How Organisations Make Cruelty Easier to Carry Out
Why “Just Following Orders” Is Never the Whole Explanation
Cruelty becomes easier to organise when authority makes harming someone appear necessary, respectable or somebody else's responsibility.
The disturbing question is not whether a cruel person can give a cruel order. It is why someone who considers themselves decent might help carry it out, then return home without abandoning that view of themselves.
Ordinary people can obey harmful instructions for several overlapping reasons: trust in authority, commitment to a group, fear of consequences, gradual involvement and beliefs that make the harm seem justified. Some comply reluctantly. Some participate enthusiastically. Others refuse. There is no single psychological switch that turns everyone into the same kind of follower.
The strongest lesson from obedience research and history is therefore conditional. Certain arrangements make cruelty easier to perform and harder to challenge. Understanding those arrangements helps explain how institutions recruit cooperation—and where resistance can interrupt it.
“Ordinary” Does Not Mean Innocent
Calling perpetrators ordinary can sound like an excuse. It should instead describe the limits of a comforting assumption: that serious wrongdoing is committed only by people whose abnormality would be obvious in everyday life.
A person can be considerate towards relatives and brutal towards an outsider. They can be competent at work, loyal to colleagues and willing to support an abusive policy. These qualities do not cancel one another. Their coexistence is part of the problem that needs explaining.
Nor does explanation remove responsibility. Establishing that a person faced pressure tells us something relevant about their choices. It does not automatically establish that they had no choice, understood nothing or should be judged identically to somebody acting under an immediate threat to life.
It helps to distinguish obedience from related behaviour. Obedience involves responding to authority. Conformity involves adapting to a group. Opportunism involves pursuing advantage. Ideological commitment involves believing in the cause. A single harmful act can involve more than one of these.
If we describe every participant as merely obeying, we may overlook those who devised methods, volunteered for responsibility or went beyond what was required. Equally, if we assume everyone acted from personal hatred, we may miss the pressures that drew reluctant participants into cooperation.
The useful question is specific: what did this person believe, what could they see, what alternatives were available and what happened when somebody challenged the demand?
What Milgram's Experiment Actually Established
In Stanley Milgram's obedience research, volunteers were placed in a situation where an experimenter instructed them to administer what they believed were escalating electric shocks to another participant. The apparent recipient was part of the experimental arrangement, and the shocks were not real.
In the famous early condition, 26 of 40 participants continued to the maximum setting, labelled 450 volts. That result became the familiar 65 per cent figure. It belongs to a particular experimental condition, not to humanity as a whole.
The finding remains troubling because participation occurred within a respectable research setting. People were asked to continue an activity presented as part of a legitimate study. Their decisions emerged from that situation, including uncertainty about what was happening and the experimenter's assurances.
Milgram conducted many variations. Archival research by Nestar Russell describes how the procedure developed through pilot work and adjustments. That history matters because the experiment was a constructed social encounter, not a neutral instrument measuring an unchanging quantity called obedience.
It would therefore be wrong to conclude that 65 per cent of people will torture a stranger whenever someone issues an order. It would be equally unhelpful to dismiss the research because it cannot support that sweeping claim. It demonstrates how a particular arrangement can draw substantial cooperation with apparently harmful conduct.
The challenge is explaining which features mattered, how they interacted and how far the findings travel beyond the laboratory.
Why the Familiar Story Is Too Simple
The popular version says that an authority gives an order and the subject's conscience switches off. That account is memorable, but it leaves little room for uncertainty, identification, persuasion or active judgement.
Psychologists S. Alexander Haslam and Stephen Reicher have argued that identification with an authority's project can be central. People may cooperate because they believe they are contributing to something legitimate or valuable, rather than because they have become incapable of thought. Their account challenges the idea of automatic, blind obedience; it is an interpretation supported by research, not the sole uncontested explanation.
This difference changes the practical lesson. If harmful cooperation depends partly on belief in a worthy mission, an institution can become dangerous while its members feel purposeful and morally engaged. The task is then to examine how the mission is being used to justify particular actions.
Later research also requires careful reading. Jerry Burger's study, published in 2009, was a partial replication that stopped at an earlier point to reduce ethical risks. It found substantial willingness to continue under its conditions. It did not recreate the entire original procedure or observe participants completing its maximum sequence.
Its results also complicate easy advice about dissent: in that study, seeing a confederate refuse did not reduce obedience in the way expected. A visible objection can matter without being a universally effective antidote.
The research supports vigilance about social conditions. It does not supply a simple percentage, personality test or guaranteed technique for predicting who will resist.
Authority Can Change What a Person Thinks Is Happening
Authority influences behaviour partly by influencing interpretation. When a trusted professional says a procedure is legitimate, a participant may assume that safeguards exist beyond what they can see. Deferring to expertise is often sensible; modern life would be impossible if everyone independently reconstructed every specialist decision.
The danger appears when that reasonable habit is extended beyond its proper limits. Expertise about a technical process does not necessarily confer moral authority over every use of that process. A manager's power to allocate work does not establish that every instruction is acceptable.
Consider a hypothetical employee asked to withhold inconvenient information from a customer. The request might be presented as routine policy, a temporary measure or something already approved by a senior team. Each explanation supplies a reason to reinterpret the action as less serious than it first appears.
The employee may also lack access to the full facts. Asking for clarification can be a responsible response to genuine uncertainty. Yet an organisation can exploit uncertainty by ensuring that nobody below the top possesses enough information to challenge the official explanation confidently.
That creates an asymmetry. The person giving the instruction knows how the process is described; the person affected by it knows the harm; the employee carrying it out may encounter neither account in full.
A useful safeguard is to connect authority to reasons that can be examined. “Someone senior approved it” should not be the only available explanation for an action that imposes a serious cost on another person.
Small Steps Can Hide a Large Change
Harmful involvement does not always begin with a demand that reveals the final destination. A person may first agree to something limited, then face another request that appears only slightly different.
Imagine a hypothetical service team that begins by shortening complaint responses during a busy period. Later, staff are encouraged to avoid recording certain complaints. Eventually, customers are told that matters have been resolved when no investigation has occurred. Each stage can be defended as a practical extension of the previous one.
The difficulty is that reconsidering the latest step may require reconsidering earlier choices. An employee who objects now might have to admit that the process they previously supported was already causing harm. Maintaining a positive self-image can make that admission uncomfortable.
This is an analytical example of escalation, not a claim that every small compromise inevitably produces serious abuse. People often stop, renegotiate or recognise a boundary. The important question is whether the organisation provides clear moments for reassessment.
Temporary exceptions are especially revealing. Do they expire? Does somebody review their effects? Are staff allowed to restore the ordinary rule when the emergency ends? Without those mechanisms, an exceptional practice can become normal simply because it remains in place.
Preventing escalation requires more than telling people to be brave at the final stage. It means making the earlier changes visible while refusing them is still relatively straightforward.
The Group Can Make Refusal Feel Like Betrayal
People depend on colleagues for cooperation, support and a sense of belonging. A harmful instruction can exploit those relationships by making dissent appear to impose a burden on everyone else.
In an illustrative workplace, a person who refuses to mislead a customer might be accused of slowing the team or leaving others to handle the difficult work. The argument shifts attention away from the customer's treatment towards the dissenter's loyalty.
That framing can be powerful because the social consequences are immediate. The customer may be distant, while the disappointed colleague is standing nearby. A person can care about both and still feel the nearer pressure more strongly in the moment.
The same process helps explain why leaving a controlling group can involve more than rejecting an idea. Relationships, identity and practical support may have become attached to participation. The situations are not identical, but both raise questions about the cost of refusal.
A healthier group can reverse the meaning of loyalty. Protecting patients, customers or the public can become part of what it means to serve the organisation properly. A colleague who raises a concern then becomes someone defending the mission rather than abandoning it.
The decisive issue is which behaviour the group actually rewards. A statement welcoming challenge means little if the first person to challenge a decision loses status, opportunity or support.
How Language Makes Harm Easier to Handle
Psychologist Albert Bandura described moral disengagement as a set of ways people can detach harmful conduct from the standards they normally apply to themselves. These include presenting harm as justified, minimising consequences and shifting responsibility. His account helps explain how a person can preserve a moral self-image while participating in wrongdoing.
Language is one route. An action described in administrative terms can become easier to discuss than the same action described through its effect on a person. Technical vocabulary is often necessary, but it can also conceal the human consequence that should remain part of the decision.
Consider the difference between recording that a case has been closed and establishing whether the person received the help they needed. The first is a measurable event in a system. The second requires attention to the outcome. Confusing them can make a process look successful while leaving the underlying harm untouched.
The appropriate response is not to ban professional language. It is to require a plain description alongside it. Who will lose access? What information is being withheld? What happens if the decision is wrong? Who can reverse it?
Those questions restore details that a label may omit. They also make disagreement more precise. A person can challenge a concrete consequence more effectively than an abstract policy name.
The moral importance of clear language lies in keeping the action available for judgement. If nobody can describe what is happening without euphemism, the organisation has already made scrutiny harder.
Dividing the Work Can Divide the Sense of Responsibility
Large organisations depend on specialisation. One person collects information, another checks it, another approves a decision and another implements it. This division can improve accuracy and efficiency. It can also allow each participant to see only a fragment of the consequences.
A hypothetical employee might say that they merely entered the data. A supervisor might say that they only applied the rule. A senior manager might say that the system produced the result. Each statement could describe a real part of the process while leaving the overall responsibility unclear.
The problem is not solved by making everyone responsible for everything. That would be impossible and could leave nobody with a defined duty. Responsibility needs to be assigned at the points where a person can understand and change the relevant decision.
For example, someone should own the accuracy of the information, someone the justification of the rule and someone the mechanism for correcting mistakes. A route for concerns must connect those responsibilities when the harm crosses departmental boundaries.
This is particularly important when the affected person encounters only the final employee in the chain. That employee may have little power to help, yet the person seeking redress may have no access to anyone else.
An institution that cannot identify who can reconsider a harmful outcome has created a practical form of irresponsibility, even if its organisational chart contains many people with impressive titles.
History Requires More Than a Laboratory Analogy
The Holocaust cannot be explained by applying a laboratory percentage to a population. It involved antisemitic ideology, state power, organised persecution and the decisions of many different participants across occupied Europe.
The United States Holocaust Memorial Museum describes collaboration ranging from administration and confiscation to deportation and direct killing. It identifies varied motives, including antisemitism, personal advancement, greed, fear and peer approval. That diversity matters: “following orders” cannot explain everyone in the same way.
Some people supported the objective. Some pursued opportunities created by persecution. Others cooperated under pressure. Understanding the distinctions helps establish responsibility rather than dilute it into a general statement about human nature.
It also prevents a misleading comparison with ordinary organisational conflict. A dispute at work is not equivalent to genocide. A psychological mechanism can appear in settings with radically different stakes, but the comparison must not erase the scale, ideology or violence of the historical event.
The transferable question is narrower: how do institutions make harmful participation possible, respectable or advantageous? Answering it requires examining rules, incentives and opportunities for refusal alongside individual beliefs.
History is especially important because it includes action beyond compliance. People can organise, improvise and pursue cruel objectives with initiative. A theory focused only on obedient subordinates would miss the ambition and creativity that can also sustain wrongdoing.
My Lai Shows Why Resistance Belongs in the Story
On 16 March 1968, US Army helicopter pilot Hugh Thompson and his crew encountered the killing of unarmed civilians at My Lai in Vietnam. The Army's account describes Thompson intervening, placing his helicopter between soldiers and civilians and helping arrange their evacuation.
This was resistance within a military setting, not an absence of institutional identity. Thompson's action illustrates how obligations can conflict: loyalty to fellow service members does not have to mean cooperation with their wrongdoing.
The example should not become an easy claim that anyone can safely resist under any circumstances. Thompson possessed particular resources, a crew and an opportunity to intervene. Other people facing coercion may have fewer options and face severe risks.
Its significance is that participation was not mechanically inevitable. People in the same broad institution could act differently when confronted with harm. That fact challenges the fatalistic conclusion that understanding pressure means accepting the outcome.
It also broadens the definition of refusal. Resistance can mean more than declining to perform an act personally. It can involve protecting someone, interrupting a process, recording evidence or bringing another source of authority into the situation.
Which response is possible depends on the circumstances. The historical lesson is not a universal script. It is that moral action can require deciding which institutional duty deserves priority when colleagues are violating it.
Why “I Would Never Do That” Is an Incomplete Defence
Confidence about our future behaviour can be comforting because we imagine the situation with its meaning already clear. We picture an obviously wrongful order, recognise it immediately and assume that refusal will be a single decisive moment.
Real situations may contain uncertainty, competing obligations and incomplete information. A person may not know whether an instruction is unlawful, whether the apparent victim has relevant information or whether challenging the process will improve matters. Those uncertainties deserve investigation rather than automatic compliance.
The point is not to insist that everyone would fail. It is that a declaration of character does not provide a plan for acting under pressure. Someone can sincerely value fairness and still lack a practical way to raise a concern.
Preparation can make the response more concrete. Identify which decisions require independent review. Know how to record disagreement accurately. Establish who can be contacted outside the immediate chain of command. Avoid accepting personal assurances as a substitute for checking consequential claims.
These are proposals for institutional practice, not guarantees derived from an experiment. They work by changing the available options and the quality of information rather than assuming that courage alone will solve every problem.
The aim is to reduce the distance between recognising a concern and being able to do something useful about it.
AI Could Give Orders a New Appearance of Neutrality
An instruction generated or supported by AI may look less personal than a manager's demand. A score, ranking or automated recommendation can appear objective even when it reflects disputed assumptions, incomplete data or a poorly chosen target.
Consider a hypothetical service where staff are told to follow an automated risk assessment. If nobody can explain the assessment and overrides damage performance ratings, the employee's formal discretion may be much smaller than it appears.
This would not mean that AI had become a moral authority by itself. People would have designed the workflow, selected the tool and attached consequences to compliance. The machine could make those human choices harder to see.
There are potential benefits too. A well-designed system could identify inconsistencies, preserve decision records or flag treatment that violates an organisation's stated rules. Whether it helps depends on who can inspect its outputs and whether findings can challenge powerful decision-makers.
The essential questions would remain concrete: who selected the objective, who checked the evidence, who can override the result and how can the affected person obtain a review? A human signature at the end is insufficient if the human lacks the means to disagree.
Make Refusal an Available Action
Institutions often praise ethical behaviour in general terms while organising daily work around speed, loyalty and numerical results. A serious commitment to preventing cruelty requires those priorities to be reconciled when they conflict.
Employees need clear authority to pause consequential work, access to independent review and protection that operates in practice. Leaders need information about outcomes, including complaints that challenge the story told by performance measures. People affected by decisions need a route to someone who can change them.
The warning in Shirley Jackson's The Lottery concerns a community that preserves a procedure while accepting its terrible result. Real institutions have an opportunity fiction can deliberately withhold: they can redesign the procedure before the next person is harmed.
The most useful test is what happens after someone says that an instruction is wrong. Are they asked to explain and given a fair hearing? Is the affected person considered? Can the action stop while the concern is assessed?
A culture's moral claims become visible at that moment. If refusal is punishable, review is inaccessible and responsibility always belongs elsewhere, decent intentions will have little room in which to act.
Sources
Haslam and Reicher — Contesting the “Nature” of Conformity: What Milgram and Zimbardo’s Studies Really Show — The experimental context and the identification-based challenge to the blind-obedience interpretation.
Nestar Russell — Milgram’s Obedience to Authority Experiments: Origins and Early Evolution — The abstract’s account of experimental variations, pilot work and procedural development.
Jerry Burger — Replicating Milgram: Would People Still Obey Today? — The abstract’s findings and limits of the ethically modified partial replication.
Albert Bandura — Moral Disengagement in the Perpetration of Inhumanities — The abstract’s account of justification, minimised consequences and displaced responsibility.
United States Holocaust Memorial Museum — Collaboration and Complicity During the Holocaust — The range of participants and motives in historical collaboration.
US Army Combined Arms Center — My Lai at 50 — Hugh Thompson and his crew’s intervention to protect civilians at My Lai.
Next Reads
The Lottery Explained: Why Ordinary People Accept a Horrifying Tradition — A fictional community in which maintaining tradition becomes a way to accept harm.
Why Do Intelligent People Join Cults—and Why Is Leaving So Difficult? — How belonging and dependence can increase the costs of refusal.
The Circle Explained: What Happens When Privacy Becomes Suspicious? — A cultural examination of institutions that redefine dissent and privacy as suspicious.
