Did trials by ordeal work?

Jonathan Adler at The Volokh Conspiracy links to a Boston Globe article suggesting trials by ordeal might have actually worked.

For the better part of a millennium, Europe’s legal systems decided difficult criminal cases in a most peculiar way. When judges were uncertain about an accused criminal’s guilt, they ordered a cauldron of water to be boiled, a ring to be thrown in, and the defendant to plunge in his naked hand and pluck the object out. The defendant’s hand was wrapped in bandages and revisited three days later. If it survived the bubbling cauldron unharmed, the defendant was declared innocent. If it didn’t, he was convicted.
....
First, consider the reasoning of the defendants. Guilty believers expected God to reveal their guilt by harming them in the ordeal. They anticipated being boiled and convicted. Innocent believers, meanwhile, expected God to protect them in the ordeal. They anticipated escaping unscathed, and being exonerated.

The only defendants who would have been willing to go through with the ordeal were therefore the innocent ones. Guilty defendants would have preferred to avoid the ordeal - by confessing their crimes, settling with their accusers, or fleeing the realm.

The next thing to understand is that clerics administrated ordeals and adjudged their outcomes - and did so under elaborate sets of rules that gave them wide latitude to manipulate the process. Priests knew that only innocent defendants would be willing to plunge their hands in boiling water. So priests could simply rig trials to exonerate defendants who were willing to go through with the ordeal. The rituals around the ordeals gave them plenty of cover to ensure the water wasn’t boiling, or the iron wasn’t burning, and so on. If rigging failed, a priest could interpret the ordeal’s outcome to exculpate the defendant nonetheless (“His arm is healing well!”).