A joke has three stages, not two. Everyone knows about the setup and the punchline. What most people miss — and what makes the difference between a joke that reliably lands and one that sometimes works and sometimes doesn’t — is the middle stage: the assumption. The setup creates an assumption. The punchline shatters it. The gap between the assumption and the shattering is where the laugh lives.
Greg Dean articulates this in Step by Step to Stand-Up Comedy with unusual precision, and when I read it I immediately recognized what he was describing — because it’s also a precise description of how magic works.
The First Story and the Second Story
Dean’s framework is built around the idea that a joke always tells two stories simultaneously. The setup creates what he calls the “first story” — the interpretation the audience builds based on the information given and the assumptions they bring to it. This story feels natural and obvious. The audience isn’t aware they’re constructing it; it assembles automatically from the words and context.
The punchline reveals the “second story” — an entirely different interpretation of the same information that was hiding behind the first. The audience’s first story suddenly collapses and the second story appears in its place. This cognitive shift, from first story to second story, is the mechanism of the laugh. The pleasure is the pleasure of the shift itself, the sudden jump between two ways of understanding the same facts.
The element that makes this possible is what Dean calls the “connector” — the word, phrase, or situation that can be read in at least two different ways. The setup directs the audience toward one reading. The punch reveals the other. Without the connector, there is no joke — there’s just two unrelated statements.
Here’s a rough example of the structure: setup — “I asked the librarian if the library had books about paranoia.” This builds a first story: a person asking a reasonable question of a librarian. The connector is “paranoia.” First story reading: someone with an interest in the topic. Now the punch: “She whispered, ‘They’re right behind you.’” The second story: the librarian is herself paranoid, interpreting the question as evidence the person is being followed. The connector, paranoia, supported both stories. The punch collapsed the first and revealed the second.
The laugh is not at the punchline. It’s at the collapsing.
The Direct Parallel to Magic
When I worked through Dean’s framework in detail, the parallel to effect structure in magic was impossible to ignore.
A magic effect also operates through two stories. The performance creates a first story — the spectator tracks what appears to be happening, builds a mental model of the situation, constructs a causal narrative. Then the reveal produces a second story — a reality that has clearly been true the whole time but that the first story completely obscured.
The connector, in magic terms, is whatever element was doing double duty throughout the effect. What the spectator was interpreting as one thing was actually something else. The reveal brings the second story forward and the first story collapses.
The experience of impossibility that magic produces is, structurally, the experience of this collapse. The first story was so compelling, so well constructed, that the second story was invisible — and yet the second story is right there in hindsight, built from the same facts. The cognitive dissonance between “I was watching the whole time” and “I had no idea” is what creates the experience of impossibility.
This is not a loose analogy. It’s the same cognitive mechanism operating in both art forms. The joke makes you laugh. The magic makes you feel impossibility. The structure is the same: two stories, one set of facts, a reveal that shifts the frame.
What This Means for Patter
Once I understood the two-story structure, I started analyzing my patter differently. Every piece of verbal material in a performance is either building the first story, protecting the connector, or revealing the second story. Anything that doesn’t do one of those three things is dead weight.
Dead weight in patter is more damaging than it sounds. It isn’t just wasted words. Patter that doesn’t serve the two-story structure often inadvertently weakens one of the stories. If you describe too many properties of the connector, you may accidentally suggest the second story early, which collapses the first story before the reveal and kills the effect. If you fail to build the first story convincingly enough, the reveal has nothing to collapse and lands with reduced impact.
The patter principle I derived from this: build the first story as compellingly and specifically as possible. Give it detail. Give it internal logic. Make the audience’s mental model of what’s happening vivid and seemingly complete. The more convincing the first story, the more satisfying its collapse.
Applying the Assumption Stage
The middle stage — the assumption — is where I find most value in Dean’s analysis. The assumption is not stated explicitly. It’s what the audience fills in from context, from social knowledge, from the specific framing of the setup. The setup doesn’t say “assume this.” It just provides information that naturally produces a particular assumption.
In performance, the assumption stage is controlled by framing and emphasis. What you emphasize in the setup shapes what the audience assumes is important. What you downplay shapes what they assume is incidental. The connector needs to be present without being flagged — it needs to be in the scene but not highlighted, so the first story reading feels natural and the second story reading stays hidden.
This is an attention management problem, and it’s the same attention management problem that any effect requires. You are directing the audience’s attention and interpretation simultaneously, keeping the first story vivid while the second story waits.
Writing Patter with the Framework
I now draft patter differently than I used to. The old approach was roughly: describe what’s happening, add some character, make it interesting. The new approach is structured around the two-story question: what is the first story this patter creates? What assumption is being built? Where is the connector? And does the reveal properly collapse the first story by revealing the second?
For mentalism in particular — where the effect is often about what someone is thinking, and the patter builds an elaborate frame around the moment of revelation — this structure is directly useful. The first story says: I am reading behavioral and psychological cues, looking for tells, interpreting signals. The connector is the spectator’s thought, which the audience assumes I’m deducing. The second story, which the audience arrives at after the effect, is constructed by each spectator according to what they believe. For some it becomes: maybe he really can do this. For others: I genuinely don’t understand how that happened.
In both cases, the first story has collapsed. That’s the effect.
The Comedy-Magic Parallel
The reason I find Dean’s framework so useful is precisely because of this structural overlap. Comedy and magic are both in the business of managing what audiences assume so that a reveal can produce a specific cognitive shift. The differences are in register, in target, and in the nature of the second story. But the architecture is recognizably the same.
Understanding this has made me more attentive to comedy as a source of craft knowledge for magic. The two traditions are more related than their practitioners typically acknowledge. Comedians have developed highly precise vocabulary for talking about setup, assumption, and reveal. That vocabulary is almost directly transferable to discussing effect design and patter construction.
The three-stage model is the kind of framework that, once you see it, you can’t unsee. I now parse every effect I perform through the two-story lens. Setup. Assumption. Reveal. What first story am I building? Where is the connector? When the second story arrives, will the first story have been strong enough to make its collapse feel like something?