In the original understanding of the Stanford Prison Experiment, the public policy take away is the anti-prison argument: you can try to set up a prison system that treats the prisoners in a kind, reforming way, but the guards will treat the prisoners in a brutal, deforming way because that is what humans with power do automatically.

The kind of experiment that would provide convincing evidence would have a pro-prison psychologist trying to rig the experiment by tipping off the guards that they are supposed to show the best side of human nature and treat the prisoners in a kind, reforming way; but it all goes horribly wrong when the guards ignore the push from the experimenter and are brutal anyway.

What do we expect would be the outcome of an experiment where the pro-prison psychologist running it tips off the guards about how they are supposed to behave? Zimbardo's work is rather ambiguous. Perhaps it generates the expectation that the guards will be brutal. Perhaps it generates the expectation that the guards will conform to the clues and hints injected by the psychologist running the experiment. The work suffers from a fundamental wrong way roundness in relation to the public policy controversy.