising in the way that is most valuable.

Thinking about alternative models, the CMA tends to take the perspective that it would facilitate choice for consumers if they are not happy with those business models. That is an effective way of promoting other business models that might be more conducive to consumers’ interests. When you look at Facebook’s business model, the key way to overcome those two key advantages is, first, how to overcome the network effects.

One approach that we suggested was increasing interoperability between Facebook and its rivals. That is a key area. I can talk about that in more detail if there is interest. The other main approach would be to tackle Facebook’s big data advantages by giving consumers greater choice over whether they provide data to Facebook and requiring platforms such as Facebook to design choice architecture in a way that facilitates genuine, free and informed choice and other measures to level the playing field for access to data.

The way in which we conceptualise it is not so much about how we can mandate a business model, because I do not think that is our USP; it is about how we can facilitate informed choice by consumers and ensure there are legitimate choices that consumers can exercise. The exercise of that choice might be in a number of directions. In this business model it is not on price. It might be on quality, it might be on privacy, it might be on standards such as the way in which a platform oversees free speech. That is absolutely key, certainly to our considerations and we would hope to your considerations as well.

The Chair: The companies that we are talking about operate on a wholly global stage. You work with EU and other regulators. I would now like to talk about some areas of international cooperation and lessons to be learned from around the world, starting with Baroness Featherstone.

Q179         Baroness Featherstone: In a way, Simeon addressed the first part of the question about entrenchment and market share that is so omnipotent. It seems to me that it becomes a self-fulfilling prophecy: the more consumers you have, the more they want to be with other consumers and no one can get a look in. I would be interested to know a bit more about interoperability and what that might look like, because that makes it very monopolistic. Unless that is changed, I do not see how this will ever shift.

The Online Safety Bill is coming before Parliament and potential new regulatory requirements. They may add to the ability of small entrants to the market in the sense that regulation might be more handleable by the big, entrenched monopolies than it will be by people trying to break into the market. We heard testimony in an earlier session that mandatory moderation might mean that the only option for new players would be to contract out their moderation to people like Facebook and storage to Amazon. You mentioned the App Store. That seems to be quite a hole to go down.

How can the DMU best promote competition in that social media market? You have touched on some of it. How will you do it and make people more interested in privacy than they are in being with their mums and dads on Facebook? As the Chair just mentioned, looking at international examples, where can we find something that might work?

Simeon Thornton: Thank you for the question. Let me expand on two areas that I alluded to in my previous answer. The first is the one that you highlighted: interoperability and how that works.

The objective of mandating increased interoperability in Facebook and rival social media platforms is to enable those positive network effects to be shared with other platforms. To get more specific, in the study we identified two clear cases where we thought that there was a strong argument for the DMU to mandate increased interoperability in this way. First, that was what we called finding friends functionalities, which means that a user could find it easy to contact their existing friends on an existing platform and invite them to join the new platform. In other words, you have contacts on Facebook; for whatever reason, you are fed up with the service Facebook is offering you and you join a new platform. You want to make it easier to invite those on your existing network to join that platform. That is called finding friends.

The second one that we identified is called cross-posting functionalities, which effectively means that you can choose the social media platform that works for you but post on different platforms to access your existing community. We thought that those two measures alone could have quite a material effect in overcoming those network effects and in improving competition in social media.

There is doubtless a whole range of use cases that the DMU will want to consider, some more complex than others, but underpinning all the complexity is a very simple overarching objective: users should be allowed to choose the social media platform that works best for them and their needs—whether that is quality, privacy or freedom of speech—rather than obliging them to choose the platform that their friends and family currently happen to use.

That is the overarching objective of interoperability. As I said, we have identified a few use cases where we think that there is now a strong case.

The second area is about data-related interventions. You asked how we can make people more interested in privacy if they are not. What we found when we looked at the literature is that, generally, when you ask people they say that they want more control over their data, but when we observe the extent to which people actually engage with existing privacy controls we see that it is very minimal, so there is a big disparity between what people say they care about and what they do.

One interpretation of that disparity is that platforms have an incentive to make the choices difficult to exercise freely. In terms of choice architecture and the default options that platforms provide, they steer users to act in a way that is consistent with the platforms’ interests rather than individuals’ interests.

One thing that we advocated is that users should have a choice over whether to provide their data for personal advertising. If you look at most social media platforms, currently they do not give that choice, particularly where you have a social media platform that has market power, because—remember, all your friends and family are on that platform—you have Hobson’s choice. If you want to contact them, you have to accept the terms and conditions. If those terms and conditions are that you have to provide that data, you really have no choice at all.

That was why we felt that one really important point of intervention would be to require that choice to be provided. We thought that that would have the effect not only of addressing competition and privacy concerns but of changing the bargaining relationship between users and the platforms, because then the platforms would have the onus to persuade the user to give them the data, if they want to, and that would open up the space for incentives—financial incentives, less advertising or different terms and conditions—and we can move away from this kind of default position in which people are taken for granted and they have to be engaged with by the platforms. We felt that that was really important.

Baroness Featherstone: I think that is very clever, actually. What about international models and the regulation making it more difficult for new entrants?

Simeon Thornton: It is a really interesting time in the international anti-trust community. These issues are being debated all over the world, and I think that we are one of those at the vanguard of that discussion.

As for a specific example, you might wish to look at the German BKA investigation into Facebook’s excessive use of data. I think it was one of the first to make a case that the terms that Facebook imposes on its users are unreasonable and excessive. That is an interesting example.

More broadly, the interface between competition concerns, consumer interests and data protection will be an increasingly important focus of the work not just of the CMA but of competition authorities around the world in years to come.

Baroness Featherstone: I do not know whether you want to add something, Daniel.

Daniel Gordon: The general message there is the importance of recognising that concepts such as interoperability are very powerful, but, as Simeon alluded to, when it comes to implementing them you have to think about and understand the business model that works in the circumstances.

That was one of the areas that we were very keen to get into in the market study and task force proposals. The world is moving in parallel—this is linked to your third question—in getting up the curve and understanding the nature of the problems. Devising remedies that will work is more complicated. That is the territory that we have been getting into to a greater extent in the context of the market study. Clearly, that is what matters at the end of the day for sorting things out.

On the international stage, as Simeon says, our work is getting a very good reception, in part because our powers have allowed us to get under the skin of some of the companies in ways that other countries have not been able to. For example, the US Senate and Congress have been drawing on those. There is a parallel path being trodden; you will see it in the reports that have come out in the past few years, and we are all trying to work out the best approaches to remedying this.

Your point about regulation and entrants is a really important one. We have heard it said that the implementation of GDPR might favour the bigger companies. Part of the concern is that the interpretation of regulations, privacy or whatever, that the big platforms put on those becomes pivotal, so if there is room for interpretation they effectively become the regulator. Those are very important issues.

We would hope to work with other regulators in the Digital Regulation Cooperation Forum and other setups to be able to recognise those issues before they take effect, because it would be far from ideal if conforming with the regulation had a further impact on competition.

Baroness Featherstone: If you ask me, both of you have a job and a half.

The Chair: Interoperability is very interesting. You have described how it might work and given some specific examples. There are other issues that we want to move on to now, but maybe you can direct us to some useful reading on interoperability and how it might work in practice that we could study. You must read through volumes of material. If you think that there is anything you could curate that would be of particular interest to us, that would be useful.

Q180         Baroness Buscombe: This is really interesting and helpful. I am now moving on to a similar question in relation to search engines themselves. How can the Digital Markets Unit best promote competition in search engine markets, given that to be effective search engines need to have access to past consumer queries on a massive scale, I suggest, and an up-to-date web index? How will this be possible? Google, for example, is often described as unassailable as well as unaccountable. With each and every move it makes, it seems clear that it is further entrenching that extraordinary power. From which international examples can you learn to tackle this difficulty in the Government delivering on the aim to have fair competition?

Simeon Thornton: You are absolutely right. If you want to understand competition search, you have to understand the position of Google. Google has over 90% of searches and over 90% of search advertising revenue in the UK. It benefits from two big and important barriers to entry and expansion that limit the extent to which competitors can compete with it. The first is the one that very rightly you alluded to, which is that it enjoys economies of scale in data. The more users you have of a search engine the more click and query data you have access to and the more you can use that data to train and improve your search algorithms, attracting more users to your search engine. There is a virtuous cycle there or, from the perspective of rivals, a vicious cycle. Having been big in the past is a strong driver of being successful in the future.

The second important one is the fact that Google undertakes activity to secure its position as the default search engine on mobile devices and browsers. Defaults, as we all know, are very important in driving consumer behaviour, and Google spends a lot of money—over £1 billion in the UK alone—to ensure that it is the default search engine on mobile devices and browsers. This drives more users to use Google and adds to the selfreinforcing nature of its market power, making it very difficult for rivals to compete. When we thought about potential interventions we focused on those two areas.

In the first case, we think that there is strong argument for the DMU to require Google to provide third-party search engines with users’ click and query data to ensure that those wider search engines can benefit from the economies of scale that Google currently enjoys by dint of being the largest search engine. The mechanics of this need to be considered, but we think that it can be implemented in such a way as not to raise privacy or other concerns.

Secondly, to address this default position that Google enjoys on mobile devices and browsers we think that there is a good case for the DMU to consider restrictions on Google’s ability to secure default positions. This could take the form of a requirement to introduce choice screens, for example, that would allow consumers to choose a search engine that it wants to be their default rather than it being imposed upon them by the mobile device operator.

We think that in combination these two measures have a strong prospect of helping to tackle the sources of Google’s market power and give rivals the opportunity to compete on an equal footing.

As for international experience, this is an area of live debate and we are engaging fully with our international counterparts in steering it. I think that the work we have done on Google and Facebook has been influential—for example, it was cited in recent US Senate and Congress inquiries.

One example of a successful intervention to tackle Google’s market power comes, interestingly enough, from Russia. In 2017, a choice screen was introduced on Android-operated mobile devices by our equivalent body in Russia. Yandex, the rival search engine in Russia, told us that this remedy was very effective in providing users with good choice and improving competition between search engines. Yandex’s market share in Russia is about 40%, which is much higher than you see for rival search engines in countries such as the UK.

The important point to make is that these interventions are not just theoretical constructs; they can and do work. They require a lot of focus and attention to the design to avoid unintended consequences, but we think there is a strong case for them.

Baroness Buscombe: Is also one of the issues the massive amount of investment required, or am I being too pessimistic in that sense? The idea of building something to compete with a search engine such as Google sounds almost like—they will hate me for saying this—privatising a state monopoly, almost an international state monopoly. In the early years we were all afraid of being seen to be uncool in taking on this extraordinary and in many ways exciting innovation. We did not want to be seen to be not innovative, but from a financial standpoint is this not a huge ask as well?

Simeon Thornton: There are traditional economies of scale certainly in search. I do not think they are insurmountable if you tackle the data economies of scale and the default positions. Let us not forget that there are other very large entities in the world, including some of the large platforms that would have the financial clout to do this if those other barriers to entry could be overcome.

It is always important not to forget that Google came out of nowhere by offering an excellent product that was at the time manifestly better than the competitors. All of us of a certain age can remember the first time we used Google and thought, “This is better. I’ll use this”. That is great, is it not? Two people with a great idea can challenge.

Our concern is that that is no longer the case, because history becomes important and size begets size in the future, and the next few people with a great idea in a garage somewhere cannot really hope to compete with Google, unless you implement this sort of regime that allows Google’s benefits in data to be shared with others.

Baroness Buscombe: Daniel, do you want to add to that?

Daniel Gordon: I would agree. I hope that it is not the case. We are saying that there are things that we should try and they are potentially quite powerful things to address that market position. At the end of the day, as Simeon says, unlike other entities where there are ex ante economic regulatory regimes, there is not physical infrastructure here. A huge amount of money does not have to be spent on putting pipes in the ground and that sort of thing; it is just not cost-effective to replicate in principle. We will find out in practice, but from where we are we should at least be optimistic about that.

Baroness Buscombe: I know others are trying to do it in a way that offers more privacy and so on, but that is very helpful.

Q181         Baroness Rebuck: Changing the subject completely, my question is about Australia and its mandatory news bargaining code. As you know, in our report we talked about the imbalance of power between the tech companies and newspapers. Simeon, you referenced dominance in advertising, which has seriously eroded revenues for newspapers. Indeed, the UK Press Gazette has been campaigning since 2017 for newspapers to be paid for their content.

We were told by witnesses that several publishers negotiated agreements in Australia before the code became law and that some deals had been signed in the UK. All the witnesses from the newspaper industry said that the kind of deals that they were able to get were possibly less than they should have been able to negotiate precisely because there was not a threat of legislation and, therefore, that imbalance of power. One witness said that some of the regional newspapers would take whatever was on the table because of their cash needs.

I am interested in your assessment of the Australian mandatory code and your view of the claim by the Canadian Heritage Minister that several countries might be looking at similar legislation.

Daniel Gordon: We would have broad agreement with the general objectives of the Australian approach. This is what we said in our market study and in the task force proposals. A code is necessary and needs to be put in place to ensure that there are fair and reasonable terms between the platforms and the businesses they interact with. It is not just newspapers; it is advertisers and other content providers, and newspapers are obviously particularly prominent in the discussion, as we have seen in Australia. Therefore, we would definitely agree with that.

The central question is what sort of approach you take to reach that outcome and get those fair and reasonable terms. You might be aware that at the launch of the shadow DMU a couple of weeks ago the Secretary of State commissioned the unit to look into exactly this question to start to explore what approaches a code might take to addressing that. We will be talking to content providers, newspaper groups and others to understand these things more.

The Australian experience is definitely a good place to start. We would agree with other important features of it. The first is that the code is statutory. We have said that the code of conduct needs to be enforceable in law, and that relates not just to newspapers groups but to the interactions with other businesses and all the other terms that would be in the code of conduct.

The second element that we would probably veer towards at this point is that it makes a lot more sense in this area for the regulator to be looking at determining what is a fair process rather than itself trying to second-guess these complicated transactions between businesses that it does not run and to impose terms.

The question then is: what sort of model of fair process would we be looking at? We will be looking in the course of the work that we are doing at the range of models out there, including the Australian one.

There are other approaches to look at as well. I fear that this is a recurring theme in our testimony, but across the world we are just learning in these areas and trying to work out the best approaches and learn from each other. There are other instances to look at. Simeon has already mentioned Germany. There is German legislation that is relevant to this point; there is experience in Spain that did not go too well but is useful to know about and learn from; and, similarly, measures are being put in place in France. We will look at all those things in the course of this work.

Baroness Rebuck: That is encouraging. Simeon, was there anything in the original draft of the Australian legislation that was edited out in the final piece that you think is worth looking at? I was thinking of the ability of news organisations to remove or filter user comments, which bizarrely has now been introduced by Facebook, but I believe that in the first draft there was also provision for more transparency and accountability of algorithms. You have talked a lot about algorithms in a different area, but that was taken out. I think that all that is left is that they have to inform these organisations a few days before they change something. Do you have a perspective on that?

Simeon Thornton: These were the sorts of issues that media organisations brought to us in the context of the study. To take them in turn, on clarity around algorithms it is understandable that news media organisations, for which Google is a very important source of traffic, will want to be able to plan and have confidence that when Google makes changes to its algorithms it is not motivated by nefarious commercial considerations. Clearly, algorithms will change; it is the nature of those algorithms that they must change over time. The position that we have reached in relation to algorithms is that there may well be merit in the DMU being able to scrutinise those to see where there are considerations being taken into account that should not be—for example, commercial considerations.

There is probably a limit to the extent to which you could legitimately expose all the underpinnings of those algorithms to all third parties. Going back to the pre-history of the internet and some of the results that you used to get by clicking on Google Search, when the algorithm is very clear to all parties, people can game it by putting in the relevant key words and coming up top of the list and it is a load of rubbish, so there is a legitimate reason for not wanting to expose all the workings to the world, but that is where trust in an intermediary such the DMU can come in. It can look at the algorithm and make sure that it does what it says on the tin rather than third parties concluding that Google must have done something for a particular reason. There is a role there for the DMU that is really important.

I entirely agree with what Daniel said about the right approach to negotiating payment for content. All I would add is that there are some subtle ways in which you might be able to shift the bargaining position between platforms and news media organisations and tilt the balance a bit in the news organisations’ favour relative to the current situation. One of the examples that we gave was: why not give the news media organisations greater influence over the way in which their content is provided on search results?

At the moment it is fairly holistic; you get the whole article and there is an incentive for Google to keep people on its entry system looking at the article. But if news media organisations had an opportunity to say, “You get only a small snippet or bit of information”, that might tip the relationship and give Google a greater incentive to bargain in good faith over payment for content.

I agree with and reiterate what Daniel said, which is good. It is very difficult legitimately for a regulator to say, “The price for this content is X and the price for that content is Y.

Baroness Rebuck: I understand that.

Simeon Thornton: It is a very difficult question, but the more that we can create a situation in which Google has genuine incentives to bargain the better.

Baroness Rebuck: Excellent. That is very encouraging.

The Chair: Sadly, we are out of time. Daniel Gordon and Simeon Thornton, thank you very much indeed for your evidence today. You have promised to follow up a couple of issues that we have discussed, which we would welcome and would add to the committee’s reading. It is kind of you to do that. Thank you both very much indeed for all the work that you do in giving evidence to this committee and engaging with us.