It’s not true, in general, that ‘nice people are Nazis’. But the converse was true; i.e., the average Nazi was a ‘nice’ and ‘good’ and ‘decent’ person, as measured by the standards of their peers. Upwards of five hundred thousand people (only half of them Germans) were involved in the Holocaust (which rendered extinct approximately twelve million people, about half of them Jewish); by far, the vast majority of these people were ‘just doing their jobs’, being nice and agreeable, attempting to make the world a better place. That was their intent (and the stated intent of every single National Socialist). That is one major reason why intent weighs very little next to consequence; sure, there’s a difference between first-degree murder and manslaughter, but that difference is much smaller than the difference between a convicted criminal and an unconvicted civilian.
Ophelia, I think your objection boils down to wariness of making a converse error; it may well be true that most moral actors are disagreeable people, but that does not logically entail that most disagreeable people are moral actors. This is basic logic, but it is beyond most people, especially those who get their news from their networks of friends and colleagues rather than checking the source material. In short, the results of this experiment may well be sound, but your own concerns are still valid–just because someone’s asocial (or mildly anti-social), it doesn’t necessarily entail that they will effect moral outcomes, even if most of those who do effect moral outcomes turn out to be asocial.
As a generally-irascible anti-authoritarian, I like to put myself in the latter camp…but such requires (at least) honest self-reflection, a working moral theory, and the ability to change one’s mind (and subsequent behaviour). Most people who subscribe to the rule ‘if people are mad, you’re doing something right!’ generally fail on these and other essential criteria, and so they generally fail to be moral actors. They do not negate the results of the experiment, but they do limit its scope, and we would be well to keep in mind the laws of logic before drawing erroneous conclusions from its results.