About a year into performing at corporate events, I started noticing a discrepancy that confused me.
People’s descriptions of the show didn’t match what I’d done.
Not wildly — nothing invented from whole cloth. But details would be wrong, sequences would be reordered, moments they described vividly hadn’t been quite as they described them. And consistently, descriptions tended to be better than the reality. Smoother. More coherent. The moments that had slightly stumbled in my experience of them were described as perfectly landing.
For a while I thought this was just polite memory. People are kind. They smooth over the rough patches when they retell things.
But it wasn’t just smoothing. The structure of their retelling was wrong. They weren’t remembering a slightly improved version of what happened; they were remembering a differently sequenced, differently emphasized version that had its own internal logic — a logic that was often more satisfying than the actual show.
Kahneman calls this the narrative fallacy. And once I understood it, that discrepancy made complete sense.
How Memory Actually Works
Memory isn’t recording. This is the first thing to understand.
When you watch a magic show, your brain isn’t filming it and storing the footage. Your brain is processing a stream of sensory information, selecting what seems important, encoding what gets selected, and then later — when you try to “remember” it — reconstructing the experience from those encoded fragments.
The reconstruction is active and creative. Your brain takes the fragments it stored, fills in the gaps with plausible information, smooths out inconsistencies, and generates a coherent narrative. The narrative feels like a memory. It feels like “what happened.” But it’s actually a reconstruction that your brain just built, right now, using fragments plus inference plus narrative logic.
This is why eyewitness testimony is notoriously unreliable. People remember with complete confidence things that didn’t happen, or didn’t happen the way they remember. They’re not lying. They’re reporting the reconstruction their brain generated, which feels indistinguishable from direct recall.
For magic performers, the implications are strange and wonderful.
The Audience Remembers What Should Have Happened
Here’s the key insight: when your brain reconstructs an experience, it applies narrative logic. Fragments get arranged in a way that makes a coherent story. Gaps get filled with whatever makes the narrative most internally consistent.
If the actual show had a small error — a moment that was slightly rough, a line that didn’t land quite right, a transition that was a bit abrupt — the reconstruction might smooth it out. Not because the audience decides to be charitable. Because narrative logic says: this is what should have happened here, given everything else that happened.
The reconstruction optimizes for coherence. A rough moment that doesn’t fit the overall flow gets softened in memory. A moment that was actually very impressive gets amplified if the narrative logic says this is where the peak should be.
I’ve had conversations after shows where someone described a specific moment with great enthusiasm — the look on their face, the emotional impact, exactly what they felt — and I know for a fact that moment was one of the more uncertain parts of the performance for me. I wasn’t sure it landed. But in their memory, it landed beautifully.
The narrative fallacy wrote the better version.
The Practical Implication for Performance
One thing this tells you: the audience’s experience of your show is partly determined by whether you give their narrative brain good materials to work with.
If your show has a clear arc — a beginning that establishes something, a middle that complicates it, an ending that resolves it in an unexpected way — then narrative reconstruction will do you favors. The brain will fill gaps with story-consistent details. Small stumbles will get smoothed into the arc.
If your show lacks a clear arc — if it’s a series of impressive moments without connecting tissue, without a sense of building toward something — then reconstruction has less to work with. The brain can’t apply strong narrative logic because there’s no strong narrative. Gaps get filled with generic “magic stuff.” The experience is harder to reconstruct coherently, which means it’s harder to remember vividly.
This is the hidden cost of setlist thinking: choosing strong individual effects without designing for arc and connection. Each effect might be excellent. But if the show doesn’t have a narrative spine, the reconstruction will be muddy. People will remember “good magic” without remembering specifics, because their brain couldn’t build a satisfying story from the fragments.
The Error That Helped
I want to be honest about something uncomfortable.
Early in my performing, there were nights when things went genuinely wrong. A moment that should have been strong fell flat. A sequence that was supposed to build to something didn’t build. And then, in the conversations afterward, someone would compliment the show in a way that felt almost irrational given what I knew had just happened.
I used to chalk this up to them not knowing what to look for, or being polite. Now I think something more interesting was happening. They were reconstructing a version of the show that made sense of all the strong moments they’d experienced, and the weak moment — if it didn’t fit the narrative they were building — got edited out or transformed.
The show they remembered was more coherent than the show I’d given them.
This could be dangerous if you use it as an excuse not to improve. “The audience will reconstruct a better show anyway, so why worry?” That’s entirely wrong. The raw materials for reconstruction have to be good. A show full of confused, unconnected, low-impact moments gives the narrative brain nothing useful to work with. Reconstruction makes good shows better. It can’t make bad shows good.
But it does suggest that a few imperfect moments in an otherwise strong, well-structured show won’t damage the memory nearly as much as I used to fear. The narrative brain is doing active repair work, and it’s good at it.
The Volunteer Distortion
There’s a specific version of the narrative fallacy that’s particularly interesting in magic: what volunteers remember about what they did.
When someone participates in a magic effect — holds a card, makes a choice, names something aloud — their later memory of what they did is often subtly different from what actually happened. Not wrong in big ways. But slightly reframed, slightly cleaned up, slightly more coherent.
I’ve noticed volunteers will sometimes describe having “freely chosen” something in a way that makes it sound more free than it was. Not because they’re stupid or because I was crude about it. Because their narrative brain reconstructed the experience as a free choice, which is the version that fits the most coherent story: “I chose freely, the impossible happened, therefore this is genuinely impossible.”
The alternative reconstruction — “something subtly influenced my choice, which then led to the impossible result” — is both less coherent as a story and more unsatisfying emotionally. The narrative brain doesn’t want that version. It reconstructs toward the version that preserves maximum wonder.
Again, this only works if the raw experience was good enough to give the narrative brain something to work with. A crude, obvious approach gives the brain materials that reconstruct into “I was manipulated.” A natural, elegant approach gives materials that reconstruct into “something impossible happened.”
What I Do Differently Now
Understanding the narrative fallacy has changed a few concrete things about how I work.
I invest heavily in the overall arc of a performance. Not just “save a good effect for the end” but genuinely: does this have a beginning, middle, and end? Does it build? Does it resolve in a way that makes the whole experience feel like a complete story?
I pay attention to transition moments. These are the connective tissue that tells the narrative brain what the arc is. If transitions are abrupt or unexplained, reconstruction will have to guess how the pieces fit. If transitions are clean and purposeful, reconstruction has a guide.
And I’ve largely stopped worrying about small imperfections. A slightly awkward line. A moment where the timing wasn’t quite what I wanted. A laugh that didn’t arrive when I expected it. These are not the things that determine how the audience remembers the show. The narrative fallacy will smooth most of them out, as long as the overall structure gives it good bones.
What matters is: does the audience have enough strong fragments and enough narrative logic to reconstruct something that deserves to be remembered?
Give them that, and memory will do the rest.