The Leviathan Blood Manifold (whatever this is) on the floor above the PCs suffers a containment leak.
A side-comment on this, and relevant to the issue of
prep: I think part of good framing is introducing story elements - like the
Leviathan Blood Manifold - that are evocative, make sense in light of genre and established fiction, and not fully defined. This gives the players something to riff on. Even something as simply as a wizard's tower (in a fantasy game) or an enemy base (in a sci-fi game) can be the starting point for interesting action declarations.
Tying everything down in advance will produce a very different play experience. It doesn't
have to negate player agency - everything in White Plume Mountain is pretty tied down, but the players can still manifest agency in making decisions about how to overcome the puzzles - but you're probably going to get more of a focus on engineering details and fewer Leviathan Blood Manifolds.
This also relates to
@Tonguez's comment upthread about player-introduced story elements. In my 4e game, much of the detail about how magic works and what can be achieved using it was introduced by the player of the invoker/wizard. But not as overt "player narrative control" based on spending fate points (or whatever); rather, he would declare actions (typically in skill challenges) and as part of that would set out his theory of how magic worked and what his PC was trying to do with it - and if the action succeeded, then his theory is validated!
I think of that as an important manifestation of player agency.
my thinking about a GM railroading themself goes along with a GM being the victim of their own Illusionism (which I mentioned elsewhere in this thread). If you as GM are in a position where you don't even have the starting position prepped, you might well end up coming to the same place no matter what the PCs do, because you are trapped in your own brain and in the moment; I'm thinking sort of when a GM decides what an "interesting result" would be.
What I'm getting at, here, is that to the extent GMing is like writing fiction, a GM who has gotten far enough past his prep might be unintentionally taking the story where he wants it to go regardless of what the PCs do, the same way a novelist who doesn't outline or sketch ahead of where he is writing would. I'm not saying it's inevitable--or even necessarily likely; I'm just saying it's possible.
I don't really follow this. A novelist who writes without outlining or sketching ahead
takes the story where s/he wants it to go. Presumably, though, a novelist who outlines or sketches ahead also
takes the story where s/he wants it to go. Why would you write an outline or sketch that isn't what you want?
But in any event GMing isn't all that much like writing fiction, because - assuming player agency is manifested at the table - the GM is not the sole or even the primary author. The players, in declaring actions for their PCs and making it clear what they hope those actions to intend, play a key role in establishing the parameters of the subsequently-narrated fiction.
Just thinking through the games I've run over the past few years, most had no starting position prepped:
*
Prince Valiant: the players created their PCs, we invented some backstory (two were father and son; the third had joined them on the road; all were on their way to a tournament); I looked through the books and chose a first episode; the game
currently has the PCs crusading in Cyprus at the head of the holy order they founded, which is not where I (or, I think, they) anticipated things would end up;
*
Classic Traveller: the players rolled up their PCs; I rolled a starting world; we invented some backstory to establish the nature of the world and explain why all the PCs were on it; I rolled a random patron on the patron encounter table; and we went from there into a bioweapons conspiracy;
currently the PCs are on the other side of a galactic rift (following a misjump) investigating a 2 billion year old alien pyramid embedded in ice;
*
Dark Sun 4e D&D: the players built their PCs; I told them we would be starting in Tyr after the revolution, and then asked them to write "kickers"; one of those kickers established that the revolution was in train at that very moment, and that it was taking place during a gladiatorial contest, and so this established the starting situation; this game has been on hiatus for quite a while now but it was heading in the direction of urban intrigue last time we played it;
*
Cortex+ Heroic Fantasy Hack: I distributed the pre-gen PCs; the players opted to interpret them in Viking rather than Japanese terms (I'd designed them to suit either); we then established a starting situation, of the players being sent north to investigate omens of trouble among the spirits and the gods; their first attempt failed when most of the PCs got lost in a dungeon except for one who robbed the dark elves of their gold and returned back south for a while; with the signs of the Ragnarok becoming more evident, they are currently in the middle of a second attempt;
*
Cortex+ Heroic LotR/MERP: the players chose their PCs from the pregens I'd prepared; we worked out why they were going to set out from Imladris; and then started playing (I'd prepared some material for a journey to Ost-in-Edhil, but the PCs headed north into Angmar because Gandalf's player explained that Gandalf had heard rumours of the recovery of a palantir in Forochel); the PCs ended up heading south in pursuit of the palantir (after they failed to stop Orcs from taking it), and currently they are in Moria after the dwarf PC led them back into the halls of his ancestors in order to escape the scrutiny of Crebain of Dunland;
I've also run Cthulhu Dark, Wuthering Heights and AD&D (the last using the Appendix A random dungeon generation process) with no starting position prepped. One Cthulhu Dark session ended up with the three PCs - a journalist, a longshoreman and a legal secretary - using a tugboat to crash a cargo ship onto rocks; in another one the PCs remained essentially strangers to one another but ended up helping thwart a colonial venture involving were-hyenas.
The only games I can think of in which there
was a starting position prepped is
The Dying Earth, where we used the example scenario in the book; and one session of Castle Amber (using AD&D mechanics).
I haven't experienced any connection between no-prep and "coming to the same place no matter what the PCs do". And I don't really see how that would even happen, for two reasons: first, if there is no prep then almost inevitably the players play a greater role in establishing the starting situation; and second, the players will declare actions for their PCs which produce consequences which in turn determine what "place" is arrived at.
I think I was more intending that the choices of the players setting things off on new branches that have not been predetermined by the GM. To use a metaphor, the PCs are blazing their own trail rather than following one of those set by the GM. As you point out, the system and the goals and methods of play will matter quite a bit in this regard.
This "blazing of their own trail" is, for me, the core of player agency in RPGing. It's why action declarations and their resolution are so fundamental. And it's why - if players are to exercise agency - there have to be limits on what can be determined via GM prep.
With regard to "system and methods of play", I'll reiterate an example I've mentioned in other threads: Classic Traveller generally is (in my view) a very robust system which reliably permits genuine and effective action resolution. That's not to say it's as tight as (say) Apocalypse World, but it's pretty good. With one exception: on-world exploration. There is no resolution framework for this beyond
the GM decides where the <whatever it is you're looking for> is and how long it will take you to find it. There are rules for chances of vehicular malfunction per day, and for encounters, which are fine in themselves, but nothing analogous to a skill challenge or similar framework for actually finding things onworld.
Since I discovered this the hard way (ie during play!), I've handled all onworld exploration simply via free narration. Eg
most recently, in order for the PCs to
find their pyramid embedded in ice, the checks required were actually social checks (to find a sect and then persuade it to hand over certain occult drawings which were actually time-series tectonic maps). But the actual journeying took place in a spaceship and was resolved via free narration: there are simply no mechanics to support making it, in itself, a focus of play.
If someone is mostly used to systems that don't have robust action resolution (except perhaps for small-group combat), then I see how they might posit that prep is
necessary as a curb on the GM, because otherwise the GM might come to the same place no matter what the PCs do. I don't know what systems
@prabe is familiar with beyond D&D; but the only version of D&D that has robust action resolution of the sort I'm talking about, across a wide range of possible action declarations, is 4e. Classic D&D is robust for dungeon exploration, and semi-robust for hex crawls, but not for much beyond that; and 5e seems pretty similar to classic (ie AD&D, B/X, etc), to me at least.
But if the action resolution rules are robust, prep is not needed to curb the GM and - if it is prep of outcomes, or of "branches" - starts to risk becoming a curb on player agency.