The first idea you have — when someone says something unexpected on stage, when you’re looking for a line of patter, when a moment calls for a response — is usually better than the one you’ll replace it with after a second’s evaluation.
I resisted this for a long time because it seems to argue against thinking. Against craft. Against the careful considered choices that distinguish professional work from amateur fumbling.
What Keith Johnstone actually argues is more precise than “don’t think.” He’s pointing at a specific mechanism: the internal censor that activates between the first response and the executed response, and the specific kind of damage it does.
What the Censor Is Doing
The internal censor isn’t evaluating quality. It’s evaluating safety. The questions it asks are: is this too obvious? Has this been done before? Will this seem uncreative? Is this weird? Will people laugh at me for this, and not in the way I want?
These are social risk questions. They have almost nothing to do with whether the response is true, whether it fits the moment, whether it will land with the audience.
The first response — the one that comes before the censor activates — is generated by intuition: a rapid integration of everything you know, everything you’ve experienced in this kind of moment, everything your body is picking up from the room. It’s fast and it’s often genuinely accurate. The speed is not a sign of superficiality. It’s the sign of a well-trained implicit system drawing on everything it’s accumulated.
The censor then evaluates this response through the lens of social safety and finds it wanting — too obvious, too simple, too vulnerable. It substitutes a more managed alternative. The managed alternative is less exposed and less true.
Johnstone made this observation through watching literally hundreds of improvisational scenes develop. The pattern was consistent enough that he built teaching exercises specifically aimed at disabling the censor — not because the censor was always wrong, but because the censor’s criteria (social safety, appearance of originality) were categorically different from the criteria that make scenes work (truth, spontaneity, genuine response).
In Performance: The Second-Guess Trap
I can identify several specific forms the second-guess trap has taken in my own performance.
The first: an unexpected volunteer response triggers a genuinely funny or apt first reaction in me, but before I say it, the censor asks whether it’s too predictable or whether it matches the more polished tone I was aiming for. The managed alternative I produce is smoother and less alive.
The second: in the flow of a routine, I sense that the moment calls for something — a beat of silence, or a particular movement, or a simpler phrasing of what I’m about to say. But I override this sense with the scripted version, because I’ve rehearsed the scripted version and it feels safer.
The third: at the end of a performance, when someone asks a question and there’s a response that comes immediately — something genuine, something true to how I actually think about what I do — I smooth it into something more considered. The considered version is technically superior by some metric. It’s also demonstrably less interesting than what came first.
All of these are second-guess failures. All of them produce outcomes that are safer and less alive.
Training the Trust
Johnstone’s approach to building trust in first responses is essentially to create conditions where there’s no time for the censor to operate. The exercises are fast. You don’t have the luxury of evaluation — you have to respond before the censor gets there.
Over time, repeated experience of first responses working builds a kind of trust. The track record of the first response outperforming the managed alternative becomes visible enough that the emotional barrier to trusting it decreases.
I built something similar through deliberate practice. When I’m developing material and I’m looking for a line or a response to a particular moment, I force myself to capture whatever comes first before I assess it. Not type it while assessing — type it without any evaluation, then look at it.
Frequently, looking at the uncensored first response with some distance, it’s good. Better than what I would have built if I’d started filtering immediately. It has a quality of directness that filtered responses lack.
Sometimes it’s not good. But even then, it’s usually the right starting material — raw rather than finished, but pointed in the right direction. The refinement of a genuine first response produces better work than the construction of a managed response from scratch.
On Stage, Specifically
The application in live performance is not that you should say the first thing that comes to mind without any evaluation. It’s subtler.
When you have time — in preparation, in rehearsal, in the construction of material — trust the first idea more. Capture it before the censor can dismiss it. Evaluate it later, from a distance.
When you’re in the moment of live performance and something unexpected happens, the choice is usually between two options: the first response that surfaced immediately, and the managed alternative your evaluating mind is reaching for. In my experience, a very high percentage of the time, the first response is the right one.
The managed alternative feels safer because it’s deliberate and controlled. But the audience doesn’t reward feeling safe — they reward genuine presence. And genuine presence often looks like the first response: direct, unpolished, real.
The Particular Value in Magic
There’s a specific application of this principle in magic performance that I haven’t seen articulated clearly elsewhere.
A magic show is constructed, rehearsed, and to a significant degree scripted. The structure is known. The beats are practiced. This is necessary — the technical execution requires preparation that can’t be improvised.
But the show’s responsiveness to the actual room — the moments of genuine contact with the specific audience in front of you — is not planned. Those moments are improvisational. And they benefit from exactly the kind of first-response trust Johnstone describes.
When a volunteer’s expression tells me something, and I have an immediate response to that expression — a movement, a word, a beat of silence — the question is whether I trust it or filter it. The filter makes the show cleaner but more generic. The trust makes the show less tidy but more real.
The performances I’m most satisfied with afterwards are almost always the ones where I said yes to the first response more often than I usually do. They’re also the performances where unexpected things happened that I couldn’t have planned, and where those unexpected things became the most memorable parts of the night.
The first idea knows something the second idea doesn’t. It knows the room. It knows the moment. It hasn’t been cleaned up into something safer and less true.
Trust it more often than you think you should.
From the spontaneity of performance response to the discipline of practice — Mihaly Csikszentmihalyi’s research on flow offers a completely different but deeply complementary lens. The challenge-skill balance he identifies maps directly onto the structure of deliberate practice, and understanding it changes how you design your sessions.