Jerome Frank, writing in the 1930s and 1940s, was of the opinion that legal rules are essentially fictional. He was not a postmodernist; his scepticism came from a solid grounding in reality and was based on the premise that no court ever makes a decision based on the "real" facts - the decision often happens months or years after the event, it often relies on suspect eyewitness testimony, and judges and juries are fallible and biased.
Rather than being based on a clear understanding of the facts, then, court decisions are instead merely a manifestation of what individual judges consider to be just and proper in that particular instance, taking into account the impact on society, with the judge paying lip service to the rules after the event. The decision-making process, in fact, is the reverse of what a textbook will tell you, according to Frank. That is, in almost all cases, the judge first makes a decision based on his own idea of what he thinks would be just, and then he searches for a rule to fit it to afterwards.
This was not a problem for Frank, and indeed it was something he thought a mature legal system should embrace. He said:
We want judges who, thus viewing and employing all rules as fictions, will appreciate that, as rules are fictions 'intended for the sake of justice', it is not to be endured that they shall work injustice in any particular case...
In other words, Frank was of the belief that in a mature legal system, the fictional veneer of predictability and robustness that rules provide is valuable, but actually all that really matters is that judges make the just decision in any given case. Applying a rule "correctly" ought never to result in an injustice. Judges, basically, ought to be free to make any decision they deem proper, unconstrained by the need for "legal clarity" or undue respect for rules and principles that have no real basis in reality.
Something similar can be said, in my view, about the role of the GM in a traditional game. I'll use Monte Cook's example as my own:
Say you've got a special weapon (magic or tech, doesn't matter) that makes foes all itchy so they are distracted. In a tightly written ruleset, the designer defines not only the effect, but what (if anything) can counter the effect. Maybe it would state that if the victim suffers a -3 penalty on all actions for the next ten turns unless he spends two consecutive turns and makes a new successful resistance roll, in which case the effect ends. In a GM-logic ruleset, you would write it entirely differently. You would explain what's going on in the situation, and let the GM handle the rest. So it would say that the victim is covered in an itchy and irritating powder and suffers a -3 penalty for the next ten turns while the powder was on his flesh and clothes. Then it would be up to the player to say, "I want to clean the itching powder off." And the GM says, "okay, if you take two turns, and make a new resistance roll, you can get it off." The advantage in the latter case is, the player could also say, "I'm going to jump in this nearby pool of water," and the GM is free to say, "Okay, that washes it off immediately." Or the player might say, "I use the water in my canteen to wash it off," and the GM might say, "Okay, that still takes two turns, but the roll is automatic now." Or the victim's mage friend might conjure a wind to blow the powder off. Or whatever.
The "tightly written ruleset" is the kind of approach that Jerome Frank argued was meaningless and self-defeating: maybe in a very limited set of circumstances it is possible to create rules which will always apply in the same way in all situations, but those rules are rare and in the majority of cases the rule will not be directly relatable to the "facts", which do not really exist in any event. It is better to recognise that, generally speaking, in most situations, the GM will make a decision based on what he or she thinks is best or most appropriate, and may then retro-fit a rule to that decision afterwards. The process for making the itching-powder decision would go:
GM thinks about what would happen if the player tried what he was trying -> GM makes a ruling -> a fictional rule is stated. (Fictional in the sense that no rule is really "real".)
The next time itching powder is used, the same process is followed. It may be that the GM makes the same ruling as before, but he shouldn't feel bound to, because the circumstances may be a little different (this time, it's raining, so the powder can be cleaned of automatically in just one turn of action, or something). He should just decide based on what he thinks is appropriate.
Frank would say we should embrace this if we want to have a mature approach to role playing. Unfortunately, not everybody does.