Well but, by initially agreeing to play the game by the rules, everyone has already come to all possible in the moment agreements, clearly. Therefore nothing requires in the moment agreements, because all of those undecided things have in fact been decided initially. (For some meaning of "initially", and/or perhaps "moment", and/or especially "example". And also maybe "agreement"? I mean we haven't established a consensus on that.)
It's certainly part of what I intended to imply concerning "just so long as one includes the prospect that the superior agreement might have been imperfectly formed" and in a subsequent post "cases we are dealing with readily stray outside our promises." Consider -
During an ordinary shopping expedition in New Pavis a player asks [RQ7] "How much for a bison-hair cloak" and their GM consults a table and says "Three lunars" so the player deducts three lunars from the total on their character sheet and writes there "bison-hair cloak".
Why don't they -
every time - get into a discussion about the availability and costs of bison-hair cloaks in New Pavis? If they have to agree in
every moment, don't they need to have that discussion?
It is Fire season and the group is in Balazar. A player asks about the weather. GM rolls a 7 and says that it is hot, clear, no precipitation.
Is it expected that -
every time - the group will get into a discussion about whether that is true? That doesn't describe very well play that I have participated in and observed of RuneQuest. But these examples have been chosen to represent what I suppose for most groups is safe ground. The tables in question give clear direction as to what to add to ongoing play. Suppose a player asked a follow-up, like "How hot, exactly?" or "Can I tell if it will rain tomorrow?" Or if due to ongoing conflict GM had earlier described that goods are scarce in New Pavis?
You used the word "agreement" and I think that requires more to be said about "our promises". Our promises are not of the form that when asked what the weather is we have agreed to say it is "hot" - i.e. it's not a commitment that a
specific detail will be true. Rather it's a promise that we'll follow the rules and where in doing so we roll on a weather table that indicates "hot, clear, no rain", we have agreed up-front to work that "hot, clear, no rain" result into our ongoing game state.
@pemerton's examples I take to test where we "stray outside" our up-front promises, which will often happen because of the open-ended contents of [imagine].
Here is one simple and in my view reasonably typical example from D&D play:
The GM tells the players that their PCs arrive at a town. One of the PCs says "I want to go and see a metal-worker, to see if I can buy a silver dagger or some silver-tipped arrows to deal with that werewolf that's been tracking us." The GM considers this action declaration, consults their notes which mention "a typical town" but say nothing about the availability of silvered weapons, decides there's a 50-50 chance, makes a roll, which comes up negative, and then tells the player "You find your way to a weaponsmith, but they don't have any silver or silvered weapons on hand." Then the player says, "We've got that sack of silver pieces, right?" and another player nods, and the first player says "Can the weaponsmith do it if I supply the silver", and the GM thinks about this . . .
No table result is available here, and it's interesting to ask what kinds of things can the GM say in response to that request? Can they say anything they like?
"Sure, but you'll have to sacrifice a finger for each arrow."
"Sure, but after doing it, it turns out that the arrows mind-control their possessor. I'll roleplay your character now."
"Sure, but [in the anticipated encounter] it turns out the werewolf is invulnerable to silvered weapons."
To my mind, we see good evidence that any moment-to-moment agreements are made within the bounds of prior over-arching agreements. (Assuming the inclusion of agenda and principles in contemporary game texts were not evidence enough!)
Here's another:
The PCs are exploring a dungeon. The GM tells them that suddenly, around the corner in front of them, comes a group of 3 ogres. The PCs are surprised! Then one of the players says "Hey, did you take it account of <this alertness-type feature that I got as a result of levelling up last session>?" The GM replies "Oops, I forgot about that. OK, so the ogres come around the corner but you're not surprised - you hear them coming and have time to ready your weapons and roll for initiative!"
There are endless variations on the above in D&D, because of all its intricate little components: did you remember my protection effect, my combat buff, the fact that I'm doing fire damage, etc, etc.
No in-the-moment agreement is being made here: we're just applying an up-front agreement to follow the rules within human limitations. "Oh, I forgot the rule says X, we need to retcon that slightly." I can extend the example to have GM ignore that up-front agreement "Nope, too late, I don't care that it's a rule. They get surprise!" That would be a negotiation... it's absence from the example reinforces my point.
Here's another:
The GM tells the players, based on extrapolation from whatever has happened so far, "OK, it's dark and you're all starting to get cold." One of the players says "We light a fire." The GM replies "OK, tell me how. Does anyone have a tinderbox? What are you using for fuel?"
The GM seeking more information from the players before assenting to their suggestion that a certain sort of craft-y or creation-y outcome is part of the fiction is very, very common in D&D play.
This is a good example of what Baker means by "negotiation" - as was the example above. Both should make clear that said "negotiation" is not in conflict with overriding up-front promises. Perhaps it could be better described by a word like "application".
Here's another:
A player says, "My magic-user wants to research a spell that will let us track the location of the <whatever>." And so the GM pulls out the rules chapter on spell research, and the player and GM start to work out what the PC has to do to research this spell.
Here's another:
The group are resolving a combat. A player says "I attack the Hobgoblin!" The GM replies "How do you do that? You've just finished killing the Bugbear, and the Hobgoblin is on the other side of the room from where the two of you were fighting?" "OK, I move across the room first. Can I do that without being attacked by the giant ape on the way?"
And another:
The GM tells the players that their PCs come to a dead-end room, with a ceiling about 20' high. The player of the Half-Orc barbarian says "The Halfling gets up on my shoulders and starts tapping the walls as high up as she can to see if they're hollow." The GM asks, "OK, what's your DEX? Is either of you trained in Acrobatics?" And the player responds, "Redgar" - the human fighter with 18 STR and 16 DEX - "joins us, so that Redgar and I are really stable with arms around one another's shoulders and the Halfling stands on both of us." And so the conversation goes on . . .
That last one is inspired by Gygax's example of play in his DMG.
As these examples illustrate, and as I hope is reinforced by the myriad other possible examples that they bring to mind, coming to agreement on what is happening in the fiction is at the core of D&D game play.
I would just slightly modify this to say that we come to agreements in the moment only to the extent that doubt exists. There are plenty of things we don't come to agreement on. For example
GM "Roll a d6"
Player, rolling "5"
GM "Nope, I'm calling that a 2"
Players "Huh?"
No one expects moment to moment agreement as to the number plainly visible on a thrown die.... but what if it were cocked? how did we decide to call for that roll in the first place? and what would a 5 (or 2!) be interpreted to mean?