To me that is still metagame - for instance, why is it a human cleric of Pelor rather than a dwarven assassin who is in the right place at the right time? Not because of any working out of internal world-logic, but because a player wanted to play one sort of character rather than another!
More generally, I don't really feel the aesthetic force of quarantining this sort of framing to setting up the game. If it's OK to have a dramatic thing happen to these people here and now because that's the premise of the game, I don't see why it's not OK for another dramatic thing to happen to them down the track.
It's not as if the causal forces in the world, as modelled through the dice and other mechanical systems, make it
impossible that subsequent dramatic things should occur to them!
In my experience, it is very hard to make much headway in 2nd ed AD&D without the GM having an overwhelming influence on the shape of events.
For instance, will or won't the PCs bump into any given NPC? There is no device for the
players to instigate such an encounter. The random encounter tables to which the GM has access tend to indicate only hostile or potentially hostile NPCs (eg they tend not to include labourers, peddlers, admirers, etc, and all the other ordinary or not potentially hostile people the PCs might meet). And even if the GM confines him-/herself to these tables, and uses random reaction rolls, these don't dictate where NPCs come from, why they might be friendly to or hostile to the PCs, etc.
But if a PC is in prison, whether or not any given NPC comes to visit him/her, or tries to free him/her, or tries to sneak in to assassinate him/her, is all pretty crucial stuff (I've just been re-reading REH's "The Scarlet Citadel" and "The Hour of the Dragon", in which visits by NPCs, freeing by NPCs, and attempts to assassinate a prisoner by an NPC, all factor into Conan's ability to escape from prison). So the GM's decisions about this, whether made up out of whole cloth or constructed out of the random tables, will have a huge influence on the shape of the events.
Other examples could easily be given.
To my mind, this isn't particularly about the GM being
neutral. Neutrality doesn't come into it, really - is it more neutral to not send an NPC to try and rescue the PC? Or less neutral? - nearly everyone has friends and admirers, after all!
It's about a lack of tools and techniques to avoid the GM having to do this heavy lifting. Classic D&D didn't have them because it wasn't really designed with those sorts of scenarios in mind, and 2nd ed AD&D didn't really add any significant new tools or techniques to the classic chassis that it inherited.
I can easily imagine this. Think about whether or not your PC has socks, or a handkerchief, or a lock of hair from his/her childhood sweetheart. I think in most games I've played PC equipment lists have not been specified to such a level of detail. Now all you have to do is generalise that to other cases - like the PC has "Adventuring gear" lists on his/her sheet, but we don't know whether or not it has rope in it. (Because the player didn't write it down, or because no one can remember whether or not all the rope got used up in an adventure we played a year ago, or . . .)
In my current campaign, I'm sure at least one PC sheet is in exactly this state.
But in any event, ignorance about roped-ness wasn't what I had in mind. Suppose we know the PC doesn't have rope. So the player declares an alternative way around the obstacle (say, an attempt to simply climb the wall Conan-style). If the GM then says no, or the player rolls a check and the GM declares it a failure, should the player keep trying to think of solutions (eg pull out the potion of levitation that was being saved for a rainy day)? Or should the player conclude that the GM is blocking any and all attempts to circumvent the obstacle, and the player is simply going to have to choose some other path for his/her PC?
In an illusionist game, including one in which the GM is sometimes blocking and sometimes not, the player can't tell. The GM could reveal, but then (i) we're out of illusionism into something else (participationism, or perhaps a trainwreck if the players arc up at the revelation), and (ii) for many players, immersion would be spoiled because the metagame has suddenly been made overt.
4e bites the bullet on (ii), by making the metagame overt: the player has to negotiate with the GM for the feasibility of what is being attempted before any check can be declared or resolved. But it thereby avoids the covert blocking issue: once a check is declared, there is a structure to establish DCs, to determine how close the situation comes to resolution (N successes before 3 failures), etc.
This is set out in the skill challenge rules that I quoted upthread: the GM first works with the player to frame the desired check within the fiction (operating under the guiding principle of "say yes as much as possible"); then the check is declared, made and resolved, with the GM narrating the consequences that flow from the framing.
If the issue is one of word-meaning, it's no skin of my nose how you want to use the word "power" or the phrase "player empowerment".
I think that, in comparison to the way that "trad" framing and skill resolution works, the players have more power because (i) there is an overt, unconcealed negotiation phase where they are entitled to make a pitch for what is feasible in the fiction, and (ii) the GM is under instructions to say yes when possible rather than block, and (iii) there are mechanical systems that tell both players and GM what to do if the GM says yes (eg no "insta-win" buttons, but instead a system of accumulating failures and successes determined within a framework of "subjective" DCs that are at least notionally designed with the maths of PC build in mind).
I think the way that (i) and (ii) confer power on the players is fairly self-evident once stated. (If you don't think someone being entitled to negotiate with someone else who is under instructions to say yes when possible empowers the first person, OK, but then I think the issue is one about word use or about the deep metaphysics of social interactions, and not about game design.)
The way that (iii) confers power on the players is, in my view, two-fold. First, it prevents GM blocking, calling for endless re-tries, etc. All the stuff that @
Aenghus and I have talked about, which for me is part and parcel of 2nd ed-style D&D.
Second, and a bit more indirectly, it supports the GM in saying yes because it ensure that saying yes has predictable, workable consequences rather than leading to game-breaking outcomes. It might seem paradoxical that the removal of insta-win buttons empowers rather than disempowers the players, but I think that it does. Insta-win buttons create the illusion of power, because of their nature within the fiction, but the fiction is not self-executing - the need to mediate insta-win buttons via GM adjudication means that, at least in my experience, they are not a source of player power but rather a cause of illusionist or adversarial GMing (with the traditional treatment of the Wish spell being just one obvious paradigm).