I think to ask that every decision is made with maximum information without any effort on the players' part to acquire that is an unreasonable ask, though. It seems there needs to be some kind of balance there for the setting fiction to actually make sense.
That is, however, partly an artifact of the fact that modern and post-modern settings are generally very high information than you'll get in most fantasy or post-apocalypse settings.
There are a billion ways the fiction can make sense. If 999 million of them involve the players making blind, or near-blind choices; or alternatively involve the players following GM breadcrumbs to try and ascertain what is really at stake in their choice; then I will look to one of the remaining million possibilities.
Otherwise, as I posted upthread, the game ceases to have the appealing features of a sandbox.
I actually just posted about this
in another thread; I'll repost here, setting out the contrast between two sandbox campaigns I ran, and what made one go bad and the other not:
The first campaign was long-running (1990-1997 inclusive) and sprawling. It started at 1st level and finished somewhere in the low-to-mid 20s. There were 10+ players over the course of the campaign, and 20+ PCs. Rolemaster is a very mechanically intensive system, and the campaign leaned heavily into that. The fiction - detail of backstory, details of play, note-keeping, etc - was intense too.
The mistakes I made in GMing this campaign resulted from not being able to manage all the backstory, and the climaxes. In retrospect I would diagnosis it as: RM doesn't have adequate formal processes for handling sprawling campaign events and stakes; and I hadn't developed adequate informal/ad hoc processes. The ones that I had developed worked OK for low-to-mid level play, but broke down as FRPGing hit the scope (mechanical and fictional) that opens up in higher level play.
Two examples:
*A massive NPC scry-teleport-fry raid on the PCs. It made sense in the fiction - the PCs had been fighting, on-and-off, with a powerful faction of wizards for a good chunk of the campaign. But my resolution of it was terrible: I statted up the rival wizards, worked out what they would be able to do, using their spells as rationally and ruthlessly as the players did theirs; and it was a massive hosing for the players (and their PCs). The mechanical framework - pure, unvarnished, hardcore simulationism - left no room for forgiveness. The whole thing left a sour taste in everyone's mouth.
*The end of the campaign. The PCs had travelled to another dimension to confront evil godlings. I had started full-time work and so was short of prep-time, but RM is not especially conducive to low-prep GMing. As the backstory became ever more convoluted, and my notes harder to follow (being scrawled down during play), and the stakes less and less clear to the players, it felt like what should have been a climax was becoming rudderless and ultimately directionless. One of the players - someone who is still one of my best friend - ended the game by detonating a massive fireball (or similar effect) so that it engulfed all the PCs as well as some of the enemy NPCs. This was his deliberate game-ending move, and it worked.
Lessons were learned. Initially they were mechanical: in our next RM campaign (which ran 1998-2008 inclusive) we excised a chunk of the scrying and teleportation magic, and also dropped a few other mechanical features (like power point multipliers) that favoured casters over non-spell-users.
But I also worked harder on the non-mechanical side. The campaign was just as sprawling, backstory heavy and intricate. But the fiction was established so as to make the prospects of overwhelming retaliation against the PCs less likely: it was largely cosmological rather than earthly factional conflict. The stakes, and how they related to the PCs, were kept clearer. This game came to a natural, and hugely satisfying (for me a least) end, around 27th level. The mistakes, of letting mechanics and fiction both get out of the group's control, had been avoided.