Yep. When people say the GM has power over the game, they mean de jure power given them by the game's rules and instructions. Everyone has agreed, that this power is mediated by de facto power of the social contract among the participants.
I don't think anyone disagrees with what is actually happening, but we again have this pointless semantic quagmire that
@pemerton seems to love. There are productive things regarding this we could discuss, like what that social contract could and should entail, but the exact words and phrases used to describe the situation is not among them.
I don't know what you mean by "social contract" here. The only usage I'm familiar with in the context of RPGs comes from The Forge, where it means (roughly) the agreement to use a particular set of constraints, rules, processes and expectations in establishing the content of the shared fiction.
The same people who coined that terminology endorse the "Lumpley Principle", which asserts that
social contract is prior to system - or in other words, that formal rules that proclaim that one party has authority can't actually do any work, if there is no agreement among participants. Which is exactly what I have been saying, and is why rules that purport to give one party absolute power in respect of the shared fiction are pointless.
And this is not just semantics - presenting it as semantics is, in effect, an attempt to put forward a particular type of play (namely, GM-driven railroading) as if it were normative or even exclusive. As soon as we have non-railroading play - that is, play where the players can actually make a meaningful difference to what happens next - then it has to be the case that the GM does not have unlimited power over the shared fiction; because it has to be the case that the
players can do things or produce results that bind the GM. (Because if they can't, then only the GM can make a meaningful difference to what happens next, and we're back on the railroad.)
I didn't learn RPGing in the 2nd ed and 3E eras. I learned RPGing primarily from Moldvay Basic. Secondary sources, for me, were Classic Traveller and Gygax's AD&D books; and also early White Dwarf essays, especially from Lewis Pulsipher and Roger Musson. None of these deny that the GM is bound by rules and expectations, or assert that the GM has unlimited power to make up whatever they want whenever they want, so that all the players can do is prompt the GM to make decisions about what to include or not to include in the shared fiction.
These differences between GM-driven and other approaches aren't just semantic. I've encountered them, repeatedly, both in actual play experiences and in conversations about them.
My view is that GM-driven play is inherently fragile in a way that other approaches are not, because
everything depends upon the GM having worthwhile stuff to say, and there is no turning to the players to produce their own worthwhile stuff. I also think that that fragility can be reduced by GMs recognising that their stuff will only become part of the shared fiction if the players accept it, and therefore taking note of what does or doesn't engage or enthuse their players, and factoring that into their decision-making about what sort of stuff to produce. A futile insistence that the GM's power to make up stuff is unlimited is not only empty - as the Lumpley Principle tells us - but distracts GM's attention from what they actually need to make a GM-driven game go, which is not
an insistence on their power but rather
ongoing success in engaging their players.
I read repeated stories on these boards about failed or unhappy play - a recent one was in this thread, with a GM poster lamenting the inanity of one of the players in his game having his PC Firebolt a random bird in the woods. There is a fruitful pathway to better play experiences for that poster, but insisting on the GM's absolute power over the fiction is not part of that pathway. Acknowledgement of the Lumpley Principle, and then the next step of thinking
How can I create fiction that my players will experience as worthwhile, so that they don't feel obliged to entertain themselves by having their PCs Firebolt random birds, is part of that pathway.
I mean, even going back to the story I told about the Kobold interrogation that brought a game to an end. Insisting that that GM had absolute power, and that we as players were all obliged to go along with the GM's conception, contributes nothing to understanding (i) what happened , and (ii) how it might have been avoided, such that everyone at that table could have had a better time. Whereas recognising that shared fiction is, by definition,
shared; and hence that uptake by the players
matters; and hence that the real issue is not any supposed absolute power, but rather
how to come up with stuff that is worthwhile, engaging and even (heavens forbid!) compelling - that actually helps identify a pathway to a solution. The first step on that path being recognising when the players are telling you that your stuff is terrible.
