The physical roll of the die isn't important, just as the particulars of the die you are rolling are irrelevant. It's all just a random number generator to determine the minutiae of who is where at a particular time.
I don't understand how this is not important. If the dice say, for instance, that the vampire is
right here, right now, then it follows that the vampire - unless spontaneously created - must have
come from somewhere. The vampire has a past. And that past has to be
written. It cannot simply be inferred by ingame causal extrapolation from other known features of the gameworld.
The DM doesn't need to alter the history of the world to account for the Lich showing up, because the history of the region already accounts for everything that could show up on that chart,
The GM has to
author new material about the past of the gameworld. That is altering (by way of adding to) the authored details of the gameworld.
Here is an example. I can't remember what colour t-shirt I wore on Friday 8 days ago (but I know I wore a t-shirt, because I
do remember that I didn't have any important functions on which would trigger me to wear a collared shirt, and t-shirts and collared shirts exhaust my shirt collection). However, because I am a real person, who really exists now and who really existed 8 days ago, and because the clothes that I own and wear are likewise real, there is some fact of the matter about the colour of that t-shirt.
But now, let's transpose that scenario into a roleplaying game. One player asks another,"What colour are the trousers that your PC is wearing at the moment? And are they the same colour as the ones your PC was wearing a week ago?" From the point of view of the characters in the gameworld, there is a fact of the matter about those things every bit as real as the fact of the matter about what colour my t-shirt was 8 days ago.
But from the point of view of reality, those of us in the real world,
there is no fact about what colour the PC's clothes are, or were a week ago, until someone makes it up. And no amount of appeal to ingame causal logic will tell you what that colour is. (At best such logic might tell you that they are not purple, because no NPCs had looked askance at the C for dressing above his/her station.)
Likewise for the vampire.
Within the gameworld, it has a history. But that history doesn't acually exist, as something that can be known by the real game participants in the real world, until someone writes it. That's a huge part of what GMing is all about. Gygax even gave advice about it (DMG pp 86-87):
Rome wasn't built in a day. . . . The milieu for initial adventures should be kept to a size commensurate with the needs of the campaign participants. . . . This will typically result in you giving them a brief background, placing them in a settlement, and stating that they should prepare themselves to find and explore the dungeon/ruin they know is nearby. . . .
After a few sessions of play, you and your campaign participants will be ready for expansion of the milieu. The territory arround the settlement - likely the 'home" city or town of the adventurers . . . and whatever else you determine is right for the area - should be sketch-mapped . . . At this time it is probable that you will have to have a large scale map of the whole continent or sub-continent involved . . . In short, you will have to create the social and ecological parameters of a good part of a make-believe world. . . .
Eventually . . . you must . . . broaden your general map still farther so as to encompass the whole globe . . . [and] consider the makeup of your entire multiverse . . . . Never fear! By the time your campaign has grown to such a state of sophistication, you will be ready to handle the new demands.
Gygax recognised that
authorship - making stuff up! - is a key GM skill, that improves with practice. As for macro-details like campaign maps, so for micro-details like the history of a newly-placed vampire: new information about the past of the campaign world has to be created all the time.
And Gygax notes that this will be driven, in part, by metagame considerations: both prosaic ones, like "what do my campaign participants need at this time" and also less prosaic ones, like "what else do I think is right for this area?"
The horse may or may not have existed all along; what the player does - and all the player does - is make that horse become relevant to the run of play. Subtle but in this case quite significant difference, I think.
The boxes may have been there all along, but if the DM failed to mention them then they just never enter play.
The point is that, unless someone actually engages in an act of authorship - actually writes the warhorse, or the boxes, into the gameworld - then they are not in any practical sense
part of the gameworld. For instance, no player will form the belief "That warhorse (or those boxes) are part of the gameworld". No player will declare an action based on the premise of their existence. No GM will adjudicate an action, or frame a scene, having regard to them.
This is one major difference between imaginary worlds and real worlds: in the real world, unknown things can still affect us (before people knew about germs, they still got sick); but if a putative element of an
imaginary world is not known to anyone, ie has not been authored, then it is irrelevant to anyone's experiences of, or imaginings about, that world.
A player (at least one worth having) is going to take game-world forces into consideration when making decisions.
This doesn't show that the gameworld exerts causal power. This shows that real things - namely, game particpants
beliefs about the gameworld - exert causal powr.
Which reinforces the point I've just made. Saying that "the vampire's history was always part of the gameworld" doesn't tell us anything about the play of the game, or the experience of the game. Up until the GM rolled an encounter with that vampire, the players were not factoring its history or existence into their action resolutions, and nor was the GM factoring its history or existence into his/her presentation of the gameworld to the players.
It is only once fiction is authored that people can believe things about it, and hence that those beliefs can influence their behaviour.
To relate this back to its origins in this thread: I stated that a strength of The Forge is that they eschew appeals to the ingame perspective as a way of explaining play experience. If you want to teach someone how to be a good GM when it comes to random encounters, getting them to think really hard about the ingame situation is irrelevant. You need to teach him/her how to
write good ingame situations. You need to teach him/her to be an
author, not an
imaginer.
Imagination may be one useful tool of the author's craft, but it hardly exhausts it, and I'm not even sure it's the most important tool. Certainly not if, by imagination, we mean not
creativity but rather
vivid mental imagary of the sort that immersionists emphasise.