Written evidence submitted by the Computer and Communications Industry Association (SMH0029)

CCIA Response to Call for Evidence - Social media, misinformation and harmful algorithms

The Computer & Communications Industry Association (CCIA) is an international, not-for-profit trade association representing a broad cross section of communications and technology firms. For more than 50 years, CCIA has promoted open markets, open systems, and open networks. Its members include multiple companies offering social media services, search engines and related digital services in the UK.

To what extent do the business models of social media companies, search engines and others encourage the spread of harmful content, and contribute to wider social harms?

There is a common misconception that social media companies, search engines and similar companies simply monetise “attention”. This is misleading. CCIA Research Center analysis of two independently-administered survey experiments finds not all attention is equally valuable, and some types of negative content (like hate speech) can both harm a social media site’s brand and significantly harm the ability of nearby ads to drive revenue/clicks.

The business models of these companies are better understood in the academic literature as being “multi-sided markets”. This means that their business is based on connecting different parties, which might include: merchants selling products that users search for; users talking to each other on social media; and advertisers seeking to reach people that might be interested in purchasing their goods and services. The operator of the service therefore has to work to satisfy those different user groups and create a platform each group chooses to use, which includes:

●        Users who want to express themselves, including expressing opinions that might (intentionally or not) represent misinformation, but also want to use platforms that they find trustworthy, safe and generally attractive. It is generally understood, for example, that one of the reasons why many users switched from Myspace to Facebook is that it offered a better-moderated experience.

●        Advertisers who want to reach users, but want their adverts placed in a context that supports their brand values.

●        Creators who want to share content that attracts users and monetise this by sharing content from advertisers.

At the same time, companies need to respond to other priorities and it would be an oversimplification to extrapolate moderation policies simply from commercial pressures.

●        Regulatory constraints: either requirements for systems and processes to moderate some content, or restrictions on the moderation of other content (both of which are included in the Online Safety Act).

●        Wider social interest: responsible social media companies and their staff are committed to exercising their role properly and this will limit their willingness to take down some content (reflecting its educational or wider social value and/or a commitment to free expression) or host other content (an aversion to potential social harms, including disturbances like the riots this year).

How do social media companies and search engines use algorithms to rank content, how does this reflect their business models, and how does it play into the spread of misinformation, disinformation and harmful content?

Any service that presents content inevitably uses some kind of process to decide how it is ranked and any digital service is likely to do so using some kind of “algorithm”. That algorithm might be very simple, with content presented chronologically in the order in which it is posted, but as services become larger and more open that will become harder for users. They will either need to work through more content to find what they want, or filter content themselves (e.g. by blocking other users they find objectionable). In its amicus brief last year to the U.S. Supreme Court in Gonzalez v. Google, CCIA noted that earlier, simpler methods of content organisation “were easy to implement but came with scaling problems and internal biases” as the amount of content available online increased exponentially but “users’ free time did not”.

Most companies presenting content online therefore use algorithms to share content users are more likely to find useful, valuable, interesting, or otherwise otherwise important to the user. As noted in the amicus brief mentioned above, this will include:

●        inventory creation: bringing large amounts of content together;

●        filtering: removing content that is irrelevant to a particular user or does not meet content moderation policies through a mixture of automatic and human-review processes; and then

●        ranking content and then reranking it to address issues including homogeneity (too much similar content listed in a row) and “borderline” dangerous content.

Diversity in digital services and their business models will reflect underlying diversity in user preferences for how content is prioritised. This can be seen in some social media platforms prioritising political content, for example, while others do not. This prioritisation of content is  also an important area for competition with platforms innovating in the nature of the algorithm, how users engage with that algorithm (often giving them new opportunities to express preferences over the content they receive) and its ability to surface new and interesting content. Users’ experience can also vary based on their choices with services increasingly providing customisation tools that allow people to reset or otherwise change their algorithmic recommendations.

These algorithms are also relevant to mitigating the spread of misinformation, disinformation and harmful content. Helping large numbers of people using the Internet to navigate large amounts of content available online is inevitably difficult and so there will always be some degree of risk that harmful content is shared, which responsible platforms work to manage. Malicious actors will actively seek to work around attempts to limit the spread of harmful content so this will inevitably be an ongoing challenge. This is why the regulatory framework focuses on ensuring that companies do what they can to implement suitable systems and processes (which might change over time as, for example, new tools for proactive monitoring become available).

The goal for platforms will be to balance the interests described in the last section and in doing so evidence regulatory compliance with Ofcom. Reflecting the competitiveness and pluralism of the sector, platforms differ in their approach to moderation, but responsible platforms generally seek to do so transparently (subject to the limits of commercial competition and not undermining trust and safety operations) and with a technical sophistication that would not have been possible before the Internet era.

What role did social media algorithms play in the riots that took place in the UK in summer 2024?

There has been extensive research over decades on the determinants of whether rioting occurs and its severity. Any attempt to understand the cause of the riots will necessarily be multivariate and is beyond the scope of this consultation and CCIA’s expertise. Attempting to isolate one factor such as social media and understand that alone is likely to be misleading, however. While many of those rioting may have used social media to communicate with one another (to be expected given that most people use social media in some form) this does not mean that there was a causal relationship between social media use in general, or any particular feature, and rioting.

A balanced understanding of the issue should reflect upon the challenges and opportunities presented by social media versus earlier forms of communication. Challenges might include:

●        Efficiency - social media might be more efficient than other potential forms of communication, which will enable the more efficient transmission of harmful content as well as beneficial content.

●        Openness - social media might be more open than other potential forms of communication, supporting people who would otherwise find it harder to engage with the wider world. However this might include some instances where people bring views that are marginalized for understandable reasons.

On the other hand, opportunities might include:

●        Transparency - in contrast with word of mouth and other means by which misinformation has previously spread (e.g. the Blackberry private messaging used in the 2011 riots), which has led to rioting in the past, it is easier for responsible actors to identify and engage with content shared on social media.

●        Moderation - social media platforms have trust and safety operations and now an extensive regulatory framework. Again this did not exist for other forms of communication used in earlier eras in which riots have taken place. Many platforms remove content before it is reported by users. For example Meta reports that more than 95% of content removed July to September 2024 as hate speech was removed before people reported it.

Considering these and other challenges and opportunities could over time yield a constructive debate around how different actors can work together to address how people might respond differently to social grievances with changes with the communications environment.

How effective is the UK's regulatory and legislative framework on tackling these issues?

It is too early to be able to give a firm view of the effectiveness of the UK’s regulatory and legislative framework for tackling these issues in that a major change is just being implemented with the Online Safety Act coming into force. This includes misinformation and disinformation where Ofcom is establishing an advisory committee to address the issue specifically. There are two important features to note about that regime:

●        It has been a major undertaking. Ofcom as the regulator and regulated companies have gone through an extensive process to implement the law. This means that there should be a premium on addressing concerns through evolutionary and collaborative work within that established regulatory framework, versus creating uncertainty and disruption in the already onerous compliance process.

●        It is intended to be flexible. This means that as evidence is developed, Ofcom should be able to responsibly take new risks into account and work with regulated companies to address them. It avoids mistakes seen elsewhere with specific timed notice and takedown regimes focused on specific harms, which often lead to over removal and disrupt overall progress in improving trust and safety processes. Again this is a reason to avoid unnecessary and/or disruptive changes to that regulatory framework.

Which bodies should be held accountable for the spread of misinformation, disinformation and harmful content as a result of social media and search engines’ use of algorithms and AI?

Different parties have different roles to play and therefore different forms of accountability.

●        If users share fraudulent, dangerous or otherwise illegal content, they should be held responsible for content they post, not third parties.

●        Digital services are responsible for developing, updating, and implementing suitable, proportionate systems and processes to mitigate the spread of dangerous content and uphold their terms of use.

●        Regulators are responsible for oversight of those systems and processes and processes and can promote best practices developed by practitioners, particularly with smaller platforms that might struggle to control such content.

18 December 2024