Where once social media was regarded as a virtual town hall, providing people with a place to share views and connect with like-minded others, it is now often weaponised to fuel divisive narratives and polarise political debate. Across our research at IDSB, we have identified several ways social media platforms distort online conversations, reshaping how political debate unfolds online.<\/p>\n
The first mechanism is what researchers have termed the \"funhouse mirror factory\"<\/a> \u2013 a phenomenon where over 90% of political content online is produced by just 10% of users. The reason this matters is that it can create the impression that certain views are shared and supported by large numbers of people when, in reality, they may be driven by a small but highly active group. This has broader implications for shaping consensus and social norms across society, but more acutely, it influences political parties and the policies they choose to pursue.<\/p>\n As a case in point, we examined a network of accounts and conversations on X<\/a> discussing\u00a0mass deportations in the UK.\u00a0 Our analysis traced these discussions back to a handful of fringe far-right accounts in November 2024. Over time, we found that\u00a0mentions of mass deportations moved from fringe online communities to\u00a0right-wing political figures such as Ben Habib and Rupert Lowe before entering mainstream political debate through Reform UK politicians. While we would always caution against drawing firm conclusions about cause and effect from social media data alone, our analysis shows one way\u00a0party policies may be influenced through the purposeful, targeted posting of a small number of fringe accounts. Over time,<\/span> this can<\/span>\u00a0distort the perception that view<\/span>s are\u00a0<\/span>shared and supported by a broader audience within the electorate.\u00a0<\/span><\/p>\n Another way social media is reshaping politics is through the algorithmic amplification of content. Social media platforms' business models depend on keeping people engaged with content. As a result, platforms are motivated to publish content that attracts attention regardless of whether it informs people and contributes positively to public debate or drives divisive political discourse.<\/p>\n During the Southport riots in summer 2024, several accounts<\/a> pivoted in their style of content to capitalise on a rapidly unfolding, trending event and boost engagement. Much of this content was later shown to be misinformation that only added fuel to the fire of an already challenging and harmful incident.<\/p>\n Algorithms track and promote content receiving high engagement. In this way, the spread of content is driven by both the motivations of users and the structure and architecture of platforms. This has been especially problematic for the promotion of violent or extreme content, with a reluctance from social media platforms to remove such content due to the high level of engagement that it <\/span>receives<\/span>. <\/span>Influencers and politicians alike can take advantage of the monetization and digital advertising that is embedded in the algorithmic amplification of content to increase the reach of their content and, in some cases, make money. In other words, there is an in-built reward mechanism attached to social media platforms whereby shocking and often extreme content is rewarded both financially and through increased engagement.<\/p>\n There is also little transparency around how algorithms work or why they promote certain content. This matters because the algorithmic structure of platforms<\/a> can influence which political voices are heard most widely. Some politicians and narratives may therefore appear more popular not because more people support them, but because platform algorithms amplify them.<\/p>\n