There is a sentence in Room Escape Artist's piece on escape room sequels this week that reads like a craft observation and is actually a statement about cognition. The challenge of a sequel, they write, "is balancing the narrative for players who have played the first game and ones coming fresh into the story." The piece is built around Immersia's The St-Onge Case, the follow-up to Weekend at the Shack, and how its designers handled exactly that split.
I want to take that sentence seriously, because underneath the narrative framing is a problem that the rest of puzzle design mostly gets to pretend does not exist: a sequel room is a single physical artifact that has to be cognitively legible to two populations whose prior-knowledge scaffolds are not the same.
The Usual Luxury of a Single Scaffold
Most puzzle experiences are designed for one assumed prior state. A standalone escape room can assume its players walk in with the genre literacy of a typical escape room customer and nothing more — no knowledge of this specific room, because there is nothing to know yet. A puzzle hunt can assume its solvers have the toolkit of people who do puzzle hunts. A cryptic crossword assumes you know the conventions of cryptics or you don't, and if you don't, you are not really the audience.
This is a luxury, and designers rarely notice it as one. When you can assume a single prior-knowledge state, you can calibrate every clue's difficulty to a known baseline. The iterative-cluing methodology escape room designers use — watch playtesters fail, adjust, watch again — works because the playtesters and the eventual customers share that baseline. The designer is tuning one curve.
A sequel breaks the single curve. On any given night, The St-Onge Case seats a team that has the full narrative and mechanical memory of Weekend at the Shack next to — or worse, in the same group as — a team that has never heard the name Philippe St-Onge. Those two teams are not running the same cognitive operation when they pick up the same prop. One is doing recognition. The other is doing first-encounter encoding.
What the Scaffold Research Predicts
Last week I wrote about a Nature Communications paper on cortical knowledge structures — the claim that pre-existing cortical structures actively direct what new concepts a person can learn and how. If that framework is even roughly right, then the returning player and the fresh player are not just differently informed. They are differently wired for intake at the moment the room delivers a clue.
The returning player has a scaffold. When The St-Onge Case surfaces a detail that connects to the first game, that detail lands on existing structure — it gets bound fast, with low effort, because the cortical structure already has a place for it. The fresh player has no such structure. The same detail arrives as raw material that must first build the scaffold it will later hang on. Same clue, same room, two genuinely different cognitive events.
A naive sequel designer treats this as an information problem and solves it with exposition: a cutscene, a text panel, a briefing that catches the newcomer up. This fails in a specific and predictable way. Front-loaded exposition asks the fresh player to encode a large block of narrative scaffold before the active solving begins — and asks the returning player to sit through scaffold they already have. It is the encoding-phase bottleneck I wrote about yesterday, inflicted deliberately: a wall of intake stacked at the front of the experience, exactly where it interferes most.
The Move Immersia Made
What Room Escape Artist describes Immersia doing is the better move, and it is better for a reason the craft language does not quite name. Rather than front-loading the backstory, they "wove the predecessor game's narrative directly into puzzle mechanics themselves," letting players learn the villain through gameplay.
Read through the scaffold framework, this is not just elegant pacing. It is distributing the scaffold-building across the encoding moments that were going to happen anyway. The fresh player builds the St-Onge scaffold one puzzle at a time, each piece arriving bound to a concrete mechanical action — picking up the object, turning the dial, opening the drawer. The scaffold gets constructed out of haptic, situated events rather than out of a paragraph read cold. And the returning player, encountering the same woven detail, runs recognition instead of construction — the callback fires, the existing scaffold lights up, and the moment reads as a reward rather than a delay.
The same design element does two different cognitive jobs depending on the scaffold the player brought through the door. That is the actual solution to the two-audiences problem: not content that serves both, but content whose function changes with the solver's prior state. The newcomer gets construction. The veteran gets recognition. Neither gets a wall.
Why This Generalizes Past Escape Rooms
I keep coming back to this because the two-audiences problem is not unique to sequels — sequels just make it impossible to ignore. Every puzzle with a heterogeneous solver population has it. A community ARG has solvers who have been following since the first drop and solvers who arrived yesterday. A long-running cryptic setter has solvers who know their personal tells and solvers meeting them for the first time. A museum puzzle trail has the family that read the introductory placard and the family that walked straight past it.
The escape room sequel just removes the designer's ability to pretend. You cannot calibrate to one scaffold when you can see both scaffolds walk into the room together. And the craft answer that emerges under that pressure — design elements whose cognitive function is scaffold-dependent rather than fixed — is probably the right answer for the easier cases too. We just don't notice we need it when the population looks uniform enough to round off.
What I find genuinely satisfying about the Immersia example is that the designers almost certainly did not arrive at it through cognitive science. They arrived at it through the iterative-cluing loop — watching returning players get bored by exposition and fresh players get lost without it, and adjusting until both curves resolved. The scaffold research did not tell them what to build. It tells the rest of us why what they built works. I would like to know whether a sequel room could be instrumented to measure the difference directly: same prop, two solver populations, and the question of whether the binding signatures actually diverge the way the framework predicts they should.