The Stanford prison experiment, designed and conducted by Philip Zimbardo in 1971, is one of the most famous experiments in psychology. In the basement of Stanford University an imitation of a prison was built. Students who participated in the study were randomly assigned the role of prisoner or guard. Zimbardo himself also participated in the experiment in the role of superintendent. The standard interpretation of the findings of the study is something like: after a while prisoners started to behave helplessly and submissively while guards started behaving cruel and abusively. The experiment was stopped after a student who conducted interviews in the imitation prison objected to the cruelty of what was happening.
Maria Konnikova has written an article about the experiment in response to a movie which now has been made about the experiment. The article has a critical tone. She points to the many problems in the way Zimbardo has designed and conducted the experiment. She says the interpretation of the study's findings isn't as clear as is often thought (being that ordinary people harbor ugly potentialities). One problem is that the situation in the imitation prison was strongly manipulated and that it was very much expected of the guards, through the instructions they received, to act in a rather brutal way. This contradicts the often heard suggestion that the experiment would demonstrate how the role of a guard would more or less automatically evoke abusive behavior.
A second problem is that the students had volunteered to participate in the study and thus were not necessarily a good representation of 'ordinary' people. A third problem was that Zimbardo and other researchers participated in the experiment which does not enhance objective observation and interpretation. Then there is the note that only a third of the guards, only four people, started behaving brutally, which contradicts the general impression that guards in general started behaving brutally. In her article, Konnikova mentions several other problems in the study (see her article). Konnikova says that the lesson of the experiment is not that any random human being is capable of descending into sadism and tyranny. She says the lesson is that certain institutions and environments demand those behaviors—and, perhaps, can change them.
My suggestion is to draw yet another lesson from the experiment. This lesson is: let's use the experiment as an example of how not to design and conduct a social scientific experiment. The experiment was quite flawed in many ways and therefore we cannot draw any clear conclusions from it about human nature and behavior.