Power is the ability to get your way even against resistance; authority is power that others accept as legitimate. The difference matters more than it looks, because the most durable influence is rarely forced. And the most famous experiment about obeying authority turns out to say something subtler, and more unsettling, than the version everyone remembers.
Power is the capacity to produce the effects you intend, especially against resistance. Authority is power that others treat as legitimate, that they accept as rightful rather than merely submit to. The gap between them is the whole game: power backed only by force is expensive, brittle, and resented, whereas authority, power people accept, is cheap, stable, and largely self-enforcing. Most lasting influence runs on legitimacy, not coercion.
Weber gave the still-dominant map of where legitimacy comes from, three ideal types of authority. Traditional authority rests on custom and inherited position ("this is how it has always been done"); charismatic authority rests on the extraordinary personal qualities people attribute to a leader; and legal-rational authority rests on impersonal rules and the office rather than the person (the authority of a form, a court, a job title). Real institutions mix all three, and Weber noted they tend to convert into one another, charisma in particular usually "routinises" into traditional or legal-rational forms once the founder is gone, because personal magic does not survive succession (Weber, 1922).
At the sharper end, French and Raven catalogued the bases of social power in everyday relationships and organisations: reward (you can give me something), coercive (you can punish me), legitimate (you have the right to direct me), referent (I identify with you and want your approval), and expert (you know things I need), with informational power added later (French and Raven, 1959). Most real influence blends several, and the softer bases, referent and expert, tend to produce more durable compliance than reward or coercion.
Then there is the evidence everyone has heard of, and it is worth actually describing rather than just naming. Milgram recruited ordinary volunteers for what he told them was a study of learning and memory. Each was assigned the role of "teacher" and sat at a console of switches labelled in rising steps, from a mild 15 volts up to 450, the last switches marked "Danger: Severe Shock" and simply "XXX". In another room, a man they believed was a fellow volunteer, actually an actor, was strapped in as the "learner", and every time he got an answer wrong the teacher was told to deliver the next shock up. No real shocks were given, but the volunteers did not know that. As the voltage climbed the learner protested, cried out, complained of a heart condition, and eventually fell silent, while a calm experimenter in a grey coat simply told the teacher to continue.
The finding that stunned the field was how far people went. In the baseline version, about two in three volunteers, 26 of 40, went all the way to the final 450-volt switch, not because they were cruel or wanted to, but because an authority in a lab coat kept telling them the experiment required it (Milgram, 1963). The result landed like a thunderclap, seeming to show that ordinary, decent people will do terrible things simply because someone in charge instructs them to. Hannah Arendt had reached for a similar idea the same year in her report on the trial of Adolf Eichmann, a Nazi administrator who organised deportations to the death camps: her phrase "the banality of evil" captured the disturbing thought that monstrous harm is often carried out not by monsters but by unremarkable people treating atrocity as a job to be done well.
A different tradition questions where power even sits. Foucault shifted attention from power as top-down prohibition to power as productive and capillary: disciplinary power that shapes people through routines and institutions, biopower that manages populations, and "governmentality" that works through knowledge and norms rather than open force, so power is not only what says no but what quietly makes us who we are (Foucault, 1975). Lukes made a compatible point with his three faces of power: the visible power to win decisions, the subtler power to keep issues off the agenda entirely, and the deepest power to shape what people want in the first place, so they never even form the grievance (Lukes, 1974).
The centre of gravity here is the reappraisal of Milgram, and it is a model of how a famous result can be real and still widely misunderstood. Working from Milgram's own archives and recordings, historians and psychologists found the "obedience" story too simple. Many participants agonised, questioned, and resisted; the experimenter often pushed them well beyond the scripted prompts; and crucially, those who continued tended to do so because they had come to identify with the experiment as a worthy scientific endeavour, not because they had switched off and become mindless agents. The better description is engaged followership: people go a long way for a cause they believe is legitimate and good, led by someone they trust as its representative (Reicher, Haslam and Smith, 2012; the archival case is Perry, 2013). This is more disturbing than the cartoonish version, not less, because it means harm is done not by robots surrendering their judgement but by committed people who think they are serving something worthwhile.
None of this makes the underlying phenomenon fake. When Burger ran a careful partial replication decades later, obedience rates were broadly comparable to Milgram's, so the effect is robust even as its interpretation shifts from "blind obedience" to "identification with authority" (Burger, 2009). The lesson changed; the danger did not.
The Foucauldian and radical strands draw a different criticism. When power is described as being everywhere, in every relationship, every institution, every norm, the concept can lose its edge: if everything is power, the word stops distinguishing anything and becomes hard to test or falsify. It is a productive lens that can tip into an unprovable worldview.
The classic obedience work is ethically impossible to run in its original form today, so the evidence base is older and thinner than the topic's fame implies. Much of the power theory (Foucault, Lukes) is interpretive rather than experimental, illuminating but hard to pin to hard prediction. And most of it is Western and institutional, so how far the models travel across cultures and informal settings is not fully settled.
Exactly which conditions tip identification-based followership from serving a genuinely good cause into serving a harmful one? How do the classic face-to-face dynamics change in digital and algorithmic authority, where the "experimenter" is a platform or a system? And can the deepest, agenda-shaping face of power be measured, or only argued about after the fact?
The usable core: legitimacy, not force, is what makes influence durable; the softer bases of power outlast the hard ones; and people are moved less by raw commands than by identifying with a cause and a credible authority behind it.
Authority genuinely persuades, but the reappraisal pins down the active ingredient: it is legitimacy, not the trappings of power. Expert and referent bases, real credentials, demonstrated competence, and shared identity, produce more durable trust and compliance than reward or coercion (French and Raven, 1959), and a "because we say so" authority erodes fast. The dual-use line sits exactly here. Authority cues work, which is why manufactured ones, the actor in a lab coat, borrowed or vague credentials, "experts agree" with no experts named, are such a common manipulation, and the popular authority principle in marketing is precisely this lever. The honest version earns the authority it claims; the manipulative version rents its appearance, and it carries a sharp trust cost when the costume slips. There is also a sober organisational lesson in engaged followership: people will go a long way for a leader and a cause they find legitimate (Reicher, Haslam and Smith, 2012), which is a warning to build cultures where questioning is legitimate too, since the failure mode is not rebellion but committed people following a bad direction.
Weber's types explain a great deal of political life: charismatic leaders electrify movements but leave a succession crisis, because personal magic does not transfer, and durable power depends on routinising into legitimate institutions (Weber, 1922). The engaged-followership finding is the sober warning of the field: ordinary, decent people can be led to do harm not by switching off their conscience but by identifying with a cause and an authority they believe in (Reicher, Haslam and Smith, 2012), which places real responsibility on how leaders frame what they ask. Lukes's third face is worth naming honestly too: the deepest exercise of power is shaping what people want so they never form the objection, and recognising it in others, and resisting the temptation to use it, is part of governing legitimately.
Legitimacy is the cheapest and strongest form of authority a state has: where institutions are seen as rightful, compliance is largely voluntary and enforcement is a backstop; where legitimacy fails, everything must be coerced, which is expensive and unstable. That makes procedural fairness and visible rule-following an investment, not a nicety. And the obedience research carries a permanent institutional caution: systems should be designed so that individuals can question and refuse harmful instructions, because the historical danger is not too little followership but too much of it aimed at the wrong cause.
Three habits. Distinguish power from authority in any situation and ask where the legitimacy actually comes from, because that tells you what is durable and what is merely being forced. When you meet an authority cue, credentials, a title, a confident expert, check whether the legitimacy is real or just costumed, since the appearance is the most common manipulation. And remember the engaged-followership lesson in both directions: it is why a cause you believe in can carry you further than you would go alone, for good or ill, so keep asking not just "who is telling me to" but "what am I identifying with, and is it worthy".
.png)