The slide from emergency command into one-person rule is a plausible pattern, but it is a descriptive model with no study behind its six steps, and the experiments usually cited for it, the Stanford prison study and Milgram’s obedience studies, support a narrower claim than they are used for: harmful conduct is easier to produce than people expect and depends on leaders who frame it as a worthy cause, not on a fixed tendency to turn cruel. The sound defence is structural, with limits on emergency power that do not rely on trusting the person holding it.
Part of Leadership Psychology in Crisis
The authoritarianism slide is the gradual conversion of temporary emergency authority into permanent unchecked control by one leader.
What The Parent Describes, And What It Is
The parent article lists six steps: a crisis demands unilateral decisions and they work; the leader extends the habit to non-crisis matters because it is faster; questioners are seen as obstructive; information is restricted to an inner circle; dissent is punished and yes-people gather; and the community is under authoritarian rule without the leader noticing how. Its prevention list is a council consulted on non-emergency decisions, protected dissent, regular confirmation and self-check questions.
The sequence is a reasoned account, not a research result. It cites no study, and it is not shown that groups pass through the steps in this order, or at all. The same is true of most stage models in social psychology (Tuckman’s forming-storming-norming-performing, for instance, is memorable and descriptive but does not reliably predict how real groups move). It should be used as a checklist of warning signs, not as a timetable.
The Famous Studies, Checked
Two experiments are usually offered as proof that power corrupts ordinary people. Neither shows that.
The Stanford prison study (1971). Philip Zimbardo had university students play guards and prisoners in a basement mock prison. It was stopped early, after about six days of a planned two weeks, and is often summarised as showing that ordinary people become brutal when given power. (Figures from memory: roughly two dozen participants selected from about seventy-five volunteers; not rechecked.) Several problems have since been documented. Thibault Le Texier’s 2019 review of the archive (published in American Psychologist, cited from memory) found that the guards were briefed to be tough and that some participants later said they were acting as they thought was expected. The sample was small and self-selected, and there was no comparison group. It is a demonstration of what can happen when a staff member tells people to behave cruelly, not a measurement of what people do on their own. A later experiment run by Stephen Reicher and Alexander Haslam for the BBC (2002, published 2006, from memory) set up a prison again without such instruction; the guards did not take on the role cleanly and no uniform brutality emerged, though the group dynamics were unstable in other ways.
Milgram’s obedience studies (1961 onward). At Yale, participants were told to give increasing shocks to a person they could not see; in the best-known condition 26 of 40 (65 percent) went to the maximum. That figure is one condition of many, and the variations are the real finding: compliance fell sharply when the authority was absent, when the setting looked less official, or when other participants refused. (From memory, a figure of about 10 percent was reported where two peers refused; not rechecked.) Archival work by Gina Perry (2012) and by Haslam and colleagues documents departures from the stated protocol and participants who doubted the shocks were real. Jerry Burger’s 2009 partial replication stopped at 150 volts for ethical reasons and found a rate close to Milgram’s at that point (from memory). The safe summary is that obedience is situational and easier to produce than people expect, never that a fixed share of humanity will obey.
What A Better Reading Suggests
Haslam and Reicher have argued that the cases involve engaged followership more than blind obedience: people comply when they identify with the cause and believe the leader represents it. This is a contested but well-supported reading, and it matters here for two reasons.
- The slide does not require a cruel leader. A leader sincerely serving what the group sees as a worthy goal can draw people into harm, and the leader is also drawn in.
- A leader’s endorsement is itself part of the mechanism. Narrow, explicit instructions on what is allowed matter more than diffuse pressure.
An older strand, Robert Michels’s iron law of oligarchy (1911, from memory), claimed that organisations tend to concentrate control in a few hands regardless of their founding principles. It is disputed as a law, but it is an observation about structure, and it supports the same practical conclusion.
Design Implications Worth Taking
Because the evidence does not support “good people stay good and bad people go bad”, the defence has to work even if the leader is well-meaning and well-informed.
Step 1: Limit the term of emergency authority. Attach an end date or a trigger at the time powers are given, and require an affirmative decision to extend. Compare Emergency Decisions.
Step 2: Make consultation mandatory, not optional. A council that must be consulted on non-emergency matters, as the parent says, is a structural rule. See Council Formation and Council Models.
Step 3: Keep information outside the leader’s control. Records held by someone else, open decisions, and the group’s access to the facts. See Decision Archives and Information Hoarding.
Step 4: Rotate or share the highest roles. See Rotating Leadership.
Step 5: Protect an exit and an appeal. Members need a way to object that does not cost them everything, and ordinarily a route outside the leader, as in Conflict Resolution.
Step 6: Name the warning signs in advance. Agree with the group what would count as the slide: questions treated as disloyalty, decisions made without the council, facts withheld. A shared definition made while things are calm is much harder to dismiss later.
What The Evidence Does Not Settle
- Whether the six-step order is typical. No data supports it.
- How much of the pattern is the leader’s character and how much the circumstances. The studies cannot separate them, and no one should claim they do.
- Whether safeguards of this kind work in small post-collapse groups. They have a long history in larger institutions; evidence for groups of this size is thin.
Common Mistakes
- Citing the prison study as a law of human nature. It should be cited as an illustration of how instructions and expectations shape behaviour, if at all.
- Citing Milgram as a percentage. Use the range of conditions, not 65 percent.
- Trusting the leader’s own sense of restraint. The parent’s last step says the leader does not realise how they got there.
- Building safeguards that the leader controls. A council appointed and dismissed by the leader checks nothing.
What’s Next
- Learn next: Information Hoarding — the fourth step, examined on its own
- Combine with: Leadership Structures — the design choices that carry the safeguards