I view that is a negative as that a convention of the game rather reflecting the reality of the setting. Similarly I am not keen on how mechanics are activated like Second Wind, or the dice pools that accompanies the 5e Battlemaster variant. Both only make sense as part of a game not as a reflection of the reality of the setting.
To be clear the reality of the setting can something fantastic like a RPG like Toon which is about roleplaying characters in a cartoon world. It not about being realistic in terms of how our world works.
Nor reflecting the reality of the setting has to be detail in the way that GURPS with all the combat option is detailed. It can be highly abstract as long it can tied back to how the setting work as if you were there as the character.
So I view mechanics like "once is all you get" as a game convention.
But you don't necessarily have to. In the case of the AW stricture, it is most likely to be a simple design construct. That is, a check like that can only be taken once because it represents the total effect of the effort of the character over the entire scene in that direction. They have done their utmost, and there is simply no more that they can do. It is like saying in D&D that the thief cannot keep rolling 'pick locks' endlessly on the same lock. That check means "you fiddled with it until you reached an end state in any attempt to pick it, there's nothing more you can do."
It might also be possible to cast some other things in that light. It is probably hard to do so with something like Superiority Dice in 5e, or encounter power slots in 4e, but what about spending an Action Point in 4e, or 5e BM's equivalent? While, in a crude mechanical sense, it is depicted as an extra action, I think it is quite possible to view it as "I am expending effort at the maximum rate, which is faster than normal." You cannot do it twice because the expending, the going faster, isn't something that simply happens all at once, it is a state of going faster which results in approximately one extra action. Remember, in games with a lot of mechanical structure like this, things like actions, attacks, hits, turns, etc. all have a somewhat abstract relationship to the fictional world (unless you hold that the world actually advances time in discrete 6 second steps as soon as someone swings a sword).
This was always my argument with 4e's A/E/D/U, plus APs and whatnot. who's to say how this looks in a narrative? It doesn't actually play out FICTIONALLY anything like there being certain discrete moves and blows made, and 'extra actions' taken, etc. If you composed it into such a narrative (during play or after the fact) and then examined it, you wouldn't be able to tell for sure what resources each PC expended. Nor for that matter how many hit points anyone lost, etc. Some of those might be a little tighter than others, but even HP are pretty suspect, given how they cannot logically possibly represent any exact physical measure of damage.
Of course, none of this is to say there are no meta-game resources or mechanics. I think its clear that, considered mechanically, MUCH of the game is of this character, at least partly. Some things pretty much entirely, like PV's Story Certificates, which do seem fairly 'meta' in that they come into existence for 'table reasons', and don't represent anything fictional when used. So they exist purely in mechanical terms (this may well make my definition incompatible with
@pemerton's, though I don't have a big problem with his either).
Events that potential or certain negative consequences for the character that they zero control over. In my experience it doesn't end will over the long haul if that handled through fiat. Players are far more accepting of the results if it occurred because of random generation. And they know that these tables are being used as part of the campaign. So they factor the risk into their plan.
Players are more aware than one would think that the referee just happened to create a forest in front of them to adventure in. It can be gotten away with is done sparely but done over and over it become a noticeable pattern. It doesn't mean it doesn't work for how you run your campaigns. But it does take away from running a sandbox campaign.
Why? Because it takes away from the challenge knowing the referee is creating something out of whole cloth right then and there.
As opposed to creating said thing last Tuesday? I mean, its a fantasy world, so nobody can say what "should" or "should not" appear in a given location, and hence no player can say "this would have existed along any path we took." So, you are really concerned about RAILROADING, as I see it. Now, in
@pemerton's case the players had chosen to "go to the Holy Land" (I think) and they didn't pick the specific route, so there's actually no question in this case of railroading, no player choice was negated! You could argue that the overall journey is partaking of more the character of a linear adventure with one path, vs a sandbox character, and I would agree with that. Still, I am sure that Pemerton's players could say "we go around the forest", although that might have entailed some sort of other cost.
Now this doesn't mean you have to make 1,000s of pages of notes. But it helps if it already on the map, and you have a sentence or two about it, even though you have to take a breather to create something or pull something off the shelf in order to supply details if the players choose to explore it.
This is based on my observation of doing this for decades with multiple groups of players. I first noticed this when I switched from using the World of Greyhawk to Judges Guild Wilderlands in the early 80s. The players considered what happened to be more fair knowing that many details were there ahead of time. That I wasn't just making naughty word up to spite them.
Keep in mind player can and do make a bad plans. Underestimate the opposition or overestimate what they can do. And suffer negative consequences for it. In short in my campaign there is the possibility of failure. But if you are going to have possibility of failure then you need to be a fair referee. And it more fair to have a certain level details already defined about the setting. In practice it doesn't have to be much.
Yes, though all of this is founded on the core notions of Gygaxian skilled play. That there is 'success' and 'failure', and a fair referee doesn't stack the deck against the players, because they should have a 'fair chance to win'. Granting that 'victory' has a lot of gradations and is a very incremental thing that is probably only rarely arrived at fully, nor are its conditions entirely spelled out (IE get a million gold, did you win the game, or do you have to be 20th level first?). Even when ultimate victory is rarely considered at all, this ethos still prevails.
I think, in my games, there isn't a contest in which anything that can be accomplished within the fiction really equates with 'victory'. Characters could achieve their goals more or less completely, but those are just things which drive the characters, not the GAME or the PLAYERS. If the players 'find out what happens' and a 'story with dramatic pacing' is perceived in play, then they succeeded (at least if it was fun). So there really isn't a sense in which placing obstacles is 'fair' or 'unfair'. The question is "did the existence of this obstacle lead to suitable and enjoyable play?" If it did, then who would criticize it? Of course if the game botched things, then boo GM!
Because the bias is minimized as a result. So the result is perceived as more fair. Provided of course the random table itself is perceived as fair. If you say on a 1 you met a goat, 2 to 6 you met Smaug the Golden. Well players will call out you out for using a dumb ass table. Unless of course is happens to be one for around Erebor. Then it fits what been said about the locale. But if a referee uses this for the Shire well they deserve the player's scorn.
Yeah, but you do run the risk of things just becoming uninteresting. If the players are hankering for social intrigue and big-city action, and all the table coughs up are airless rocks and TL2 worlds with no spaceport and a law level of A, they are probably going to think that's boring. Anyway, I don't think it is a problem at the scale of a sector/sub-sector map because it is such a vast region, SOMETHING interesting can be scared up. But that again speaks to how much Traveler leans on "the ref can add stuff that makes things interesting" (though to be honest, TAS, patron tables, random encounters, bureaucrats, etc. goes a long way, space is rarely a snooze for long).
Anyway, as I said above, if the goal is interesting story, and the story tracks the desired themes and interests of the people in the game, then this 'bias' simply isn't a meaningful measure, in my style of play. The GM can be a total rat bastard and just lay it on thick and then dish a whole nuther load on top of that, if it results in good enjoyable play!