Contemporary electoral democracy emerged and has been concentrated mainly in the global North. That may be changing. There have been significant recent shifts towards more authoritarian politics and governance within Northern democracies. There are a range of apparent causes, and no clear indications of the likely long-term trend. However, the adverse impacts on the politics of digitalisation, especially the explosion of social media use, justify some pessimism. Even at its best, electoral democracy has always been a very imperfect mechanism for ensuring government by the people. Large proportions of voters routinely vote expressively – i.e. according to their identity, beliefs, or values – rather than instrumentally, i.e. to maximise their chances of influencing the outcome. Digitalisation increases pressures to vote expressively and reduces the scope to vote instrumentally. There are signs of contemporary convergence between formally democratic and evidently authoritarian regimes. In both, governments rely increasingly on digital media to confuse their voters and citizens.
Democracy, voting, social media, digitalisation, political communication, informational autocracy, expressive voting, instrumental voting, authoritarianism.
The 1995 IDS Bulletin‘Towards Democratic Governance’ (Robinson 1995) was produced in a heady period in the history of governance and development. The Soviet Bloc had disintegrated a few years before. The third wave of democratisation had just rolled through Eastern and Central Europe, Latin America, and much of sub-Saharan Africa. There was an air of triumphalism, and not just in the ‘West’. Words and phrases like accountability, transparency, and good government had suddenly become commonplace in the global South. Large numbers of people were savouring liberation from repressive, military, and authoritarian regimes and relishing the possibility that they might have a say in the ways in which they were governed.
These Southern voices were not represented in the 1995 IDS Bulletin. Its authors were predominantly white male researchers from the global North who were engaged, supportively or critically, with the emerging ‘good governance’ and democratisation programmes of the bilateral and multilateral aid donors. Relative to the size of the aid-recipient economies of the global South, aid from these Western sources peaked around this time. Donor agencies had money to invest in their new ‘good governance’ agendas. No longer inhibited by competition from the Soviet Union for the United Nations (UN) votes of the countries of the global South, Western and Western-influenced aid donors now had the confidence to lay down principles about how those countries should govern themselves if they were to continue to receive aid.
For many of those on the receiving end, this felt like a systematic violation of sovereignty. And did the West really have the competence and experience to engage effectively in the large-scale political and institutional engineering that was implied? A few months previously, Mark Robinson, editor of the 1995 IDS Bulletin, had sat down with me and a blank sheet of paper on which we had begun to sketch out the first detailed list of the dimensions and indicators of good government to be used by the United Kingdom (UK) aid agency (then the Department for International Development, DFID). At that moment, this seemed an ambitious and exciting initiative. The hubris became clear only later.
The contributors to the 1995 IDS Bulletin were broadly representative of their (Northern, aid-oriented, policy-wonkish) tribe in that (a) they had enthusiastically welcomed the third wave of democratisation; (b) they saw their mission as helping to strengthen and deepen it; (c) they believed or hoped that Western bilateral and multilateral aid agencies would be up to that task; and (d) they assumed that the countries of the global North would remain firmly democratic and would be the lodestar for poorer societies that were democratising or trying to do so. Politically, the global South would want to be more like us, and we would help them become so.
How times change. The overconfidence and the impracticality of this aid-driven good governance agenda now merit little more than historical footnotes. It matters more that the Third Wave of democracy has turned to an ebb tide: authoritarianism seems to be on the rise again in the global South. Perhaps even more disturbing are the growing signs of the degeneration of democracy in the global North – and not just in the United States (US), where democracy is visibly endangered. The global North is still the world’s main reservoir of democracy, and has been a major driver of its expansion elsewhere. If democracy is in trouble there, it is likely to be in trouble everywhere.
The signs of degeneration in ‘Northern’ democracies are diffuse, varied, and overlapping. They include the following:
Taken individually, some of these dimensions of democratic degeneration might not seem particularly surprising or worrying. The history of elections and democracy is peppered with episodes of violence, paranoia, polarisation, majoritarianism, and repeated corruptions of the rules of the game. Large-scale immigration has almost always been unpopular. Levels of transborder population movements have been particularly high in recent decades across much of the globe. The human and economic toll of the Covid-19 pandemic has been severe in many places, as have the effects in poorer communities of recent waves of addiction to new manufactured drugs, especially opioids and fentanyl in the US.
It should be no surprise that many social democratic parties have lost much of their former working-class electoral base to more nativist or populist parties. There has long been truth to the claim that the more educated and articulate professionals who have come to dominate social democratic parties have paid limited attention to their ‘traditional’ working-class base. In turn, they have also often failed to appreciate the alienation and resentment caused by attitudes, rhetoric, and policies that seem to prioritise the needs of foreigners, immigrants, ethnic or cultural minorities – and (more educated) women (Gethin et al. 2021).
It would not be unreasonable to retain some optimism about the future of democracy in the global North, on the grounds that the threats listed above simply reflect the coincidence of a series of short-term factors. The historically anomalous lengthy period without endemic threats of inter-state war in the global North may also be a significant factor. It has been argued that this permits frivolity and irresponsibility in electoral politics because voters believe they are not risking anything fundamental by voting for politicians who are unlikely to form competent governments.
But it is now more than a decade since Russia occupied a chunk of Ukraine, and three years since its large-scale invasion. Political frivolity seems not to have diminished in that time, except perhaps in Poland and the Baltic states, which are right on the front line and routinely targeted by cyberattacks, sabotage, and other short-of-war assaults. It is difficult to be optimistic about the prospects for electoral democracy. It may be experiencing serious, long-term deterioration. In particular, the vote may be losing value because it is becoming increasingly difficult for the average citizen to use it in an informed way to choose representatives and governments. Before developing that argument, and to protect ourselves against nostalgia for a golden era of electoral democracy that never actually existed, we need to take a realistic view of how in practice the vote is used.
In the contemporary world, electoral democracy is one of the most widely accepted creeds. Its devotees account for a much higher proportion of the population today than a century ago. The vote is its primary icon. It is often represented as the means to achieve ‘the government of the people, by the people, for the people’. The strong version of the creed is that peoples will govern themselves if every adult is given the right to vote and allowed to exercise it. That is impossibly optimistic. The mismatches between aspiration and reality have been of major concern since (a somewhat imperfect) nationwide electoral democracy was established in the US at the end of the eighteenth century. Leaving aside the many illegal and illegitimate practices that encrust so many electoral democracies – gerrymandering; vote buying; pressure on voters from family members, co-ethnics, or religious figures; voter suppression through registration procedures; booth-capturing; counting fraud, and so on – there are four main reasons why, even when practised in full accordance with the law, electoral democracy with universal suffrage will never fully equate to government by the people:4
Expressive voting – and expressive popular political activity more generally – have almost certainly become more frequent since the emergence of social media around 2005. For Britain, we have strong evidence of increases in the frequency with which people vote against a political party rather than for one (booing rather than cheering) (Rivas and Rockey 2021). Much of the previous theoretical, conceptual, and academic literature on voting was produced by serious-minded scholars who tended to assume that voters cast their ballot instrumentally, to maximise the chances that they could influence public policy in their own interests.5
By contrast, it has become clear more recently that raw emotions, especially negative emotions such as resentment and hostility against other social groups, are becoming more common drivers of voting behaviour. There is more on this in Sections 5 to 7. Note for the moment that it can broadly be explained by the spread of popular online political engagement, especially social media. The psychology of how humans process information and make decisions can in turn explain the causality involved.
One of the most general findings from extensive psychology research is that, ultimately to economise on mental processing energy and time, humans typically do not engage in the extensive appraisal and evaluation of new data that comes their way. Instead, if they pay any attention at all to that data, they (a) use a whole series of heuristics to make it consistent with what they already believe; and (b) make decisions based more on emotions than facts. People tend to welcome and absorb new data that can be fitted into their existing understanding of the world, and to ignore or reject data that does not fit (confirmation bias). Normal human reactions to new data are very different from the scientific norms: ‘Falsifiability may provide the cornerstone of the scientific method… But believability constitutes the hallmark of a good narrative. When a fact is plausible, scientists still need to test it; that is the purpose of hypothesis generation and testing. However, when a story is plausible, many people will believe it is true’ (McDermott 2019: 221).
Evolution has left humans, individually and collectively, with a major systemic cognitive vulnerability: they are consistently primed to believe information that they find convenient, comfortable, and consistent with their values rather than to make the effort to subject it to serious scrutiny.6 Some humans have learned how to take advantage of that gullibility in their interactions with others, including in the political and electoral arenas. But digital technology, especially social media, now provides would-be manipulators with the tools to exploit these vulnerabilities on a massive scale, by flooding individuals with targeted, emotive, and endlessly replicated ‘information’. This threatens to further increase the gap between the ideal of ‘government by the people’ and the nearest approximation that can be achieved through electoral democracy. Because there has always been a gap, I feel uncomfortable with more ambitious definitions of electoral democracy that locate its essence in popular control over governments. I settle instead for Charles Tilly’s more realist process definition: ‘A regime is democratic insofar as it maintains broad citizenship, equal citizenship, binding consultation of citizens at large with respect to governmental activities and personnel, as well as protection of citizens from arbitrary action by governmental agents’ (Tilly 2000: 4).
He summarises democracy as ‘protected consultation’, i.e. voters have institutionalised influence over the people that rule them. My argument is not that we are facing a sudden precipitous fall in the capacity of voters in electoral democracies to choose governments and influence public policies. That capacity has always been limited. The risk now is that it is being reduced further. The risk arises from changes in the dominant modes of political communication.
What is political communication, and how do we distinguish it from other kinds of communication? There is no simple answer. If we adopt the expansive definition of politics as something that permeates most human interactions, then ‘everything is political’, and it follows that all communication is (actually or potentially) political. For present purposes, I define political communication as communication that might affect voting choices. We can usefully distinguish between two broad channels of political communication:
It is the latter that have changed most over human history, as summarised in Table 1. The world has moved steadily towards modes of political communication that are increasingly large in terms of scale and reach; rapid; diverse in content; and pluralistic – and often opaque – in terms of source. The focus in this article is on the consequences of the shift in recent decades from Stage C, where TV, radio, newspapers, and film – and earlier also the post and the telegraph – were the main channels of political communication, to Stage D (internet, social media, artificial intelligence (AI), and deep fakes).
| Communication technology stage | Direction of communication | Rate of spread of ideas and information (I&I) | Diversity of sources of I&I | Reach of I&I messages | Origin date(s) | Political significance |
|---|---|---|---|---|---|---|
| A. Talk, ritual, imagery, symbolism | Top-down | Slow | Narrow | Narrow | ||
| B. Printing press | Top-down | Slow | Wider | Wider | Late 1400s | Enables Protestant upsurge in Europe |
| C. Telegraph, post, newspapers, radio, film, TV | Mainly top-down | Much faster | Much wider | Much wider | 1850s–1950s | Mass national politics, membership parties, and trades unions |
| D. Internet, social media | Multidirectional | Near-immediate | Very wide | Very wide | 1990s | (See main text) |
Source: Author’s own.
There is a massive literature on the impact of digital technology on politics and governance generally. The literature grows fast partly because of the constant evolution of the technology itself and the ways in which people use it. I do not deal here with the impact of digital technology on politics and governance generally. My focus is on the impact on voting, and specifically on the reasons to be pessimistic about that impact. There are two stages to my argument. In this section, I make the case that, regardless of context, the spread of digital technology itself poses a significant threat to clear-eyed instrumental voting because it makes large-scale voter manipulation cheap and easy. In Section 6, I argue that these adverse consequences have been exacerbated by recent changes in patterns of work and social interaction.
At the most basic level, the emergence and evolution of digital technology poses a threat to intelligent, instrumental voting choices because it increases the scope for some (powerful) actors to manipulate the ways in which voters understand the world. That emerges from the interaction of the following:
Digital communications media greatly increase the scope for some actors to manipulate other people’s world views on a large scale and at low cost. But what are the manipulators’ incentives to exploit these possibilities by shaping content to maximise user engagement? They are both money and politics, in varying mixtures:
In sum, there are a range of ways in which a variety of commercial and political actors benefit from manipulating online digital content to generate more user engagement. In Section 6, I explore the question of what the actual or potential constraints on the extent of manipulation are. Before that, I present the case that the problems outlined in this section have been exacerbated by contemporary changes in patterns of work and social interaction, especially in richer countries.
The populations of the global North, in particular, have become more open and more vulnerable to (misleading) online content designed to engage them because their alternative sources of information and political communication have changed. One set of changes relate to the world of work and the other to social relations outside work.
In consequence of de-industrialisation, men in particular now spend less time in populous workplaces that are conducive to face-to-face interactions with diverse people who may not be from the same family or the same immediate residential neighbourhood. These kinds of social interactions, taking place especially in factories, mines, ports, railway and post office networks, militaries, and large administrative offices, provide a standing challenge to parochial world views. Further, many of these work environments nurtured ‘proletarian’ perspectives in which were often embedded notions of the unity of humanity and the commonality of the interests of people located in very different places but similarly subjected to capitalist employment relationships.9
Proletarian consciousness was of course not always strong or altruistic.10 But, reinforced by mass circulation newspapers, radio, and then TV, trades unions, and mass membership political parties, it provided a degree of insulation against the kind of aggressive political narratives familiar to us today that locate the perceived problems of one social category in the malevolence of members of another group, who, because of education, culture, or place of origin, sound, look, or behave differently.
Post-industrial workplaces, that are mostly focused on service activities, engage employees less. They spend less time at work, originally because of reductions in working hours, and now also because of the spread of working from home. And car ownership and general increases in personal mobility decrease the likelihood that people will socialise with co-workers. Social interaction with co-workers is less influential in forming world views to challenge narrow or parochial perspectives.11
That would be less of a problem if people were engaging in more social interaction outside of work, especially social interaction of a collective nature in clubs, societies, and associations – and thus outside the immediate family. But the opposite is the case. Especially in the US, but also in many other countries, engagement in associational life of this kind is in steep decline. The alarm about this was first sounded publicly by Robert Putnam in his 2000 book Bowling Alone (Putnam 2000). More recent research shows that the decline in associational activity that he identified has continued throughout this century and seems almost to be a global phenomenon (Putnam 2023).12
The decline of sociability outside the household is not balanced by increased sociability within the household. The long-term decline in birthrates and average household size anyway militates against that. There is increasing evidence – again, especially, but not only, for the US – of big increases in both (a) the time that people spend alone; and (b) the reported incidence of feelings of loneliness (Thompson 2025; Burn-Murdoch 2025). The two are not identical: people can be physically alone but socialising online. Unfortunately, at least for younger people, the amount of time spent online has increased considerably since the take-off of social media around 2010, and is statistically associated with unhappiness and poor mental health. At the very least, online socialisation is not compensating for the reduction in real face-to-face social interaction, and is probably exacerbating the incidence of alienation more generally.13 The link with politics and voting is direct: people who are lonely or alienated more broadly are more vulnerable to recruitment into online networks peddling toxic content: conspiracy theories, misinformation, and hatred (Ebner 2020).
Within the digital sphere, there is a degree of natural immunity to the proliferation of toxic content. Some professional and business users are only willing to use platforms that contain reliable information and/or have concerns about the reputational impact of being associated with toxic content. After content moderation was largely abandoned on X, the number of subscribers began to decline and alternatives, notably Bluesky, grew fast. Some platforms will continue to self-regulate to some degree. Unfortunately, there is little reason to be optimistic that this natural immunity and self-regulation will play more than a marginal role. The financial and political incentives to produce and promote toxic content seem overwhelmingly powerful.
We then turn to government regulation. In practice, this is largely a matter for the US government. It is one of only three governments that have the political power, organisational capacity, and economic clout to be significant players in this field. The European Commission is trying to regulate but is hobbled by (a) the fact that none of the big digital platforms or digital technology companies is European; and (b) the reality that economic power is shifting away from Europe to the US and to China. Rates of economic and productivity growth have been significantly higher in the US than in Europe for about the last two decades. Relative to Europe, the Chinese economy has been growing faster for longer. It is the only country other than the US to host large digital platforms. But, except for TikTok, the others, such as Baidu, are confined largely to China. China might begin to rival the US, especially if recent evidence of Chinese technological leadership in AI proves well founded.
Currently, the US dominates. It seems very unlikely that any US government will try seriously to regulate American digital platforms for toxic content:
Short of a major political upheaval in the US resulting from a popular revolt against the current political dominance of large companies, digital platforms, and private fortunes, there seems little prospect of serious government efforts to regulate digital content. Without that, online activities seem likely to continue to undermine the capacity of individual voters to use their vote in an informed and instrumental way – in the US, in other countries in the global North, and increasingly worldwide.
This article focuses on voting. It summarises worrying evidence that the combination of digital technologies and recent trends towards reduced levels of face-to-face social interaction poses threats to the ability of (and incentives for) voters to use their votes instrumentally to influence the choice of political representatives and governments. This is most evident in the US but also visible in a wide range of other countries. It is one dimension of the current global trend towards less democratic governance.17 But what kind of ‘less democratic’ governance? The world is not generally undergoing a reversion to familiar old authoritarianisms heavily dependent on the deployment or threat of force, i.e. Communist or military rule or personalistic and familial regimes like those of the Assads in Syria between 1971 and 2024. It has been convincingly demonstrated that we are moving into an era of spin dictators and informational autocracy, where ‘authoritarian’ rulers stay in power less through coercion and more by manipulating information, and keeping us confused and divided (Guriev and Treisman 2019). This raises the prospect that, just as the digital sphere blurs the boundaries between politics, entertainment, gambling, investing, advertising, news, and opinion, in the future we may find it much harder to distinguish democracies from autocracies. Both may appear to be driven by contested elections, but the quality of the contests may generally be low across the board.
Brechenmacher, S. (2024) Why Gender is Central to the Antidemocratic Playbook: Unpacking the Linkages in the United States and Beyond, Carnegie Endowment for International Peace, 25 November (accessed 19 March 2025)
Burn-Murdoch, J. (2025) ‘Young People are Hanging Out Less – It May be Harming Their Mental Health’, Financial Times, 18 January (accessed 19 March 2025)
Carothers, T. and Hartnett, B. (2024) ‘Misunderstanding Democratic Backsliding’, Journal of Democracy 35.3: 24–37 (accessed 19 March 2025)
Chen, B.J. (2025) ‘What Does Silicon Valley Want? Personal Observations on Tech and Policy’, Data and Society, 22 January (accessed 19 March 2025)
Downs, A. (1957) An Economic Theory of Democracy, New York NY: Harper & Row
Ebner, J. (2020) Going Dark: The Secret Social Life of Extremists, London: Bloomsbury
Financial Times (2025) ‘Musk Expresses Support for Rival to Reform UK as Feud in Farage’s Party Intensifies’, 10 March (accessed 19 March 2025)
Gethin, A.; Martínez-Toledano, C. and Piketty, T. (2021) Brahmin Left Versus Merchant Right: Changing Political Cleavages in 21 Western Democracies, 1948–2020, World Inequality Lab Working Paper 2021/15, Paris: World Inequality Lab (accessed 19 March 2025)
Goldthorpe, J.H.; Lockwood, D.; Bechhofer, F. and Platt, J. (1968) The Affluent Worker: Industrial Attitudes and Behaviour, Cambridge: Cambridge University Press
Guriev, S. and Treisman, D. (2019) ‘Informational Autocrats’, Journal of Economic Perspectives 33.4: 100–27
Hacker, J.S. and Pierson, P. (2020) Let them Eat Tweets: How the Right Rules in an Age of Extreme Inequality, New York NY: Liveright Publishing Corporation
Hayes, C. (2025) The Sirens’ Call: How Attention Became the World’s Most Endangered Resource, Brunswick: Scribe
Hertel-Fernandez, A. (2019) State Capture: How Conservative Activists, Big Businesses, and Wealthy Donors Reshaped the American States – and the Nation, New York NY: Oxford University Press
Khemani, S. (2019) What Is State Capacity?, Policy Research Working Paper 8734, Washington DC: World Bank (accessed 19 March 2025)
LSE (n.d.) Professor Sonia Livingstone, London School of Economics (accessed 19 March 2025)
Margolin, J. (2024) The Wagner Group: Inside Russia’s Mercenary Army, London: Reaktion Books
McDermott, R. (2019) ‘Psychological Underpinnings of Post-Truth in Political Beliefs’, PS: Political Science and Politics 52.2: 218–22, DOI: 10.1017/S104909651800207X (accessed 19 March 2025)
Mejias, U.A. and Couldry, N. (2024) Data Grab: The New Colonialism of Big Tech (And How to Fight Back), Chicago IL: University of Chicago Press
Putnam, R.D. (2023) Join or Die?, Netflix documentary (accessed 31 March 2025)
Putnam, R.D. (2000) Bowling Alone: The Collapse and Revival of American Community, New York NY: Simon and Schuster
Reuters Institute for the Study of Journalism (2024) Digital News Report 2024, Oxford: Reuters Institute for the Study of Journalism (accessed 19 March 2025)
Rivas, J. and Rockey, J. (2021) ‘Expressive Voting with Booing and Cheering: Evidence from Britain’, European Journal of Political Economy 67: 101956, DOI: 10.1016/j.ejpoleco.2020.101956 (accessed 19 March 2025)
Robinson, M. (ed.) (1995) ‘Towards Democratic Governance’, IDS Bulletin 26.2 (accessed 1 April 2025)
Salazar-Miranda, A. et al. (2024) Shifting Patterns of Social Interaction: Exploring the Social Life of Urban Spaces Through A.I., NBER Working Paper 33185, Cambridge MA: National Bureau of Economic Research
Thompson, D. (2025) ‘The Anti-Social Century’, The Atlantic, 8 January (accessed 19 March 2025)
Tilly, C. (2000) ‘Processes and Mechanisms of Democratization’, Sociological Theory 18.1: 1–16, DOI: 10.1111/0735-2751.00085 (accessed 19 March 2025)
Insert credits text.