Designing Scenes That Respect Working Memory: A Game Master's Design Perspective

Most advice about information overload at the table is reactive. A game master notices the quiet, the repeated questions, the glazed looks, and then reaches for a recap or a name card. That works, and it is worth knowing. But there is an earlier question: what would a scene look like if it had been designed to be easy to hold in the head?

This article takes that design perspective. It treats working memory as a constraint on scene design, the way a carpenter treats the width of a doorway. It does not give a rescue kit for a stalled session; it asks how to shape scenes, encounters, and information flow during preparation so the table has less to carry before anyone rolls a die.

A note on scope. The ideas below borrow from cognitive load theory, which comes from education research. The foundational paper is Sweller (1988), which examined how the demands of problem solving affect learning. Gaming is not a classroom, and the theory was not built for it. What follows treats it as a lens, not a ruleset, and where something is a matter of table preference, it says so.

Working memory as a design constraint

Cognitive load theory starts from the observation that people can hold only a limited amount of new information in mind at once while they work on a problem. Sweller (1988) is the standard citation for the origin of the theory as applied to learning. The practical takeaway for a designer is modest and useful: if a scene asks players to track many new things at the same moment they are being asked to decide, something will give.

The theory is usually discussed in terms of different kinds of load. Some load comes from the inherent difficulty of the material. Some comes from how the material is presented. A designer cannot remove the first kind without changing the game, but the second kind is largely a presentation choice. That distinction is the heart of this article: you can keep a scene's fiction rich while making its presentation light.

It helps to separate three things a scene asks of the table:

• Holding: keeping facts in mind (who is who, where things are, what the party wants).
• Choosing: weighing options and committing to one.
• Acting: declaring what a character does and resolving it.

A scene that front-loads holding and then immediately demands choosing is the classic stall. Design can sequence these so the table is never doing all three at full intensity at once.

Principle one: budget your new information

Every scene introduces new things. Names, places, factions, rules quirks, objects, rumors. A useful design habit is to give each scene an explicit novelty budget: a rough count of how many genuinely new items the players will need to retain to act sensibly.

There is no universally correct number, and no source cited here supports a specific figure, so treat any number as a table-by-table judgment. What is defensible is the practice itself: decide in advance which new elements are load-bearing (the players must remember them) and which are texture (they enrich the scene but can be forgotten without harm).

A simple sorting exercise during prep:

1. List everything new in the scene.
2. Mark each item must-hold, nice-to-hold, or texture.
3. Keep the must-hold list short.
4. Present texture lightly, and let it be forgotten.
5. Move excess nice-to-hold items to a later scene.

This is close in spirit to ideas about prep economy. Shea (2024) writes about focusing preparation on the material most likely to matter at the table, and the same filter applies to information, not just prep time. If an element will not affect a decision, it does not need to be learned.

The reusable-name trick

One way to lower the budget is to reuse. A minor character who shares a role or a visual tag with someone already established costs less to remember than a wholly new face. Designers sometimes build a scene's cast from two or three archetypes the table already knows, adding variety through manner instead of through fresh names. This is a stylistic choice, and some tables prefer a sprawling cast. It is offered here as an option, not a rule.

Principle two: make the structure legible

Players hold less when the scene's shape is obvious. A scene with a clear boundary, a clear goal, and clear exits gives the table a frame to hang details on.

Alexander (2010) describes node-based scenario design, in which an adventure is built from discrete situations connected by information, as opposed to a single plotted sequence. Whatever one thinks of the broader method, a node has a useful property for working memory: it is a bounded situation. Players know where they are and what is in front of them, and they do not need to simulate a branching future in their heads.

For scene design, legibility has a few practical components:

• A stated purpose. The scene exists for a reason the table can state in a sentence.
• A visible space. A rough sketch or a verbal anchor so players do not rebuild the room in their minds.
• Obvious pressure. Something that makes the next decision feel urgent without requiring a lengthy explanation.
• Clear exits. The table knows how the scene ends or how they can leave it.

The point is not to railroad. A legible scene can still be open. It simply does not make players reverse-engineer its structure.

Principle three: stage information over time

The single most common design error in this area is the info dump, where the game master delivers a large block of description and then asks for a decision. Alexander (2017) discusses evocative description as a craft, and one reading of that craft is that selective detail beats exhaustive detail. A few vivid specifics are easier to hold, and easier to act on, than a complete inventory.

Staging means releasing information in waves, matched to when it becomes useful:

1. The opening wave: only what the players need to orient and take a first action.
2. Reactive waves: details that emerge in response to what the characters do or ask.
3. Reserve detail: held back until it is relevant or until someone seeks it out.

The Angry GM (2025) writes about knowing when to stop narrating, and this is the performance side of the same design idea. A designer who has already decided what belongs in the opening wave has a natural stopping point built into the scene.

There is a trade-off worth naming. Staging gives up some of the pleasure of rich, uninterrupted description. Some tables enjoy long scene-setting, and a game master with a gift for it may reasonably keep it. The design perspective does not forbid that; it asks that the cost be a conscious choice.

Principle four: shape choices, not just scenes

Working memory is taxed by options as well as by facts. Every available action is something a player may feel they need to weigh. Designing a scene includes designing the decision surface, the set of things players perceive they can do.

This connects to long-standing guidance about presenting clear options. Alexander (2008) describes the Three Clue Rule: for any conclusion the players need to reach, provide at least three clues pointing to it. The rule is about redundancy so that a single missed clue does not stall the game. From a working-memory standpoint it carries a second benefit. When clues are redundant, players need not catch and retain each one perfectly, because the scene is built to tolerate lossy attention.

That is a useful general stance. A well-designed scene is fault tolerant: it still works when the table forgets, misreads, or tunes out for a moment. Ways to build in tolerance:

• Repeat load-bearing information through more than one channel (spoken, written, on a map).
• Let a character or object restate the goal naturally in play.
• Avoid designs where one forgotten fact makes the next decision impossible.
• Offer a small, visible set of salient options, while leaving the table free to improvise.

On that last point, a shortlist of options is a presentation aid, not a fence. Walmsley (2009) argues for improvisation and for resisting excessive planning, and a shortlist sits comfortably with that view as long as players know they may step outside it.

Principle five: design pacing as load management

Pacing and cognitive load are closely related. A scene that is perfectly lean in isolation can still overload a table if it arrives immediately after three dense scenes.

Alexander (2013) treats pacing as an art, and Shea (2022) offers practical advice on improving pacing at the table. Both are worth reading for the general craft. Through a working-memory lens, one useful reframing is to think of pacing as the rhythm of demand:

• High-demand scenes introduce much that is new or require hard choices.
• Low-demand scenes let the table consolidate: a quiet travel beat, a camp conversation, a straightforward fight against a familiar foe.

Alternating them gives players room to absorb. This is a design heuristic, not a measured law, and different groups will tolerate different rhythms. A veteran group that has played together for years may carry far more than a newly formed one.

The Angry GM (2016a) writes about keeping pace, and The Angry GM (2016b) writes about structuring a session as episodes. Both frame a session as having shape. If you plan a session's shape in advance, you can place your densest scene where the table is freshest and your lightest where energy tends to sag. Whether energy actually sags at a given point is a matter of the particular group, so observe your own table instead of assuming.

Principle six: externalize memory on purpose

A design that relies on players remembering things is a design that spends working memory. A design that puts those things somewhere visible spends less. The craft is deciding, during prep, which pieces of the scene will live outside heads.

Common externalizations:

• A sketch map left on the table.
• A short list of present characters, each with a trait.
• A visible statement of the current goal.
• Handouts for any text players will need to reread.

This is the planning counterpart to table-side practice. The tools themselves, and how to use them mid-session, are a separate topic from this article's focus on design; what matters here is deciding in advance what to externalize. Prep guides often point at this kind of economy. Shea (2018) lays out a lightweight prep approach for game masters, and one way to read that spirit is that physical aids should be prepared only where they earn their keep.

A caution: externalizing everything creates its own load. A table buried in handouts has traded one burden for another. The design question is which few items genuinely deserve a physical home.

Principle seven: protect the newest person

A scene's load is not distributed evenly. Someone new to the game, to the setting, or to the group carries more, because they lack the shared background others rely on. Designing with that person in mind often improves the scene for everyone.

Two design moves help:

• Give every scene an entry point that needs no backstory. A concrete problem in front of the characters is easier to join than an allusion to prior events.
• Offer role-appropriate hooks. Shea (2023) discusses session-focused character hooks as a way to give each player something personal to engage with. A hook that ties a character to the scene's immediate stakes gives that player an anchor.

Group norms and expectations also matter here. Resources on setting expectations, such as the session zero guidance in Haeck (2024), point toward discussing how the table wants to play. Whether a group prefers dense, lore-heavy play or lighter, faster scenes is exactly the sort of thing worth surfacing before the design choices above get made, since the right novelty budget depends on the people at the table.

A design checklist you can adapt

The following is a prep-time checklist, not a procedure to follow rigidly. Trim it to fit your style.

1. State the scene's purpose in one sentence. If you cannot, the scene may be carrying too much.
2. Sort new information into must-hold, nice-to-hold, and texture.
3. Trim the must-hold list and defer extras to later scenes.
4. Decide the opening wave of information and the stopping point of your narration.
5. Plan two or three salient options, and remember players may do something else.
6. Build in redundancy for anything load-bearing, in the spirit of Alexander (2008).
7. Choose what to externalize: a map, a cast list, a stated goal.
8. Check the rhythm against the scenes before and after.
9. Consider the newest player: is there an entry point that needs no backstory?
10. Leave slack. A scene with a little unclaimed room is easier to adapt than one packed to the edge.

Where the lens stops working

It would be a mistake to treat working memory as the whole story. A few honest limits:

• Cognitive load theory is a learning theory. It was developed to explain how people learn from instruction (Sweller, 1988). Applying it to play is an analogy, and an analogy can mislead. Players are not only learning; they are also socializing, performing, and having fun.
• Some load is the point. Many players enjoy rich, intricate material, and for them a complex political web is a pleasure, not a tax. Designing for lightness is a preference, not a virtue.
• Tables differ. Experience, familiarity with the system, and mood all change how much a group can carry. Nothing here should be read as a claim about how most groups behave.
• Prep has its own costs. Elaborate load-management design can itself become heavy. If a technique costs more prep time than it saves at the table, it is probably not worth it.

The design perspective is best used as a set of questions, not answers: What does this scene ask the table to hold? Is that worth it? Where could it live instead?

Example: The Lantern Road Company

This is an illustrative scenario, not a report of an actual game. It shows how a design-first approach might look for a fictional group.

The Lantern Road Company is a four-player party escorting a merchant caravan through a mountain pass. The game master is preparing the opening scene at a crowded waystation and wants to design it for light load from the start, instead of fixing problems at the table.

Step one: the novelty budget. The game master lists everything new the scene could contain: an innkeeper, her husband, three rival caravan masters, a tax collector, two travelers, the layout of the common room, the weather, and a rumor about bandits. Sorting the list, the game master decides what is must-hold: the innkeeper (who wants help with a broken cart), one rival caravan master (smug, wants to leave first), and the tax collector (nervous, wants an escort). Everyone else becomes texture, mentioned in passing or dropped. The husband, two of the rival caravan masters, and the two travelers are deferred to later scenes.

Step two: structure. The scene's purpose, in one sentence, is to give the party three small problems at the waystation while they stay aware that the pass will close. The game master sketches the common room as a rough rectangle with the hearth, the door, and the stable yard marked, so the space does not have to be rebuilt in anyone's head.

Step three: staging. The opening wave contains only the sketch and a single sentence about the crowded room. The three must-hold characters are introduced one at a time, each in a short burst, as the party moves through the room. The rumor about bandits is held in reserve, to be offered only if a character asks around.

Step four: fault tolerance. The party's goal, reaching the next town before the pass closes, is designed to surface through more than one channel. It appears in a one-line recap, again when the tax collector worries about the closing pass, and once more when the innkeeper mentions the weather. If a player misses one, the others carry the information.

Step five: rhythm and the newest player. Because the group includes a newer player running a cautious scout, the game master gives that character an entry point that needs no backstory: the scout can notice from the door that the rival caravan master's wagon is already half-loaded. The scene is placed after a quiet travel beat and before a denser negotiation, so it sits in a moderate position in the session's rhythm.

Nothing about the fiction was simplified. The waystation still has its crowd and its tension. What the design changed was how much the table had to hold at any one moment, and where the rest of it lived.

A closing thought on craft

Designing for working memory is not about making games smaller. It is about spending the table's limited attention on the things that matter to the story, instead of on bookkeeping. Prep tools can support this, whichever form they take. The Dungeon Trainer (2026) is one resource that aims to help people play together and improve their craft, and its AI Dungeon Master is optional; the choices above remain yours and your table's either way.

If you take one habit from this article, make it the novelty budget. Before a session, look at each scene and ask what the table genuinely needs to hold. The answer is usually shorter than the draft.

Sources and Further Reading

• Alexander, J. (2008, May 8). The Three Clue Rule. The Alexandrian. https://thealexandrian.net/wordpress/1118/roleplaying-games/three-clue-rule
• Alexander, J. (2013). The Art of Pacing. The Alexandrian. https://thealexandrian.net/wordpress/31509/roleplaying-games/the-art-of-pacing
• Alexander, J. (2017, September 4). Random GM Tip: Evocative Descriptions. The Alexandrian. https://thealexandrian.net/wordpress/39227/roleplaying-games/random-gm-tips-evocative-descriptions
• Alexander, J. (2010, May 27). Node-based scenario design, part 1: The plotted approach. The Alexandrian. https://thealexandrian.net/wordpress/7949/roleplaying-games/node-based-scenario-design-part-1-the-plotted-approach
• Haeck, J. (2024, March 11). How to run a Session 0 for your D&D game. D&D Beyond. Original work published January 8, 2021.
• Shea, M. E. (2018). Return of the Lazy Dungeon Master. Sly Flourish.
• Shea, M. E. (2022). Improving Pacing. Sly Flourish. https://slyflourish.com/pacing.html
• Shea, M. E. (2023, June 19). Calm Pre-Game Nerves with Session-Focused Character Hooks. Sly Flourish.
• Shea, M. E. (2024, June 3). High Value Prep. Sly Flourish.
• Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257–285.
• The Angry GM. (2016a, March 23). Keeping pace. The Angry GM.
• The Angry GM. (2016b, June 8). When episodes ain't episodes: How to structure a session. The Angry GM.
• The Angry GM. (2025, July 31). When to stop narrating. The Angry GM.
• The Dungeon Trainer. (2026). The Dungeon Trainer.
• Walmsley, G. (2009). Play Unsafe. Graham Walmsley.