World Adhesion and You

Between The Reporter and the Machine

Bounded Agency: How Rules and Prep Power the Illusion of Total Freedom

Firstly a shout-out to The Disoriented Ranger for being the origin on this article and also the foil to my own thinking. As one iron sharpens another.

When a player sits down at a tabletop RPG table, they are typically looking for one thing above all else: ‘agency’ (We could perhaps argue over this but if your actually reading this I think you won’t). They want to know that their choices actually matter, that the world will react dynamically to their intent, and that they aren’t just passengers on a hyper-railroaded plot.

But behind the screen, a perpetual debate rages among Gamemasters: “How do we actually deliver that freedom?”

Let’s ignore all those reasonable people in the middle and look at the two most extreme ends of the spectrum.

On one side, you have the advocates of runtime generation, the “Jazz Musicians” who believe a great GM must be a master of pure, on-the-fly improvisation to handle the infinite chaos players bring to the table.

On the other side, you have the system purists, the “Simulation Engines” who argue that rigorous preparation, massive setting databases, and meaty, procedural rulesets can handle almost any curveball without the GM needing to invent a single thing out of whole cloth.

To understand why both sides are actually right, we have to look outside of gaming entirely and examine two concepts from sociology and organizational theory: Bounded Agency and Structured Improvisation.

The Illusion of Total Freedom

To the player, their agency feels absolute. If they want to crash an alien spaceship into a high-fantasy ocean and blow the airlock, they expect the world to handle it.

But in sociology, researchers have long understood that human willpower never exists in a vacuum. The term Bounded Agency, popularized by sociologist Karen Evans in the late 1990s, describes how individuals navigate their lives. Evans argued that people don’t have free-floating, infinite agency to do or become anything. Instead, their choices are bounded by their environment, past experiences, and institutional structures. Yet, ‘within’ those strict boundaries, their creative choices remain highly meaningful.

At the gaming table, the rulebook and the setting lore are those structures. The players can’t just wish the water away, but they have bounded freedom to solve the crisis using the tools allowed by the environment.

Without structure, agency loses consequence; without flexibility, agency loses possibility. If you blow an airlock underwater and a low-prep GM just says, “Uh, okay, you all magically wash up on a beach because I don’t know what else happens,” your choice didn’t actually matter. The system’s rigid boundaries: hydrostatic pressure, drowning timers, hull integrity ratings, etc are what give your chaotic choices their teeth.

The Jazz Framework

This brings us to Structured Improvisation, a concept shared by ethnomusicologists studying jazz and organizational theorists studying fast-moving tech companies in Silicon Valley.

When a jazz musician improvises a jaw-dropping solo, they aren’t hitting random notes in a vacuum. They are bound by a strict, mathematical framework: the chord progression, the tempo, and the key signature. Without that rigid structure, four musicians playing together would just produce unlistenable noise.

Similarly, running a meaty ruleset (like AD&D or HackMaster) combined with deep setting literacy doesn’t kill creativity; it fuels it. It shifts the GM’s mental workload from Generation (inventing a universe out of whole cloth) to Compilation (applying a robust database of existing rules and lore to a new variable).

When the spaceship crashes underwater, a hyper-prepared GM doesn’t have to panic or “bullshit” a narrative. They look at the environmental hazards table, check the hull specs, and run the calculation. The structure provides the underlying logic, which frees up the GM’s cognitive bandwidth to focus entirely on the dramatic performance.

Behind the Screen: The Five Sources

The fundamental trap of the Prep vs. Improv debate is that we often view it from the wrong side of the screen. The game ultimately exists to serve play first. The player nursing an ale at the tavern doesn’t care about the GM’s epistemological source; they just want a fast, believable answer to a simple question:

“Who is the guy sitting to my right, and what’s his deal?”

Behind the curtain, the GM might pull the answer from five entirely different sources:

1. The Prepared Note: “That’s Joran, a retired caravan guard.” (Pure index retrieval)

2. The Random Table: *Rolls dice* “That’s Kellan, a former sailor.” (Procedural interpretation)

3. Setting Extrapolation: “That’s Bren. He’s a local woodcutter arguing about river tariffs because this town’s economy is based on lumber.” (Systemic calculation)

4. Pattern Extrapolation: “That’s a wealthy merchant having marital problems.” (Internalized genre tropes)

5. Pure Creation: “That’s Vargo. He’s secretly a red dragon in human form.” (Pure runtime invention)

From the player’s seat all five of these answers are completely indistinguishable. The source of the data is entirely invisible. The magic circle remains intact as long as the answer arrives quickly and maintains the internal consistency of the world.

```

[Player Input] ──> │ Opaque GM Screen │ ──> [System Logic / Lore / Improv]

│

▼

[Consistent World Reaction]

```

If the source of the data is entirely invisible, it suggests that from the player’s perspective, the acts of prep, random generation, setting knowledge synthesis, pattern recognition, and improvisation are not five different acts but one: World Response Generation.

The player only observes a handful of outputs: response speed, consistency, plausibility, consequence, and GM buffering (the observable hesitation or rustling of papers while finding or creating the answer). Because of this, we can compress the larger, academic categories of improvisation and extrapolation into a single, functional act: Inference.

The Fluency Paradox

A native speaker of a language doesn’t consciously plan the grammar and syntax of every sentence before they speak; their fluency makes the underlying structure invisible. They don’t memorize static sentences. Instead, a veteran GM has compressed setting knowledge, procedural rules, genre expectations, and narrative instincts into a unified mental model that can be queried in real time.

The act of creative output from this compressed model becomes so deeply informed and so fast that the illusion remains seamless. The GM stops appearing as if they are “making it up” as they speak.

Preparation and improvisation are not rival techniques. Prep is the scaffolding that allows high-quality improvisation to happen. Sweat in production saves blood in battle; smart preparation converts blind invention into logical inference. The player experiences both invention and inference exactly the same way: “the world answered.”

The Illusion of Constraint

This reveals an important division in player psychology. We can simplify players into two distinct camps based on how they perceive the GM’s authority:

1. The Author Camp: The player feels like the GM is just arbitrarily making things up to suit the moment.

2. The Reporter Camp: The player feels like the GM is merely reporting on a concrete world that already exists independently of the table.

Objective reality tells us that both worlds are equally fabricated. The player isn’t actually judging the act of creation itself; they are judging ‘Grounding’.

If a player feels a GM is just “pulling it out of their ass,” it’s usually because the answer lacks structural friction. If you want the cheat sheet to reporting worlds that feel real, the secret lies in providing answers that include the implicit constraints and realities of the broader environment.

A: A Weak Answer (Authorship): “That’s Bren. He’s an angry guy in the corner.” (Provides the bare minimum information; feels like arbitrary creation to fill a vacuum.)

B: A Strong Answer (Reporting): “That’s Bren. He’s a local woodcutter arguing with the barkeep about river tariffs because the logging guild is trying to squeeze the local transit routes.” (Provides the information, anchors it to regional economics, proves the presence of outside constraints, and naturally leaves the player with space for follow-up questions.)

The ultimate purpose of preparation is not to eliminate invention. It is to make invention appear to your players as if it were completely inevitable.

Weak answers feel arbitrary, strong answers feel grounded in larger structures.

World Adhesion

When a player accuses a GM of “making things up,” they aren’t critiquing creativity or novelty. Knowingly or not, they are judging the density of connections that ground a fact within the larger whole.

The more grounded the connections, the more “right” an answer feels. We might have just invented Bren the logger six seconds ago, yet because Bren is grounded by forces larger than himself (within the established reality of our world) the players experience him as something discovered rather than GM-created.

A believable world is not one where every possible question is pre-answered, but one where everything is grounded in the same root logic and interconnected with the rest of the setting.

In short, a world starts feeling real when things stop existing in a vacuum. Bren is grounded in economics, geography, history, and the actual mechanics of play. These facts “stick” to each other and to Bren because they are internally consistent and externally logical.

One fact becomes a cluster. In the example above, we established that an entire logging guild is tied to the river, and that there is an ongoing political dispute. We only asked about the guy at the bar, but when we pulled that thread, we unspooled a whole web of information about the local scene.

The Engine of Adhesion

We started by debating prep versus improvisation, but both styles are just different engines generating the exact same output: a dense web of connections.

The Prep-Heavy GM builds adhesion structurally: relying on established lore, written procedures, defined factions, and session notes.

The Improv-Heavy GM builds adhesion dynamically: relying on callbacks, strict internal consistency, pattern recognition, and inferring relationships on the fly.

Most GMs will fall somewhere between these two extremes.

Same outcomes. Different methods.

The Player’s Audit: Context Over Content

Players do not care how many hours you spent writing a backstory. When they interact with the world, they are unknowingly measuring fictional mass: “How many other things seem attached to this THING?”

A dragon disguised as a merchant may feel infinitely more real than a meticulously prepared blacksmith if the dragon has actual adhesion to the surrounding world. If that merchant’s presence ripples through the local economy, aligns with recent political shifts, and explains the burn marks on the old bridge, they exist. If the blacksmith exists in a vacuum, they are just a cardboard cutout.