And now instead of the referee neutrally adjudicating the fiction, we've got the GM's adjudication establishing the fiction!
And that is the crux, isn’t it?
Neutrality of refereeing is rendered out of existence in such a scenario.
When you have codification of action resolution process, you’re building from the middle out (what does OBS4 mean? What Position and Effect be for this situation? Here are the spread of results for all moves). You don’t have to be peerless/unchallengeable when establishing obstacles. Here are the action resolution mechanics < move backward from them to framing obstacles/situation > move forward to action and consequence.
When you’re building and iterating the resolution method in real time for each obstacle/situation, you’re starting at the the beginning. What is phenomenon we’re modeling (obstacle/situation) > build resolution scheme that manages appropriately integrating all the parameters of the model > hopefully you’re a peerless expert and now we move on to action and consequence.
Of course if you’re not a peerless expert or the “high-trust” dynamic at the table is wobbly or outright compromised due to an abundance of (often high ego) experts in related fields, before action and consequence, you would have to append:
* change situation to be possessed of worse or better framing based on the potency of the dispute of your action resolution model (you lose neutrality of GMing here!).
* defer to received input or vehemence from table experts who are disputing your model (the “high trust” parameter is gone!).
The selection pressures that would lead to table dysfunction are significant. The functional table had to have been the very rare breed.