I've seen little about what sorts of rules do or don't work, but Temple Grandin's _Animals in Translation_ (generally recommended for anyone who's interested in animals, emotions, mind-body interaction, or cool facts) has a very interesting section.
Temple Grandin is a high-functioning autistic who specializes in designing humane slaughterhouses. She invented an animal welfare audit for the U.S. Department of Agriculture based on the idea of finding a few key measurable facts which imply a lot about whether a facility is well-run. For example, if animals are limping, it might be a problem with their food, genetics, flooring, health, or treatment. If you enforce a limit on the amount of limping, then you don't have to have rules for all the factors that affect limping--you can just tell the farm owner to get the proportion of limping animals down.
She says that rules need to be few and clear enough to be comprehensible, to measure outputs (not inputs or paperwork), and to focus on big problems. The rules need ongoing enforcement, but a well-designed rule set makes enforcement relatively easy.
In her opionion, the problem is that verbally oriented people tend to make rules which are vague, unmanagably numerous, and not focused on relevent outcomes--she gives herself as an example of a visually-oriented person who's come up with an excellent rule system. I'm not sure that the real divide is verbal vs. visual. I suspect it's caring vs. not caring, though bad rules tend to make people not care.
Dog and cat breed standards might be an example of rules made by visual people which don't work out well, and which fail to be relevent to anything important.