In (2) the GM can enjoy embodying the simulation, i.e. instantiating models in their mind and notes, and having ideas or rubrics for causality.
GM can experience the simulation unfolding. I have certainly experienced that when I have been GM, and I have heard others attest to it.
Vibrant imagination - a feeling of
what else would it be or
of course, THAT's what follows - doesn't make it a simulation or a model. That's still authorship.
In this thread I've given concrete examples: Gygax's advice to GMs, in his DMG, on how to create "living" dungeons. Those are not accounts of how to instantiate models. They're advice on how to author certain things. In each of those scenarios, there are many other ways things could unfold. For the players, the GM's decision "reveals" how the internal causes unfolded. But what is the GM's rubric? Referring to "rubrrics for causality" gets us nowhere in my view. Because - as
@AbdulAlhazred likes to point out, and as I've just pointed out in relation to Gygax's examples - from a given state as described in a RPG, just about anything can happen!
Here are examples of the sorts of principles that occur to me, but given that I'm not the one advocating for this play approach, it seems to me that others ought to be able to do a better job:
* In your extrapolation, don't have regard to off-screen causal factors that are not known to the players, unless the players are culpable for their ignorance of them;
* In your extrapolation, don't rely on magical phenomena that a skilled player of the game can't be expected to have regard to;
* In your extrapolation, don't rely on mapping mechanical edge-cases to the fiction (eg don't rely on the fact that falls can't be fatal to high level D&D NPCs, nor that single crossbow shots can't kill them, nor on the difference in most versions of D&D between dodging a giant's boulder and balancing on a narrow beam);
* In your extrapolation, don't rely on physical laws that are put into obvious question by the fantasies in the game (like FTL starships, flying dragons, etc).
That last one is of course participant-relative (what's obvious to a group of engineering PhD students may not be obvious to a group of 12 year olds).