Let’s compare a role playing game to someone telling a story. When someone is telling a story, they are likely in complete control of the narrative. They talk, others listen. They dictate what happens in the story. The audience are just passive observers who have no control at all on what happens or the outcome. The narrator has complete control over the story and it’s characters.
You could also consider a shared story. One person might be the primary narrator, and through some means the audience can have some impact on the story itself. It could be as simple as a binary yay/nay, up/down, and the narrator tunes the story as he or she goes, influenced to some degree by the audience who is now interacting. The narrator still has complete or near complete control, but has agreed to give the audience some voice in the story.
A more shared story might go round robin, where one person tells some story, and then the next person continues telling the same story. In this case, each person who is at that moment telling the story is in complete control. Thus participants have complete control over the story when it is their turn, and no, or little, control when it is someone elses turn.
In a typical D&D style RPG, the GM/Narrator is assumed to have complete control over the story. The players don’t dictate what happens, the GM does all of that - but the GM might give players a chance to influence the story in small ways, no directly on the story but instead indirectly through their characters. Thus you find that the players can perform specific actions in the hope of having some limited control over the story. Their interface to the story is through their characters.
In other, more ‘indie’ style games, you might find an arrangement where the GM is considered to be in primary control of the story, but the players are also given some narrative ability. The difference in this and the previous D&D example is that in D&D, the players can try to convince someone of something, or to perform some physical feat, or to kill a monster, or save a friend, but that’s the limit of their interaction. In this indie example, they might still control a character in the story, but they might have the opportunity to have more narrative control - to perhaps change the story itself in some mechanically appropriate way. The electronic game Scribblenauts comes to mind.
GM: “Men in armor with swords burst into the room. They seem to be searching for something, or someone…”
Player, exercising narrative control: “They bear the colors and sigil of Duke Harald, and the man leading them is my father!.”
I theorize that a story that lets multiple participants have the control to introduce elements or in some fashion to narrate will be a better story. When we play, we all aspire to be like an author in some fashion. But unlike writing, role playing is live - you don’t have the luxury of an editor or rewrites.
Players can gain some level of narrative control - or influence perhaps - when they say to the GM “Hey, I’d like to do so and so in the game.” And the GM says sure, and later incorporates some element of that into the game. This is just a slightly different way of giving the player some control.
Some systems specifically give the players means to control the story, or rather to move it in directions that they think are interesting. There are a few systems that allow a player to say “I know that person” when an NPC is introduced. Thus they can make the story more interesting, and feel/be more involved with the story.
I experimented with this in a Shadow of Yesterday game. The party was searching for someone, and had a few clues, and then wanted to go to dice to determine whether they found the person. Dice rolls are commonly seen as having a binary outcome. Did you find the person, yes or no. The players want to find the person, and the GM wants the players to find the person too, but the GM wants they players to have to work for it. A dice roll is commonly seen as that work. But what if they dice are unkind, and they say no? Instead of simply deciding yes or no, let the dice be an arbiter of what’s going on - you could say that a failure results in a twist. “Did we find uncle Bob?” If the results are a success, then great - there he is. Story continues. If the result is a failure, it does no good for them not to find uncle Bob. The whole story revolves around him, or he’s the next step in the story. But instead throw in a twist. Yes, you found uncle Bob, but he’s about to be killed by evil aunt Wilma, or his house is burning down when you pull up in front of it, or he’s just plain dead. At best, it’s a conflict/encounter that may be interesting, but again has the potential to be show-stopping - what if the party gets TPK’d or the fight doesn’t go as planned? If you go for heavy twists (bob is dead), you can’t just drop that on your players without a plan for them. There needs to be a lead - can he be resurrected? Can the information that they needed from him be found some other way? A twist should be interesting and exciting, rather than frustrating and disappointing.
An interesting story is generally going to be one in which some sort of adversity was overcome. No one wants to hear a story about how you went to the grocery store - unless you were mugged upon your arrival. A good story is always going to have at least one conflict. Two sides who want two different things. It may not involve violence or war, but it could instead involve some sort of negotiation or compromise.
If you think of stories in terms of RPGs, you can imagine the dice rolling behind any blocking or conflict scene. Is that door locked, can the party get through it? Can the party escape from their pursuers? Can the Duke be convinced to support the claim? Will the owlbear kill the party? Can the group find its way through the labyrinth? Each of these are questions that have the potential for more than one answer. The door is locked or barred and the party must find another route. What happens when the pursuers catch up to the party? And so on.
RPGs are still basically a game about storytelling. The GM is telling a story and the players are helping. Stories require details. Details make a story come alive. Details make a story interesting. They introduce new elements or possibilities.
Player input and dice rolls are commonly going to be informational or factual. Informational might be whether they spot a secret door, or pending ambush, or interesting clue. Factual is whether they were able to succeed or fail at some task. Climbing a wall, riding a horse, killing an orc, convincing a shopkeeper. There is some logic that insists that informational dice rolling be dispensed with.
There’s another facet here - and that’s random twists. Much like the random encounter chart. Consider that the party is traveling through a dark and sinister looking forest. A player asks “Do I hear anything, or notice anything suspicious?” The GM has not intended to drop an encounter here, and can simply say “no”, and move on. Or the GM can have them roll a dice, the dice roll determines whether there IS in fact something to be encountered. Random twists can be fun, but they can also distract when there is a story being told. Some games could be very nearly one random twist after another - but this will result in a disjointed story.
So, what kind of control do the players want? How do they want to interact with the story? Do they simply want to be told what’s happening, and try to push buttons to influence the story? Or do they want more control, and to be able to modify the story in significant ways.
When players create characters, they are making decisions about the kind of game that they want to play, and how they want to play it (and how they want to interact with the world around them, and the game before them).
Most popular RPGs, no matter the genre, boil down to the same effective level of control for the players. They are provided with opportunities to tweak the story in small ways by the mechanics and by the GM.
Consider that there are already many elements of decision making done by the GM via fiat. Is there a car on the street? asks a player. Is the castle gate open? Do the orcs notice us? Is the Baroness mad? All of these could be determined by GM fiat, or a random (or semi-random) dice roll, or even by player input. “Say yes or roll dice” is a line from Dogs in the Vineyard. That game encourages a GM to give the players what they want in the story, and if the GM is unsure of the outcome, to go to dice and let the game’s mechanics sort out the answer.
Those of us who are familiar with RPGs oftentimes have learned how to play “correctly”, or the way that we are accustomed to. People who are new to the game will be interested in exploring how much control they have and what they can accomplish.
What type of story or game is the GM going for though? If they are trying to tell a very specific story, then giving the players more control can be problematic for them. If the Lord of the Rings were taking place in an RPG environment, and the GM had already planned out much of the story, letting the players have more control could derail the story. The GM wants it to be about the difficult journey to Mordor, fraught with difficulties, and the fate of the Ring and the protagonists. Player control could send the story spinning off in some other direction though.
The back and forth of player input and GM narration can be configured in a variety of ways. Consider the question of “How will the party get to Mordor?” The party can likely choose a variety of paths to get to their destination. Each choice may result in descriptive and narrative differences, and in conflicts or further decisions. In an RPG environment, this can be approached either with the “all roads lead to rome” philosophy in which the story still progresses in much the same way. A better example is perhaps a party in a cave system. They arrive at a fork in the tunnel and can choose to go left or right. In some games, the GM may have prepared for each choice, having descriptions and encounters and such ready for either choice. This requires more work certainly, and leaves the possibility of some of that work going unused. Another approach is for the GM to let the players make their choice, and to run descriptions and encounters much the same either way. The hazard here is if somehow both choices are invoked. “Well we chose the left path, but lets go back and do the right path instead.” This forces the GM either to make sometimes arbitrary and meta-seeming decisions: “Rocks fall and block your way back.”, or to come up with something for the other choice.
A common pitfall (or feature, depending on your perspective) for a new GM is giving the players a choice, and having designed their game in such a way that the players are required to make a certain decision. If the players make a different decision, or ignore the GM’s offer/bait, then the GM is potentially left struggling with something important that was missed. “The party completely ignored the basement, where they were supposed to find the ghost!” The simplest answer to this sort of problem is to simply present what you need/want the players to find. You WANT them to find the ghost in the old house. If for some reason the ghost needs to be in the basement, then the players need to be guided to the basement, or placed into the basement (the floor collapses, or what have you). Having a McGuffin that the players need to find can result in a boring search, especially if the players realize that they are trapped in the scene - and even worse if the players don’t know what is required to leave the scene.
The party finds itself in the old McLauren house. The GM wants the players to find the ghost of old Ms. McLauren in the basement so that they can get a quest from her and put her to rest.
That’s the story that the GM has prepared for the night. The players are initially unaware of the ghost though. One player says that they want to look for the way out. The GM is naturally opposed, so has 3 choices. Say yes, say no, or roll dice. The GM in this scenario is unlikely to say yes or roll dice, since he wants the players to remain in the scene and find the ghost. The players may feel frustrated, and try increasingly escalating methods of leaving the house, once they see that they are blocked.
The doors are all stuck? Fine, we’ll chop them down.
Now the GM is forced to start using weird meta devices, or to simply tell the players that they can’t leave, which is still frustrating for many players. Instead, give the players what they need! Have the ghost pop right out, or give them a trail of breadcrumbs to the basement. But remember that players may ignore breadcrumbs - either they fail to notice the trail, or they decide that they’re not interested, or they think that it is unimportant.