"In practice, there were apparently edge cases (students with ambiguous evidence..."
These where exactly the cases that I focused on. I honestly thought I was playing in a "Numb3rs" episode. Constructing models for detecting the probability that an individual student would structure an Excel spreadsheet in a specific way, measuring likelihood of sentences using Google N-grams, and other things that are not the obvious things that the honor council would immediately understand. At the end, the students were admitting guilt but at a great personal cost for me.
When nobody else is fighting like that what is the incentive for me to continue? Not all students go through my classes, so that all have equal chance of being caught. In a (very) perverse way, it is a matter of fairness to the students for me to be as vigilant as everyone else.