Trial by ordeal—it evokes images of a darker time, where social outcasts drowned amidst accusations of “Witch!” or carried hot coals to clear their names. But this week’s Law and Economics brings a novel twist to an old subject with Professor Peter Leeson’s paper, Ordeals, available to read at http://www.law.uchicago.edu/files/files/Leeson.pdf.
What It’s About
To a modern secularized society, trial by ordeal may seem particularly insensible as a method of adjudication. It forces the innocent and the guilty alike to undergo the cost of an ordeal and, short of actual divine intervention, the outcomes won’t have much correlation with actual guilt or innocence. However, Professor Leeson takes the unique position that, for a highly religious or superstitious society, an ordeal system may produce more accurate results in adjudication by leveraging people’s strongly held beliefs to elicit confessions and improve factfinding. In other words, ordeals may have simply been the best option for a society where more sophisticated methods of evidence-gathering do not exist.
Here’s how it works. Suppose priests could manipulate the results of trials by ordeal. Professor Leeson notes that priests had discretion to lower the temperature of the hot iron as they prepared the ordeal, to pray a longer invocation to allow boiling water to cool, to douse an innocent suspect with cold holy water, or to decide arbitrarily when the accused had sunk far enough in the water to prove his innocence in a water-dunking ordeal. Suppose, also, that priests recognized themselves as “agents” of God and exercised this discretion to sort out people they believed were innocent from those they believed were guilty.
If belief that God will intervene in an ordeal is strong in the community, innocent people would always choose to undergo an ordeal, believing that God will protect them from the heat of the hot ordeal or from floating in the water-dunking ordeal. Guilty people would always choose to confess or to settle rather than undergo the ordeal, be exposed by God, and suffer a larger punishment after experiencing the pain of the ordeal itself. Priests would adjust ordeal outcomes to make most ordeal results show innocence, and they would have elicited valuable information from the defendants about their actual innocence. However, the community’s belief in the ordeal system is not likely to be perfect. If everyone who undergoes an ordeal is declared innocent, this might induce some guilty skeptics to try their luck with the ordeal. So, the priests would have to maintain some non-zero rate of guilty results.
Professor Leeson’s paper presents a mathematical model suggesting that priests could select an optimal “guilty-result rate” that raises the cost of undergoing an ordeal to just a high enough level to cause guilty defendants and innocent defendants to act differently. Although rigorous models are available in the paper, the basic idea is this: Medieval defendants, in reality, probably had neither perfect belief nor perfect skepticism in the ordeal system. To the extent that they are skeptical about the system, the guilty and the innocent can anticipate the same result. They face the same random chance of being declared guilty under the guilty-result rate, so the cost of undergoing the ordeal is equal for both types of defendants. So, the guilty and the innocent perceive a difference in cost only to the extent that they believe that God might intervene. We can imagine a world where the alternative to ordeal (settlement, confession) costs some amount C and defendants believe in some P % chance that God will intervene, so that the cost of ordeal becomes too high for the guilty (because they factor in the chance that God will get ‘em), but the cost of ordeal becomes acceptable for the innocent (because they factor in the chance that God will stand up for them). The guilty will confess or settle, while the innocent will still go through with the ordeal. The stronger the society’s belief in God’s intervention, the lower the guilty-result rate the priests have to maintain to cause the guilty and the innocent to act differently. As long as trial by ordeal is only used as a “tiebreaker” when the evidence is balanced on both sides (or completely absent), it may be the best available method of adjudication for a religious society—certainly, its information-inducing effect produces more accurate results than the alternative of doing a coin-flip.
What Was Discussed
Professor Leeson started the discussion by summarizing the paper and analogizing the lessons learned from medieval superstition to the modern belief in jury factfinding and lie detector tests. Sincerely held beliefs, whether true or not, can have actual effects on the truth-finding ability of a system by affecting the way believers disclose private knowledge. As he paraphrased, at one point, how well a legal system works depends on how well people think it works.
Immediately, one participant noted the “irony” of a system that requires laity to be strong believers and priests to be non-believers in order to work properly. How could a priest that actually believed in God’s role in the ordeal dare to intervene enough to maintain the necessary guilty-result rate? That would mean that, every once in a while, the priest would have to declare that a person who sank X feet in the water-dunking was guilty, knowing full well that he had earlier declared some other person who sank the same number of feet innocent. Professor Leeson maintained that a sophisticated priest could appreciate that God works through the priest as an “agent” during the ordeal, while keeping consistent with his belief that God endorses the ordeal. A few participants still remained unconvinced on the religious state of mind of the priests, asking whether there was any direct historical evidence that priests actually deviated from the standard ordeal procedures. For example, on participant suggested that Professor Leeson could look at whether there was any correlation between guilty-result rates and decline in the belief/use of the ordeal. Professor Leeson admitted that there is very little historical evidence, just because records of ordeals from the medieval period are sparse and non-systematic, but offered that the small proportion of guilty results, especially given the clearly harmful acts prescribed in the heat-related ordeals, are in itself evidence suggesting that the procedures must not have been followed strictly. Skepticism about the priest’s knowing complicity in the sham ordeal resurfaced several times during the course of the workshop. Several participants expressed wonder at how a priestly class could maintain the secret of the ordeal without the laity figuring it out, to which Professor Leeson suggested that ordeals may have been used more often with the lower class for that very reason. One particularly strange piece of historical evidence is that about 63 percent of ordeals surveyed seemed to exonerate the defendant. Given the harmful activities (being boiled, carrying hot metals) and subjective evaluation required in the ordeal (declaring whether a blister is clean or unclean, assessing whether a defendant has sunk into the water enough), it seems surprising that such a high number of people were exonerated unless the priests had some intent to reach such an artificially high result.
Another participant asked Professor Leeson to elaborate on why a method as costly as an ordeal was necessary, when a spiritual oath might be as effective in a world where religious beliefs are strong. Professor Leeson explained that the scarier the punishment gets (the higher the cost of ordeal relative to the cost of confession/settlement), the lower the guilty rate needs to be to elicit a different response between guilty and innocent defendants. The idea came up that the cost of the ordeal should be the same for anyone who faces the ordeal, whether guilty or innocent, since the same pain awaits them. However, Professor Leeson clarified that perceived costs may be different between the guilty and the innocent, to the extent that the innocent believe a certain probability of being spared by God and the guilty perceive a certain added probability of being found guilty by God. (The implication to the original question posed seems to be that the difference in perceived probability is multiplied by the cost of the punishment, so that a painful ordeal can be much more effective at magnifying the difference in cost between guilty and innocent people than an oath.)
Participants next expressed the concern that, in practice, differentiated behavior between the innocent and the guilty might not happen, as evidenced by modern plea bargaining statistics. (For example, risk-averse defendants may avoid ordeals even when innocent; their faith may fail them as the boiling cauldron steams up or they may want to spare their families of unwanted controversy.) Another participant added that trials and ordeals might not even be for factfinding purposes—perhaps they are just employed to “legitimate” a result and build community consensus. Professor Leeson acknowledged that this is a different historical use of ordeals; however, he distinguished consensus-building ordeals, such as witch-hunting ordeals of the 16th century, from the factfinding ordeals of the medieval period by noting that consensus-building ordeals have a high guilty-result rate, while factfinding ordeals require a low guilty-result rate to work. As mentioned in his paper, high guilty rates increase the chance that a “God-ordained” ordeal result may be disproved by later exonerating evidence as more random innocents are found guilty under the high rate, eventually eroding the faith that is necessary to maintain the system after disbelief reaches a certain threshold. This observation led Professor Leeson to hypothesize at one point that superstition-based systems such as ordeals may be inherently unstable after it reaches a certain threshold of disbelief, since inaccurate results from the superstition lead to disbelief, which raises the guilty-result rate required to differentiate innocent and guilty defendants, which in turn leads to a higher probability that inaccuracies are found out (and the process repeats). After belief drops to a certain threshold, there will be a vicious cycle where most people don’t believe in the ordeal and the ordeal no longer becomes viable as a factfinding tool.
Although several other topics were discussed, the workshop came to a close with many questions still remaining, and the moderator closed with a parting reminder of two classic examples of ordeals from the Bible—the trial of Isaac, in which God induces Abraham to reveal his loyalty by nearly sacrificing his son, and Solomon’s proposal to slice a disputed baby, in which the King induces the true mother to reveal her identity.
Any way you slice it, an interesting discussion! I’m afraid I have to agree with many of the participants in feeling somewhat unconvinced about whether the ordeal was consciously used by priests as an information-inducing tool. Professor Leeson argues that most priests who administered the ordeal could reconcile a genuine belief in God-sanctioned ordeals with the fact that they are manipulating results using a kind of “agency” rationale. However, even then, it seems difficult to believe that priests would have tolerated playing dice with individual cases of guilt and innocence (even if it were for the noble purpose of obtaining better results in the aggregate), while maintaining belief in an omniscient God concerned with and capable of individual judgment. To draw a modern legal analogy, judges may feel that they are “agents” of legislative will, but they recognize that they have more discretion in applying standards than in applying rules. Similarly, a priest may believe that God will use him as an agent to transubstantiate the communion wine and bread, but that does not necessarily mean that a priest will feel empowered to the point that he could promulgate new doctrine, or even deviate from an established liturgical procedure. It is one thing for a priest to think he is God’s agent in administering the ordeal; it is another to have a priest self-consciously deviate from a set of precise procedures handed down from the higher-ups.
On a parting note, I end by questioning whether accurate factfinding is really the end goal that ordeals, whether in the form of water-dunking or modern jury trials. We live in a society where it is generally agreed that letting ten guilty defendants go free is better than letting one innocent be condemned. How should this affect our attitude toward the random guilty-rate method of truthseeking? And how easy is it, really, to draw the line between the arguably socially useful kind of ordeal that elicits hidden information and the bad kind of ordeal that stops inquiry, builds social consensus, and condemns? To what extent, for example, did medieval society’s invocation of God’s good name disincentivize further investigation? In the paper, it seemed as though ordeals were envisioned as a kind of last-resort tiebreaker when evidence doesn’t exist or is ambiguous. But if the public believes the local authorities when they claim they could just ask an all-knowing, fair judge for all of the answers, why would they bother with investigation at all—even for the easy cases?