Select Committee on Democracy and Digital Technologies
Corrected oral evidence: Democracy and Digital Technologies
Tuesday 4 February 2020
Watch the meeting
Members present: Lord Puttnam (The Chair); Lord Black of Brentwood; Lord German; Lord Harris of Haringey; Lord Holmes of Richmond; Baroness Kidron; Lord Lipsey; Lord Lucas; Lord Knight of Weymouth; Baroness Morris of Yardley.
Evidence Session No. 10 Heard in Public Questions 124 - 135
I: Pascal Crowe, Data and Democracy Project Officer, Open Rights Group; Ailidh Callander, Legal Officer, Privacy International; Jon Lloyd, Head of European Campaigns, Mozilla Foundation.
Examination of Witnesses
Pascal Crowe, Ailidh Callander and Jon Lloyd.
Q124 The Chair: Good morning and welcome to the Committee. I am sorry about the slight faffing around, but that is the nature of committees. I have to make a formal statement to begin with. As you will know, this session is open to the public. A webcast of the session goes out live and is subsequently accessible via the parliamentary website. A verbatim transcript will be taken of your evidence and put on the parliamentary website. You will have an opportunity to make minor corrections for the purposes of clarification or accuracy.
Q125 Lord Black of Brentwood: I start with the question of the information that the public should have access to about political campaigns. We touched on it at the end of the last session. We all know the public have limited attention for these matters and probably limited knowledge about detailed digital issues. From your point of view, what are the most important aspects of political campaigning, about which the public should be made aware? You can look at a number of different areas, if you want: where the money is coming from and what it is being spent on; data, or what campaigners know and how; and then speech, or what is and is not an advert.
Ailidh Callander: That is a really important question, because it gets to the heart of transparency. You can also look at who is providing that information to the public. It is important that, if an individual wants to know, they have the information that explains to them why they are seeing a particular message. That requires clarity on what sources of data were used to target that message at them and full targeting criteria to be available for why that message was targeted at them. They should also have information about how to exercise their rights and how to prevent such targeting, and access to information about the cost of that targeting.
We also need information and transparency from other actors, such as political parties, about the tools and services they are using; the companies they are contracting with, for what services and at what cost; the messaging they are using; granular information about the data, from the sources to what they are doing with the data, who they are targeting and under what criteria; and the campaign groups and third parties with which they are affiliated. We need much more information about how companies are targeting information that is not ads, pushing content and recommendations. All this information must be available in a persistent and continuous way, because we need information available to us as individuals and voters, but also for further study, so that others, including civil society, academics and journalists, have access to it, because transparency is also a tool for accountability.
Pascal Crowe: I am the data and democracy project officer at Open Rights Group. On the kind of information the public should have access to, there should be wider recognition and an information push to ensure the public understand that this problem of data-driven political campaigning is not limited to the use of political ads on social media. In many ways, political ads are the end of a production line that begins with electoral roll data taken by the political parties, which is mixed with sources of commercially available data, such as Experian credit history information. That is processed and turned into a profile, which is used as a basis on which to start uploading to Facebook, mixed with information on social media. That then gets turned into ads that are served and targeted.
Secondly, there needs to be a wider understanding among the public that social media platforms are not the sole arbiters of our democracy. Political parties bear a significant responsibility in this as well. They are the customers providing the initial sources of data and data profiles. Really, it is difficult to expect political parties to regulate against their own interest but, in many ways, the responsibility lies at their door as much, if not more, as it does for social media platforms.
Jon Lloyd: The question of political campaigns’ transparency is fundamental to democracy and covers many issues. While Mozilla supports strong transparency requirements on all actions by political campaigns, our work this year has been, like my remarks today will be, focused on online political advertising transparency. I reinforce the point that Ailidh has made about disclosure to users on the platforms and clear labelling obligations, including why users are seeing ads. That is essential to this. While user transparency is good, it is putting too much pressure on an ordinary person — you or me — to protect ourselves online.
We advocate for an obligation on the relevant platforms to disclose all political and issue-based ads, and publicly available ad archives. These archives should generally include information about the advertiser, how much they spend and any information available to advertisers themselves. More specifically, several pieces of information are critical to making these archives successful. First, these repositories should include the advertisers’ targeting parameters. This is necessary to understand how information flows across these platforms and can be disclosed without any substantial privacy risk. Platforms have thus far been unwilling to disclose this information and we do not find their arguments against disclosure to be particularly compelling.
Secondly, it must include all demographic information about who is seeing and clicking on ads. This information could pose some privacy risk if disclosed incorrectly but, to avoid this risk, disclosure obligations should focus on providing aggregate, high-level statistics about protected classes and demographics. Thirdly, the Government should seriously consider requiring all ads to be disclosed, not just a narrow class of political ads. Focusing on political ads allows adversaries to game the system to avoid disclosure and puts platforms in the untenable position of deciding what a political issue ad is.
Lord Black of Brentwood: On the digital literacy and education point, I think everybody here would agree with what Pascal said. On access to data, disclosure and user transparency, the question is how it is done and who does it. There are a number of bodies in this area, the most obvious of which is the Electoral Commission. We have heard evidence from the Advertising Standards Authority relating to its possible role in this. How can we best go about this? Could the existing bodies be given the power to do it or is it going to need a new system of regulation? That would require legislation, which is not going to be quick.
Jon Lloyd: The upcoming duty of care for Online Harms Bill should include a code of practice for online targeting, and any new regulator should require full ad transparency, as a way to address targeting practices.
Ailidh Callander: The focus should be on the implementation and enforcement of existing frameworks, as well as updating them. Transparency is a requirement of data protection law as it stands, and we have some mechanisms to make this meaningful. Enforcement of data protection law is key to all the questions the Committee is looking at. How do we make that transparency more granular and mandate it? For example, the ICO code of practice, currently in draft, is one tool, but should be made a statutory code, so that that transparency can be mandated. As well as making that principle a reality, so mandating transparency for the different actors involved in political campaigning, we can look at electoral law and disclosure requirements in terms of the detail of transparency about campaigning activities.
Q126 The Chair: Jon, can I ask you one favour? You mentioned that non-disclosure by the big corporates is an issue, and that you regard their arguments as not being particularly valid. It would help the Committee a good deal to understand your counterarguments. This is a battleground, and I would love to get my head around why the companies are hiding behind privacy considerations, which in fact are not really valid.
Jon Lloyd: It is important for the transparency of political ads that we should not stop at how users are being targeted by advertisers. A recent study from Northwestern University in the United States showed that Facebook’s algorithms can dramatically skew the delivery of ads, even if the advertiser did not intend it. That it is why it is critical to analyse both the potential reach of a political ad, so the criteria chosen by the advertiser and how much the advertiser spent, and who their political ad actually ended up reaching. That is a function of the platform’s algorithm and is why we are advocating for those disclosure obligations.
The Chair: By penetrating the algorithm, how much commercial damage would we, collectively, potentially do to the companies?
Jon Lloyd: I do not think increased transparency through a comprehensive ad archive will fundamentally undermine the business model.
Q127 Lord Knight of Weymouth: To follow that up, my understanding is that the Cambridge Analytica scandal—or whatever we choose to call it—came about in part because of researchers being able to access Facebook data and, to some extent, understand the algorithm better. Is there a risk that, as transparency increases understanding, bad actors understand how to circumnavigate and exploit the platforms?
Jon Lloyd: Cambridge Analytica is an example of only partial transparency. Advocating for full transparency can have the opposite effect. Pascal was talking about data being a supply chain. As the food and fashion industries have been transformed by the transparency of their supply chains — for example, we now know that children are not stitching our shoes — we need to look at transparency in this case as a means to an end. It is to help us identify the intervention points to make the entire system stronger, rather than the thing we are aiming to achieve in the first place.
Q128 Baroness Kidron: My question is somewhat narrower and I accept that it sits in the bigger case that you have all made. Last year in May, the Government committed to introducing digital imprinting by the end of the year. We had an election and it did not happen, and we have not yet heard anything post-election about what exactly they intend. I accept that Jon has answered some of this, but I would love to hear what features digital imprints and ad libraries should have, to be useful. I was interested in the point about where the boundary between a political and non-political ad is, and whether imprinting is an effective tool beyond political advertising. Maybe you could start, Jon, as you have said a lot about this, and then we will move on.
Jon Lloyd: My answer is going to focus on ad libraries themselves, as they were the centre of our advocacy campaigns in 2019. We do not yet have a detailed position on digital imprints, at this point, but the key transparent requirements that I mentioned earlier would be a good start. On the ad libraries specifically, Mozilla, along with 10 other researchers, helped create a set of guidelines that lay out the minimum standards of what an ad library or API could look like. Last year, we focused on political content but, as I mentioned, there is an argument for all advertising content to be included. For us, it would include comprehensive advertising content, the content of the ad itself and information about the targeting criteria, which I mentioned. That could be the text, image or video content, information about where the ad appeared, the targeting criteria used by the advertisers to design the campaign, and then who the ad actually reached.
It also needs to include the functionality to empower and not limit researchers to do analysis. That means each ad needs a unique identifier associated with it to allow a trend analysis of the advertiser over time. All images, videos and other content need to be in a machine-readable format accessible via some kind of programmatic interface. We need the ability to bulk-download data for all relevant content and search functionality by the text of the content itself. We also need up-to-date and historical data access, which means the availability of ads within 24 hours of publication, the ability to search for ads going back 10 years and that the tools researchers use are promptly fixed when they are broken. Most crucially, we need public access, so that any tool or data collected from that tool should be accessible to anybody in the general public.
It is also important to note that we assessed Google’s and Facebook’s ad archives, according to the guidelines that I have just outlined for you, ahead of the 2019 European Parliament elections. Google’s ad API met four of the five minimum standards, but with a few important omissions. Namely, there was no information on its targeting criteria, which is vital; nor does the ad API provide engagement data, so clicks, essentially. That targeting and engagement data is critical for researchers, because it lets them see what types of users an advertiser is trying to influence and whether those attempts were successful.
Facebook met only two of our five minimum standards. Like Google, it failed to produce targeting and engagement criteria. It is impossible to determine if it is providing comprehensive ad data, because the current ad API design puts huge constraints on researchers, for example search rate limits or the fact that, a lot of the time, the archive did not work at all. These findings were highlighted by the New York Times last year and backed up by conclusions made in a similar study by the French Government.
Ailidh Callander: I will just add to what Jon has said about what is required in those ad libraries and the current deficiencies. Much more information is required about the ads that are there. We also need to ensure more ads are captured in those archives. The platform is a piece of the picture, and there are many tools and ways that we can be targeted with messaging. This is an important step, but cannot be the only one. It has to be seen as part of that wider picture. That is where it is important to ensure transparency from other players, including political parties.
For example, we and other civil society organisations wrote to all the political parties in the UK, in advance of the December election, asking them for transparency and 10 clear points that link back to data protection and electoral law, but none of them was forthcoming. None was proactive in its transparency about how data is used in conducting its campaign. We need to work on improving the ad libraries and ensuring they include all the ads and the information that we need. We also need to think about that wider picture and transparency from all the actors.
Pascal Crowe: The Electoral Commission has been calling for digital imprints since 2003. It is frankly incredible that we have got to this point without any kind of parity, at a minimum parity between online and offline political ads.
Baroness Kidron: I am curious that neither of you mentioned the imprint itself. Is that because you think it is useless or because it is a simple matter?
Ailidh Callander: In this case, it is important to listen to the regulator, which is an expert in this. As a minimum, the regulators must get the resources they are calling for, but that would only be a tiny step in the bigger picture. It is one thing that should and can be considered, but we need to think more widely than the imprint.
Q129 Lord German: Is there a case, in your view, for extending the current 12-month regulated period before a UK general election? If there is, would you counter that with the costs that would fall on those bodies, including political parties, which would have to find the necessary funding, as there is a cost to extending it beyond 12 months? This only occurs in UK general elections, not other elections in the UK.
Pascal Crowe: Yes, there is a case for that. Part of the reason is that so much of the data gathering and preparation for the campaign happens outside the regulated period. We need a way to make sure that the regulators in particular can hold political parties to account over their use of personal data, the value that might bring to their political campaign and ways it might help them save money in other areas of their campaign. For example, we have proposed not just that the regulators can conduct joint data audits to make sure that, outside the regulated period, parties abide by spending limits on the data they accrue; but also that, during elections, they can conduct raids, essentially like the drug tests you see in the Olympics, to make sure that political parties abide by the spending limits as they stand.
Ailidh Callander: When we talk about elections, we talk about the electoral cycle. There is recognition that the electoral cycle does not stop and start within that narrow period. It is an entire cycle. We need to change regulation to reflect that. Digital campaigning and the changes we have seen enable almost a permanent campaign. Looking beyond that restricted period is essential. On the question of offsetting costs, for example, part of this is already a requirement of law. It needs to be done for transparency and other requirements of data protection law. We need to think about the importance of this to us as a society. What are the costs? Are they minimal in comparison to what we are really talking about here? Thinking about the electoral cycle as a whole is the key point.
Jon Lloyd: We are not convinced that expanding the regulation of political campaigns beyond the current regulated period will be efficient at tackling some of these issues around disinformation in particular. Rather than extending the regulation period of political campaigns to tackle issues such as disinformation, instead we recommend focusing on the structural issues that are causing it. That means a greater focus on bulk ad disclosure and limits on microtargeting, rather than ad moratoriums.
While disinformation itself is a broad and complicated issue, there is a range of short-term and long-term measures that could be implemented to address both the symptoms and the root problem of disinformation. This could include platforms appropriately tweaking recommendation engines to reduce the virality of false and misleading content. It could be closing fake accounts and bots that are creating false amplification, and transparency about the political or possibly all ads on these platforms. Our understanding is that the main concerns about harming small campaigning groups and other bodies relate to calls for a total ban on online political ads or other requirements that increase costs but, as of today, we do not support such an outright ban on online political ads.
Lord German: I want to come back to you two, who support the idea. There have been severe complaints from charities and third-sector bodies that want to campaign on issues. If you have a permanent role, in the way you are describing, all those charities will have to look at everything they do online. Crisis, for example in talking about homelessness, has a big campaign at the moment. That campaign would count as being political forever, so they would have to identify all the costs derived; the Electoral Commission would then have to audit them, as well as everybody else. Is that a price worth paying? Jon has said perhaps not; you are saying it is a price worth paying, but they would have to cover all those costs, as well.
Pascal Crowe: As with all these things, the devil is in the detail and how it would be implemented. We also have to consider the current status quo. From our perspective, the main political parties, by which I mean the Liberal Democrats, the Conservative Party and the Labour Party, have by far and away the most developed datasets or assets accrued outside a regulated period. This means that, going into a general election, the main political parties have a massive advantage over the smaller political parties and can use the assets they have accrued outside the regulated period to save costs and campaign more efficiently within it. If there is no proper capture of the data assets that are being accrued outside the regulated period, we risk a narrowing of political campaigning, reducing it to an even more elite activity than it already is.
Lord German: You have not answered the point about Crisis or any other campaigning organisation.
Pascal Crowe: In terms of the burden on charities, that is a reasonable point to make. Perhaps you could make it as cheap and easy as possible for them. A cheaper, simpler Companies House-style organisation is a possibility.
Ailidh Callander: Going back to Pascal’s point, it is in the detail. What do we mean when we talk about expanding the regulation of political campaigns beyond the period? How do we define that? We are concerned that political campaigns do not start and end in the period immediately before the election. There needs to be wider thought about how we expand that, but that does not necessarily mean it applies to all actors, in the same way. Our concern is that a number of activities, whether ads that have run or others, are not captured and reported on, because they have fallen out with the period.
Q130 Lord Harris of Haringey: I am interested in this. Traditional, pre-digital analyses of electoral behaviour suggested that, in fact, most people made their minds up 18 months or so before an election itself, and the campaigns themselves made very little difference, so this question of the arbitrary cut-off when things start is probably not entirely helpful. The position of charities is different, in that they are regulated by the Charity Commission as to what they can and cannot do. Plenty of other actors have an interest and are not regulated at all. I am just about old enough to remember the campaigns of Mr Cube — anybody else? — against the nationalisation of the sugar industry. I did not even know that was a thing. It was not in the Labour manifesto last time. The point is that that campaign was run over many years to influence public opinion. I am interested in whether there are ways, or if it is even legitimate, to try to regulate and control things that are not necessarily run by charities and subjected to those constraints or run by political parties, but are designed to deliver a political objective. Maybe that is what democracy is all about.
Pascal Crowe: To follow up on the previous point, if I recall correctly, you have to meet the public test and the purpose test, under the Electoral Commission guidelines, to be included in a regulated period. I have been to training sessions by the Electoral Commission, in the run-up to an election, about what we can and cannot say. There is a degree of leeway about whether you can say you think something is good or bad and to what extent it is encouraging individuals to vote a certain way. There is a degree of discretion there. Charities are already very careful about what they do and do not say, and whether it can be construed as encouraging people to vote a certain way.
Ailidh Callander: To take that back to the transparency point, it is why there is a wider argument that, when talking about ad libraries, we need to ensure we include more than just political ads as defined by the companies. We need wider definition and wider transparency to capture political advertising that might not be run by a specific political party, taking a wider view. It goes back to what the problem we are trying to solve is.
Jon Lloyd: Currently the focus of the discussion is often on political parties or advertisers. I would emphasise even more that platforms, such as Facebook and Google, are equally if not more responsible for the process of delivering these ads. They often have way more information on people than political parties and especially campaigning groups, for example Crisis, will ever have. That makes it possible for them to optimise ad delivery and target groups, using lookalike audiences. That means political parties or campaigning groups do not need to collect any data on people; Facebook just does the job of choosing the right ones for them.
As Ailidh just said, this stuff is not happening in a silo. Why is it not okay for us to be targeted with political ads, when it is okay to target me with an ad for insurance or a job? The ad tech ecosystem needs reforming on the whole, which includes a strict and thorough implementation of GDPR or other data protection laws. We need that better transparency and accountability, especially towards things such as recommendation engines and what the platforms are choosing to show us. That is just to better understand and fight the spread of disinformation.
Q131 Lord Holmes of Richmond: Building on the point just made by Lord Harris, I am tempted to ask you how you think Mr Cube fits into the political sphere, but I will not. Should there be limits on the types of information that can be used for the microtargeting of political adverts or advertising across the piece, not least given what was recently seen from Google and Facebook?
Jon Lloyd: As a point of clarification. Google has placed restrictions on the way ads can be targeted, but more could be done. Facebook has only announced new controls that allow users to turn off custom audiences from a list. This might seem like a small positive step, but in practice it puts the onus on consumers or users to protect themselves. It leaves people vulnerable to microtargeting with custom audiences.
Mozilla believes that there should be reasonable limits on microtargeting and we are working within our organisation to determine what they are. Microtargeting is too broad a term, without an agreed-upon universal definition. Any limits on microtargeting need to be established with care. Some forms of targeting, based on where you are at the moment or current context—for example, if you are searching for shoes and are shown ads for shoes—should probably be acceptable, but targeting based on behavioural or interest profiles, collecting vast amounts of information about your interests and inferring other information from that, is driving most of the problematic behaviour in the ad space, right now. The Government should explore limits on those forms of online targeting.
Ailidh Callander: To add to what Jon said, we need to look at the limits set by data protection law. It is imperative that the law is enforced for the outstanding complaints against data brokers, the wider ad tech ecosystem and a number of the platforms. It is essential that the law is enforced there, which puts some limits, but we agree there should be restrictions. Part of the problem in defining, first, what we mean by microtargeting, and then what the red lines are, has been this absolute dearth of transparency and the unwillingness of those involved to play ball, whether the platforms, political parties or other companies within this ecosystem. We cannot rely on their business decisions as to the criteria they choose today or tomorrow. We need to define those limits. That is why transparency is so important in ensuring we have the information available to draw those lines and take this forward positively.
Pascal Crowe: I definitely second Ailidh and Jon in saying that data protection law is the first imperative we should follow in this. In an age where the system of value for campaigning is data and datasets, we need to find a way of incorporating the value that data bring to a political campaign more fully into the spending regulation framework we have. Go back to the fifth report of the Committee on Standards in Public Life. That was set up essentially to limit the war chests of campaigns, based on how much they could afford to spend on newspapers or billboards. Now, that reach is not being determined so much by money spent on newspapers, but how large or good your dataset is, and the value that is bringing to your campaign. That is often not being captured by the spending limits.
After determining whether the data that campaigns are using is legal and obtained legally, secondly, there should be an assessment of whether they have declared the value of the dataset and how much they paid for it. If the data have been obtained freely and legally, for example through canvassing, there should be a way of appraising the value of that dataset and the value it brings to a campaign financially to see where it fits within the spending limits. We have proposed that a basket-of-goods methodology could be used to apportion value to datasets, in that way.
Q132 The Chair: Do you think the present regulations relating to spending limits are sensible? Are they enforceable? How far are we from having our own version of Citizens United, whereby corporates are somehow able to get free of spending limits?
Pascal Crowe: In answer to the first point, a key concern is that, in an age of data-driven campaigning, it is very difficult to know, first, what value data really brings to a campaign, which is why we propose this basket-of-goods methodology, and secondly that campaigns are being honest and truthful about what they are using and how. That is why we have recommended that the ICO, in collaboration with the Electoral Commission, ought to be able to carry out data audits, so it knows how much data a campaign is going into an electoral cycle with. It should also be able to carry out raids much more easily and quickly than it currently can so, if there is a hint of something illegal happening, it does not happen after the fact of an election.
An election is quite different to a commercial transaction. If you are mis-sold a pair of shoes, you can go back to the shop and return them. Good luck to any of us trying to return the result of an election. That is generally deeply problematic. In terms of how far we are from a Citizens United-style environment, I would not go that far, but it is apparent that there is a loosening at the edges of what is considered acceptable and how far campaigners and parties are willing to push the envelope. The reasonably limited, although not non-existent, actions of the regulators are not doing much to deter them, until this point. Of course, the regulators are constrained by statute.
The Chair: Do you agree, as Lord Hennessy has put it, that we are moving quickly away from the “good chaps” concept of running fair elections?
Pascal Crowe: It seems that way.
Q133 Baroness Kidron: I am thinking about your various answers to the last two or three questions, and Jon’s excellent list of what is required from the platforms. Is there a case for not only regulation of the levels of transparency among platforms, in all the things you have brought up, but a code of conduct that the people placing ads—political parties, charities or those campaigning—commit to, so you can see that they have made a commitment that is not about money and regulation? Maybe it is underpinned by regulation, but the regulatory piece goes in the engine. What everybody has said all morning is that we need transparency, data transparency, audit and all these things. Is that a fair way to think about it, rather than ad spend or putting a lot of burden on the people advertising?
Jon Lloyd: A report came out this morning from the Centre for Data Ethics and Innovation, which recommended that a new regulator should require the platforms themselves to provide the ability for anybody to look under the hood, so to speak. That is likely to be through a meaningful ad archive. That would address targeting practices.
Baroness Kidron: You are not as worried about the front end, the campaign, the political party and so on. I wonder whether there is another way of getting consensus from that group of people, if we have sorted the back end.
Jon Lloyd: I do not have an opinion on that, right now.
Ailidh Callander: We are extremely concerned about the back end, because it determines the front end. I would labour the point that we need to look at data protection law and making it a reality. Obligations already exist in law for transparency, so what can be done? Under data protection law, there are issues in terms of the political sphere. There is a condition that allows political parties to process special category personal data without consent. That particular condition is inherently problematic. The Government decided not to implement the derogation in GDPR that would have allowed civil society to bring actions. Often we see across the board, in both the political sphere and the wider sphere of the digital ecosystem, that it is extremely difficult for an individual to take any action. That is another change.
As I mentioned before, it is interesting to look at the code of practice by the ICO in terms of the concerns around the political arena. That code of practice should, and I hope will, apply to a wide spectrum of actors, from platforms to political parties and all those in between. By making that a statutory code of practice, it would be much easier to enforce. Really, we have an enforcement gap. Looking at what the Electoral Commission and the ICO have been calling for, and have been for years, and making that a reality is the first step.
Pascal Crowe: I second that. Probably the most difficult part of this is asking political parties to legislate against their own self-interest. Certainly a good start would be putting the ICO’s framework code of practice on to a statutory footing to give colour to the kinds of activities that Open Rights Group has seen the parties carrying out. We have recently run a campaign on subject access requests, which is a right under data protection law that allows you to see the data an organisation is holding on you. Essentially, you send a letter to the data protection officer of a political party.
We have some interesting results. There are a lot of propensity scores, about how likely they believe you are to think a certain way. To give an example, we have seen the Labour Party attempting to guess what ethnicity you might be. We have seen the Conservative Party trying to guess what newspapers you are likely to read, and the Liberal Democrats trying to guess whether you are a pragmatic liberal — I am not really sure what that means — or a soft Tory.
It is also interesting to note that, quite often, the profiles the parties are building on us are incorrect. Quite a lot of the people who sent subject access requests had been mis-gendered, for example. That might affect the messages they get and the engagement they receive. Really, the question is not whether the parties are reading your minds or profiling correctly. The first point is whether they are following the law. Secondly, what effect is this having on the narrowing of our political debate and ecosystem?
Q134 Lord Knight of Weymouth: I have some concern about third parties, individuals and other unregulated organisations using political advertising. I am interested in the restrictions you think there should be on who can use political advertising on online social media and platforms. Should all advertisers defined as political have to be registered with the Electoral Commission?
Ailidh Callander: That is an issue where it is important to look at the different definitions and ways the platforms have ended up defining political ads, issue ads and political issue ads. We need to be much more comprehensive and coherent there to ensure that we capture what we want. It goes back to the point raised about the need to capture all the ads, but then to look at what other safeguards there are for political ads.
On registering with the Electoral Commission, again the question is what we want to achieve from that. Is it transparency or accountability? Answering that helps decide whether that would make the changes that we want. What does that registration effectively do? How can we ensure that there is accountability, at the end of the day?
Jon Lloyd: This question is not something we have looked at in detail, at the foundation. At a high level, we certainly believe there is a need to look at the current legislation on political campaigns and update it to account for the changes and possibilities brought about by new tech. On your second point, if offline political advertisers have to be registered with the Electoral Commission, online political advertisers should probably have to be registered as well. The definition of political is being left to the platforms themselves right now, in defining what a political ad or a political issue ad is. We are encouraging the Government to consider including all ads in any transparency disclosure measures.
Pascal Crowe: On a more general point, we contributed to the report from the APPG on Electoral Campaigning Transparency that recently came out. That has some good specific recommendations on this point. More generally, under the Political Parties, Elections and Referendums Act, you can withdraw yourself from being communicated to by political parties after one initial free post, but it is very difficult to do that online. In terms of what restrictions there should be on who can use political advertising on online social media, there is also a question about whether you should be able to cut yourself out of that environment if you want. You can take yourself off the electoral register or anonymise your presence on it but, once you are in the online sphere, it is very difficult to remove yourself from the conversation, if you want to.
Lord Knight of Weymouth: As a supplementary, is there a danger that we overregulate? We have the challenge of trying to anticipate where technology might get to by the next general election, in four years. We could overregulate and then the political machine takes it in the direction of travel that we see already, with a lot of conversation moving into private WhatsApp-type platforms. Then we have no transparency at all and no regulation of that.
Ailidh Callander: The first step, as I have said already, is enforcing the existing regulation, and then making slight changes to those existing frameworks to update them. They will stand strong for years to come. On the question of private messaging apps, one of the issues is that data is still involved. Often that is someone’s phone number, so where did that phone number come from? This has been studied extensively in Brazil, and that was a heavily underregulated country that had only just brought in its data protection framework. The scraping and wide abuse of phone numbers was used there, so this comes back to the abuse of data. The proposals to update the e-privacy framework, which sits alongside data protection law, have stalled at European level, but we must also make sure we update our law in that regard and all the law that regulates direct marketing. That will help with that concern, which goes back to the phone number example.
Jon Lloyd: We do not have a particular opinion about the overregulation of the market, at the moment.
Pascal Crowe: I echo Jon on that.
Q135 The Chair: I will ask a last relatively simple question. If there was any one thing that we could recommend or could be done to improve the regulation of political campaigns in a digital age, what would it be? What is your one wish?
Pascal Crowe: Change the system of value for electoral regulation to one that focuses on data, rather than purely on cost, or incorporates data into the cost-based system.
Ailidh Callander: The struggle is to find just one thing to make the change that we need. If I were to have one thing, it would be to make the changes to strengthen the data protection law that we have and make it easier to enforce. That is the derogation for political parties’ collective redress and making a statutory code of practice.
The Chair: You mean that we could possibly enforce the laws that are already there.
Ailidh Callander: Yes, that is the number one.
The Chair: That is not a bad starting point.
Jon Lloyd: The Government should require platforms to disclose the ads on their platforms in publicly available ad repositories, where they can be analysed by regulators and watchdog groups. They should include the targeting parameters for the ads and aggregate demographic information about who sees and clicks on those ads. The Government should seriously require all ads to be disclosed, not just political ads.
The Chair: Thank you very much indeed. You have been very helpful.