Power: A Radical View began its surprisingly long life as a very short book (now the first chapter of this one) in 1974 and its third edition, incorporating a series of revisions, elaborations and applications, appeared in 2021—four years ago as I write this preface to its republication. What I aim to do here is to take stock of how I now see its arguments and claims, in the light of the various criticisms and reactions it has evoked and other related accounts of and debates about power over the years since it first appeared. For whom is it written and who should read it? I suggest that any reader new to the book’s arguments and to those debates should skip this preface and, if so inclined, go straight to the first chapter and maybe turn to it later—if for no other reason that that the journey it records might be more interesting, and would in any case make more sense that way.
Much has happened in the world in those four years since 2021; it is as though history has suddenly speeded up. I cannot resist using Antonio Gramsci often-quoted sentence, despite the vastly different context, to describe the present moment: ‘[t]he crisis consists precisely in the fact that the old is dying and the new cannot be born; in this interregnum a great variety of morbid symptoms appear.’ (Gramsci 1971:276) Two developments in particular appear to be transforming our world: the election of Donald Trump as President of the United States in an ever-more reactionary, authoritarian world, and the exponential growth of hyperconnectivity through social media, together with current developments supposedly hurtling towards the emergence of advanced generative AI. In what ways do these developments bear upon how we are to think about power?
I will here start with what we mean by ‘power’ at its most general, in abstraction from any particular context. Then I will turn to what mostly interests us when we speak of power in social and political life. The book proposes that power is best viewed three-dimensionally. The suggestion is that doing so enables one to see further and deeper (which is what I meant and mean by ‘radical’ when writing of a ‘radical view’). After all, if you can view something in three dimensions, why would you settle for only one or two? I will indicate what viewing social and political power three-dimensionally amounts to and conclude by suggesting ways in which exploring the third dimension—the theme of the fifth chapter—needs to be taken further in the light of the transformative changes taking place in our world, referred to above.
Power at its ‘thinnest’ (Abend 2019, Kirchin 2013)–most abstracted from any context and presuppositions–simply amounts to the capacity of some source to produce effects of some kind. We speak of horse power, electric power, brain power and market power. We are interested in it—in locating it, harnessing or restricting it and assessing its extent, even quantifying it when we can, when we see those effects as significant in one way or another. In social and political life, we are interested in power when it consists in the capacity to secure the compliance of another or others—whether or not this works to their advantage or to their disadvantage or neither. Consider the following three ways of conceiving how power works. First, we tend to talk and think about power when the powerful are seen as, in one way or another, able to bring others under their sway or control, subordinating them and perhaps (but not necessarily) disadvantaging them or acting against their wishes or interests. That idea is captured by the locution ‘power over.’ Sometimes, second, we focus on power as cooperative and constructive, of actors acting together, in concert or in solidarity, combining to achieve shared outcomes, when we speak of ‘power with.’ And thirdly, the expression ‘power to’ suggests ‘empowerment,’ focusing on what can enable actors to overcome hindrances and obstacles, rendering them capable of bringing about significant outcomes. All three ways are worth extensive analysis and discussion, and can, obviously, coexist and interrelate in any given situation. But what I summarize here and elaborate on throughout the book is the first of these understandings of power. We are to consider, in short, what it means to view power over another or others in three dimensions.
Debates about Power
Before doing so, however, I should draw readers’ attention to some other debates that the topic of power has generated. These debates are ongoing, with no resolution in sight that will satisfy all. Let me note in passing that I don’t see that as a matter for regret. Sometimes it may be that there is more to be gained by a dispute continuing than by either side prevailing. As I believe is the case here, the continuing debates can generate insights through the ongoing contestation that might otherwise not emerge. The contest may be perpetual because, as discussed in the text, the concepts in question are ‘essentially contested’—that is, concepts ‘the proper use of which inevitably involves endless disputes about their proper uses on the part of their users’ (Gallie 1956:169).
Alongside the debate over power itself, to which the arguments in this book have contributed and to which we will turn, I have in mind those over domination and over exploitation and the discussion over the nature and ethics of nudging. Power is, in a sense, basic in all these three cases, since each of these is a way in which the social power of some over others is at work.
In the domination debate (among political philosophers and intellectual historians) the key idea is the so-called ‘neorepublican’ notion of freedom as independence or non-domination. Drawing inspiration from classical republicanism it views freedom as the absence of dependence on the arbitrary or uncontrolled will of others—not just as non-interference, or the mere absence of physical or coercive constraint (Skinner 2025). Freedom as non-domination is robust across counterfactuals, across variations in how far others are hostile or friendly. Doors are open and the doorkeepers are restrained. You recognize it when people ‘securely enjoy resources and possessions to the point…where they can look each other in the eye,’ when they are free from ‘the fear or deference that a power of interference might inspire’ (Pettit 212:105, 84). Domination, in short, is ‘a kind of power, and usually social power, over other people’ (McCannon 2018)–an asymmetric interrelationship involving imposition and an imbalance of social power; it is arbitrary or uncontrolled and relatively stable over time; and it is morally serious, constituting grounds for resentment and indignation on the part of those subject to it (who, however, may not experience these because they see it as legitimate). The domination debate is entirely agent-focused, involving disagreements about what counts as ‘arbitrary.’ Extended to cover patriarchy, colonialism and various forms of private, social and public tyranny, the unifying theme is that to have ‘a master—good, bad or indifferent—is to be dominated’—that is ‘subject to the will or pleasure or the power holder’ (Lovett 2012:121, 112). And domination is an ever-present potential, at work even when dormant and unactualized.
In the exploitation debate (mainly among Marxism-influenced economists, and analytic-Marxist and liberal philosophers) the key idea is extraction of surplus from the labor of others: exploiters benefit from what the exploited produce, often, and typically under capitalism, with their willing consent. On the Marx-derived view, surplus value is extracted from workers even if they are paid a fair wage by the standards of the capitalist system because the means of production are owned by a few. Exploitation in general exists because capitalists own assets and don’t have to work, while workers lack them and must work for others (Roemer 1986). This unequal ownership is perpetuated by power, seen as structural or institutionalized, in the form of the ‘mute compulsion’ (Mau 2023) of the relations of production. On a liberal view, there is no exploitation if the consent is genuinely voluntary and mutually beneficial, as expressed in contracts, unless there is force, fraud or deception. Some, however, see such consent as merely formal and so not genuine when given by those unfairly disadvantaged because of a weak bargaining position or because they are ill-informed, desperate or lack better alternatives (Werthaimer 1996). Theories of exploitation are concerned with the control and distribution of the surplus from transactions that are mutually beneficial and consensual; and the debate has been over what nonetheless can make them exploitative, and perhaps the most plausible ties it to domination, as the abuse of power where A exploits B if A benefits from a transaction in which A dominates B.. On this account, exploitation ‘is the dividend A extracts from B’s servitude.’ (Vrousalis 2023:73).
The debate over ‘nudging’ is over the power to influence people’s choices by designing environments that steer their decisions, without restricting their options, imposing significant costs or changing their incentives. Nudging is defended on the ground that it enables them to make better decisions, in their own interests, enabling them to avoid errors, procrastination and risks. Critics argue that it bypasses rational deliberation by exploiting cognitive biases and involves manipulation without their awareness. Is this an abuse of power, a covert kind of control, a form of libertarian paternalism, deciding what is best for people independently of their own judgment? Is it effective in the longer run? Does it reinforce inequality, working to the advantage of governments, corporations and employers? Who gets to nudge whom? Or can one see nudging as a beneficent, albeit paternalistic, form of power over others, serving their interests? Should we not recognize the case for paternalism and accept that people are driven by cognitive biases and poor reasoning? Or is nudging inherently manipulative, violating people’s autonomy, using psychological trickery, relying on people’s biases rather than seeking to ‘boost’ their decision-making and reasoning capacities across different contexts? Research, it is reported, shows that teaching people simple actionable skills, such as understanding risks and probabilities, can outperform nudge-based interventions (Grigorenzer 2015 and Grigorenzer et al 2018).
Turning now to the debate over power itself, we should recall that viewing power three-dimensionally is to go beyond what is apparent because directly observable. The first dimension reveals what is overt: what is, so to speak, on the surface, answering the question: who prevails when there is an observable conflict or resistance? Conflicts can range from the interpersonal to the international, and from battles of will to decision-making situations. One-dimensional power can be explicitly coercive, where people comply because threat induces fear. Or it may simply consist in winning by following a power-conferring procedure such as voting, where one side defeats the other. In the second dimension, power is covert, but still observable: what, upon investigation, becomes observable is how the surface is configured from beneath, determining what is at stake when there is observable conflict on the surface. Who decides what is to be decided and what excluded? Grievances may be suppressed or neglected. Who controls the agenda? This has been called the ‘mobilization of bias’.
As for viewing power in the third dimensions, exploring what lies beyond direct observation is what it requires and invites, revealing what is hidden or latent. This is therefore especially challenging to the social scientist seeking empirical evidence for its presence. When speaking of power, we always mean, I assume, to attribute it to agents (individual or collective), not impersonal structures or forces (as in the writings of Michel Foucault) but it may exist without being deliberately exercised and, when it secures the compliance of another or others, that may serve the desires and purposes of the powerful or else their interests. Consensus may seem to prevail: observable conflict and resistance may be concealed, in hiding, even largely absent, and compliance may only appear to be willing. Here people may want what others may want them to want and behave accordingly. In this dimension symbols and rituals–such as a flag and singing the national anthem–and cultural influences shaping what people believe and are committed to, come into view.
Furthermore, the dimensions are mutually inseparable. Power viewed from within each significantly affects power viewed from within the others. For example, censorship can induce anticipatory compliance (what Timothy Snyder has called ‘obeying in advance’) as in the case of self-censorship, which can in turn lead to belief: people may be overtly coerced into accepting boundaries to what can be said within public discourse and then come to internal acceptance, even the embracing, of what those boundaries enclose. Conversely, symbols and rituals can incubate resistance to prevailing power relations, leading people eventually, when the circumstances look propitious, to engage in overt conflict, transgressing the boundaries and defying the coercive power of the powerful by overcoming their fear of their threats.
These three-dimensional features are present in the power exemplified in the three debates just considered. Neorepublican power as domination may lurk unexercised as an ever-present potential. And yet that potential may only be effective in inducing compliance if coercion, or force if coercion fails, is exercised from time to time (pour encourager les autres, as Voltaire so memorably put it). Exploitation is, as indicated, especially on the Marx-influenced account, a matter of impersonal compulsion not directly attributable to the agents who benefit from it, who may themselves be dominated. And the nudging may be built into the prevailing choice architecture and thus unattributable to any particular agent’s purposive manipulation.
Power in the Interregnum
In the United States and across the world we have over the last decade seen the rise across the world of authoritarianism for which ‘fascism’ is an imperfect label, not least because of the ways in which power has, without the massive deployment of military force, secured not just the compliance but the consent of majorities.
I shall focus here on how what is distinctive and sometimes innovative in the ways power has been understood and deployed in the United States in the early months of the second Trump administration in the United States, across all three dimensions. I do so partly because of its disproportionate global impact, and partly because of what these might portend for the future, in the United States and elsewhere.
Consider first President Trump’s own understanding of power, which is all about winning and losing. He is often correctly described as ‘transactional.’ This amounts to seeing life and the world one-dimensionally, and thus reductively, in terms of power as zero-sum, where one party wins and the other loses and, moreover, where what the one gains the other loses. Furthermore, if there are additional, ‘positive-sum’ gains to both and/or to others, these are seen as byproducts, as successes to be chalked up to the credit of the winner rather than being the goal or point of the transaction. Among countless examples of this conception of power in action, consider Trump’s tariff policy, in which he embraces ‘eighteenth-century mercantilism as his economic philosophy, viewing all international economic relations as bilateral, win-or-lose scenarios, determined by who has a favorable balance of payments (based on the exchange of goods but not services).’ (Browning 2025),
Machiavelli famously wrote that for the Prince ‘it is much safer to be feared than loved when one of the two must be lacking,’ though ideally it would be better to be both loved and feared. That ideal is fully actualized at the time of writing: the Trump administration has hitherto successfully and overtly deployed its power of fear through extensive intimidation and of love through the magic of Trump’s continuing charisma. The intimidation has been widespread, notably in academia, with the early capitulation of Columbia University, and the legal profession, with that of the legal firm Paul Weiss. It has pervaded all government agencies in the face of mass firings, the elites of the Democratic Party and even the Justice Department. It has resulted in widespread compliance and has inhibited resistance. And Trump rallies are festivals, collective effervescences, where love is repeatedly and explicitly expressed by the leader and returned by the followers.
The second dimension of power exposes the underlying setting of the agenda within which first-dimensional power is exercised. In the debate hitherto, this aspect has focused attention on selection, on the sidelining of issues inconvenient to the powerful, on what, as mentioned, E.E. Schattschneider memorably named ‘the mobilization of bias.’ What we have been seeing under the second Trump administration is in a way the opposite, namely what Steve Bannon recommended as ‘flooding the zone’ (‘The real opposition is the media. And the way to deal with them is to flood the zone with shit’). This way of setting the agenda could in turn be named ‘demobilization through chaos.’ Within a few days of inauguration, the New York Times reported:
Firing inspectors general. Sweeping clemency for Jan. 6 defendants. Investigations of perceived enemies. A federal hiring freeze. Moving to end birthright citizenship. An immigration crackdown. Terminating diversity, equity and inclusion programs. Revoking security clearances.
On Tuesday, just when Democrats thought they might come up for air, news broke that Mr. Trump had ordered a freeze on trillions of dollars in federal grants and loans, prompting a new round of outrage (Broadwater 2025)
The effectiveness of this way of deploying power relies, not on the content of the flood but on the continuing distraction of attention (Hayes 2025). Focused attention is, of course, a precondition for any approximation of democratic deliberation.
But this raises the question of how much of such deliberation had in any case survived in the public sphere into the second Trump era. Across the deep divide between the parties due to ever-deepening polarization there was precious little and ever less. In their book Democracy for Realists Christopher Achen and Larry Bartels observe that for ‘most citizens for most of the time, party and group loyalties are the primary drivers of voter choice.’ (Achen and Bartels 2017:299). As polarization has intensified, politics at the elite level, in the legislature and other public forums, becomes ever more purely confrontational, and compromise across party lines disappears, while citizens largely inhabit separate informational ecosystems. In an evenly divided electorate, all that is required to achieve an electoral outcome with massive, transformational consequences, is the power to win the votes of a sufficient number of voters to achieve victory, especially if institutional and constitutional constraints then fail to restrain their implementation.
That power, wielded by Trump and the hollowed-out Republican party, achieved victory in 2024 with a margin of 1:5% and has largely retained the support of its voters since. That power is in several respects, third-dimensional. Their support is, to all appearances, consensual. It is secured by mechanisms and processes of which they are largely unaware, or to which, where they are aware, they are indifferent. Moreover, it works against many of their interests (such as social security and public health) and favors those of the powerful.
What is distinctive and innovative in the deployment of this power is a kind of epistemic liberation. This is a development unfamiliar in modern democratic states. It dates back to the very beginning of the first Trump administration, when the newly elected President made a wildly exaggerated and demonstrably false claim about the numbers attending his inauguration. This began the practice, that has continued ever since, of a rhetoric freed from factual constraints. This deploys signals and dog-whistles evoking sentiments rather than providing information and making arguments to encourage belief. Already in 2005 the comedian Stephen Colbert named this ‘truthiness’ but its repeated use during the first and now second Trump administrations, delivered from the government and accepted as normal by a sufficient proportion of citizens, is a significant American contribution to what political scientists call ‘democratic backsliding.’ Is the backsliding now a forward march into a new social and political order?
Another innovation, encouraged by the Republican party, likewise relying on signaling, aims to build the future on social media, notably Tik Tok and X (formerly Twitter), engaging the party’s young extremist fringes. A New York Times article ‘Trolling Democracy’ describes a ‘new way of pushing for real change through memes and jokes,’ thereby ‘updating a dark art for the platform era—one that is ruthless, inflammatory and designed for maximal viral reach.’ (Pemberton 2025)
I shall purposely refrain from commenting on the implications of the current rapid and unexpected developments of Artificial Intelligence for how we are to think about power, except to observe that they obviously empower those with the expertise to use it over those who lack it. This reluctance is because there is no way to make confident assertions about the impact of these developments while they are happening, especially given the intensity and extent of the debates they have occasioned and the extravagant claims alongside anxiety and skepticism about what lies ahead.
On one issue, however, it is worth commenting, by way of conclusion to this Preface. I suggested above that the concept of power in social and political contexts is linked to that of agency. We attribute such power to agents, individual or collective, not to impersonal structures or forces–or, for that matter, to machines or algorithms. We hold the powerful responsible and accountable and assume that their power advances their interests. We are interested, in short, in the power of humans. The elation and anxiety in the debates around AI have, among much else, been about whether with large AI models we are on the cusp of creating autonomous, intelligent—even super-intelligent—agents with artificial general intelligence (AGI).
It has been argued that the very idea that this is possible is ‘fundamentally misconceived’ (Farrell, Gopnik, Shalizi and Evans 2025). The argument is that large AI models constitute a cultural and social technology—the latest way of accessing and organizing information across space and time. There are technologies like printing, film and video and from libraries to newspapers to internet search—with profound effects on human thought and society. Likewise, the social institutions of markets, states and bureaucracies co-ordinate information-gathering (all imperfectly) and decision-making. These cultural and social technologies are only possible because humans have distinct capacities characteristic of intelligent agents, with a conception of truth and falsity (and the ability to distinguish the one from the other), building new models of the external world in the light of new evidence, and developing new goals and acting on them. Large models lack these features, even when ‘scaled.’ They ‘simply sample and generate texts and images;’ they summarize and extract statistical patterns and reconstruct unmanageably large and complex bodies of human-generated information, abstracting it and rendering it tractable. They also rely on human judgment and knowledge reinforcement learning from human feedback.
These immensely impressive developments do not, however, license our natural temptation to view them anthropomorphically (Marcus and Luccioni 2023), and for a further reason. What all AI models also lack is a further feature that is species-specific, distinguishing humans not only from machines but also from non-human animals, namely the capacity of independent judgment: to stand back and distinguish what is moral from what is merely customary and conventional (Turiel 1983, Lukes 2025).
These last observations lead me to reflect on the project of this book–to explore and render explicit what it is to have power over another or others. That question is of the deepest human concern, a concern that is perennial. It requires us to look further and deeper beyond what may appear to be a prevailing consensus into ways in which the power of human agents can have significant consequences without being directly observable and without even being exercised. Such a project does, of course, lead, to facile and unsupported claims of responsibility and, at the extreme, to conspiracy theories, and so it poses especially stringent challenges, especially at the present time, to the inquiring social scientist.
