Fair enough, and I don't think there's anything wrong with asking the question "what might happen here?" or "what might be the consequences of this action?" I mean, at some level those are questions that HAVE to be asked. Again, there's a need for coherency, so at least some sort of plausibility is implicit in that.
I do rebel against this notion, which is pretty commonly brought up here, that somehow GM's are "just figuring out what would really happen" or that they "must simply follow the consequences" with the idea there is even any way to determine that which is not 99% simply what they want to see happen next (for whatever reason, fun presumably). I'd say that this stance is not based on 'physics' (of any kind particularly) either. It seems to be based on, actually I don't know what. It claims literally that there is some 'natural progression' of a given 'world state' that can be 'worked out', at least in some degree.
Honestly, I'm not sure where you fall in this, and I'm not that interested in lambasting or jousting with anyone over these sorts of things. We all have various notions, and if we have fun playing based on them who really cares, right? I have just felt a bunch of heat for pointing out this "but he's not wearing any clothes!" thing (in a few threads, this one, and the parent thread probably, maybe a couple others over the past few years).
I do however use random dice rolls when determining what the players encounter while exploring. But I always give myself the freedom to ignore the outcome, if it doesn't seem fitting.
I also think the method can matter. Some other posters seem to disagree, though.
Both these posts seem to asssume that there are only two possible resolution systems for determining if the PCs find sect members at the teahouse: the GM decides based on his/her beliefs about the gameworld, or the GM "says 'yes'".
That is, they seem to assume that play will be driven simply by GM decision-making.
I find that to be an odd assumption to make, but unsurprisingly I agree that running a game that way will tend to make for a mediocre play experience.
(One reason I find it an odd assumption: the first RPG system I know of that explicitly deals with the issue of trying to find certain sorts of people in urban situations is Traveller (1977), and it assumes that the outcome of such attempts will be affected by rolls that are affected by skills like Admin, Streetwise and Leadership, with subsequent supplements adding further relevant skills like Carousing and Recruiting. It doesn't say anything about the referee just decding what happens.)
But others might argue our death mechanic is less realistic. That we don’t get to choose when we live or die and that it should be the dice that do so.
One quick comment on balance. This comment will be invoking 4e (because that is what has been invoked), but at its heart, its a design question (as an input) and the related product of play (output).
4e's balance often gets invoked as if its (a) some sort of retardent to dynamism and (b) some form of perpetuator of status quo.
That isn't correct for 4e (its actually not even in the realm of correct and the inverse is provably so) and its not correct as a product of design aimed at balance.
4e has balance on 4 different axes:
1) Broad intraparty balance at the site of the encounter.
2) Broad intraparty balance at the site of the adventuring day.
3) Expectant results of a 5 player party (which covers all the Roles) vs unremarkable deployment of a same level encounter budget.
4) Expectant results of an archetypal workday for a 5 player party (which covers all the Roles).
Because these 4 design aims are explicit and were achieved, there seems to be this designation by some (typically those that didn't play it very much or didn't play it at all), that such (achieved) design aims must yield a play culture that stays tightly within the boundaries of (3) and (4). Then, following from that, there is this assumption of my (a) and (b) above (lack of dynamism and boring, uninteresting status quo).
The problem with this is simple. The idea that (3) and (4) are actual play culture fundamentals is absolutely wrong. They are balance calibration features of design. Further, those two as play culture artifacts are completely at at tension as (4) assumes you aren't doing 3! Further still, both DMGs go on at length of how to perturb that balance calibration archetype and what the implications of such perturbance will be...therefore assuming you're going to be doing just that!
Further, further, further, still...
If each Role can be thought of as a different Magic the Gathering deck (and that is exactly how they should be thought of - I guarantee that was a design impetus if not THE design impetus; eg a "Monored Aggro Burn vs Jeskai Midrange"), then significantly varying the exact same encounter budget and using different battlefield qualities (terrain, obstacles, distances, terrain powers, Hazards/Traps) and different objectives (eg "Hold the Line vs Waves", "Escort/Protect the Minion from here to there", "Deal with Interference While Completing/Foiling the Ritual", "Defeat the Enemy Before X Rounds", etc) is going to change the dynamics of the combat significantly and introduce variables that will play to the strengths or to the weaknesses of different character builds and group builds!
Just because the baseline has been calibrated such that the GM can predict the outputs within a reasonable margin-of-error, doesn't mean that there ceases to be variables x, y, and z that can be perturbed to create significant dynamism at the encounter level and for the adventuring day. If anything, it emboldens GMs to perturb that x, y, and z because they can foresee the potential outputs of those inputs within a reasonable margin-of-error.
That is why GMs such as myself appreciate rigorous baseline calibration, especially if a system has a robust range of x, y, and z.
That isn't what I said, though. This is what I said, "However, if we start trying to play combat like real life where everyone can react in real time to what creatures are doing near them, the game would bog down to the point where combat is simply unplayable."
That involves a lot more than just having a chance to escape. Goblin one moves, then 19 other goblins and the PCs have a chance to react to what is happening. Then all the goblins and the PC start reacting to each other. Perhaps the goblins all start to rush. But what if 3 pull out crossbows? Goblins will react by getting out of the way. The PC will react by trying to get to cover or low to the ground. The crossbow goblins maybe aim lower, or maybe move to get better position. And on and on. That just can't be effectively modeled and even if you try, it will take huge amounts of real time to play out a combat like that.
Actually that’s roughly how I run combat. Of course it’s not real time in terms of each swing, but we don’t use initiative and don’t really use rounds either. We go with a general idea how long an action might take, and if there is time before that action is completed other things happen.
And they are free to change their actions in response to other actions. It’s not unplayable at all, and helps move things along and keep people involved. It’s not as tough or complicated as it sounds.
If you have a trust problem with GMs, then that sounds like a you issue to me. It may be safe to say there are some GMs out there who might choose what happens based on what they want to happen - but I also know there are a lot of GMs out there who take the idea that they should be impartial seriously. Frankly, I'm a little more suspicious of the "Say Yes or Roll" mentality than the "Say Yes or No when appropriate for the situation" mentality because I don't feel the former gives the setting/mysteries/NPCs an even break with the PCs.

(Dungeons & Dragons)
Rulebook featuring "high magic" options, including a host of new spells.