The Danger of Nothing

In May of 2014, on the final leg of a trip from Edmonton to Milan, I opened my laptop and typed out a “manifesto” on the art of improvisational theatre—theatre that is created and performed at the same time. I was traveling to present research at a biomedical engineering conference, but my mind was elsewhere.

I recently excavated the manifesto. It’s long, it’s rough, and it’s beautifully messy. At the time, I worried I might be over-intellectualizing comedy. But looking back from where I sit now—-working with generative models for creativity—-I realize I wasn’t imposing artificial, unnecessary order on the art. I was trying to figure out how creative engines work.

I was trying to find order in chaos.

The Paradox of Safety #

One of the first sections of the manifesto is a dissection of a favourite theatre warm-up game of mine called Knee-Freeze Tag. The mechanics are ridiculous: you touch a knee with your hand to freeze someone; you touch their knee with your knee to unfreeze them.

It’s a fun game, try it sometime! I was more interested in the systemic failure of “winning”. A common pattern I would see when teaching the game is that players would try to win by freezing everyone. But by winning, we lose. If you “win” the game by freezing everyone, the game is over. The movement stops. To keep the game alive, you have to actively work to unfreeze the other players, a choice that inherently forces you to put yourself back in danger.

That right there? That’s the secret to generative creativity.

Just like in the game, a generative system dies if it seeks total certainty. If a model only predicts the most “correct” and safe next token, the output freezes, it becomes repetitive and flat. To generate something novel, the system (or the player) must accept the risk of the “unfreeze.” You have to walk the line to know where the edge is.

I wrote a story about this to illustrate the point. An improviser I know well went parachuting one day… as he fell out of the plane, his parachute failed to deploy. He didn’t panic. Instead, he started pirouetting. He accepted the reality of the fall so fully that a flying V of Canadian geese swooped in under him to cushion his landing with the air pressure of their wings. He didn’t fail at skydiving; he succeeded at falling.

The Service of the Scene #

I spent pages trying to deconstruct what makes for a great improvisor. In my lcasses people would point to the funniest, fastest, loudest person in the room. My notes argue the opposite.

For me, the best improvisor is the person everyone else wants to work with. They know how to set the ball on the tee so that anyone else on their team can hit a home run.

I explored this thinking in a show I directed called 15 Minutes of Fame. I would pull a complete novice from the audience onto the stage. The goal was to prove that if I did my job correctly–—if I acted as the enabler, the support, the floor underneath them–—then this untrained stranger would look like a genius.

I gave them only three rules:

• Say Yes.
• Don’t ask questions; make decisions.
• Care passionately about whatever you are doing.

But to pull this off, they inevitably have to get out of their own heads. This requires a specific type of focus. And it oftne took a few minutes to get there.

I called it The Collective Audi-Eye. And I often explained it breifly as follows: When you look down, you tend to think about the past. When you look up, you ruminate or plan the future. But when you look at each other, or out into the into the Audi-Eye, we stay present in the moment. In the here and now. We close the feedback loop. We stop performing at the audience and start performing with them.

The Geometry of Story #

Even in 2014, I was resistant to the standard “Hero’s Journey” or “Conflict/Resolution” model of narrative. They felt forced. Instead, I stopped looking at theatre scenes as story beats and started seeing them as vectors. Direction. Magnitude. Velocity.

Stories are geometrical. You have a Platform (Point A: the status quo), and you have a Tilt (Point B: the deflection). By viewing the classic Keith Johnstone concept of the “Tilt” as a geometric vector, we stop guessing the destination and start plotting the trajectory. It makes narrative actionable because we aren’t looking for a “good idea”; we are simply calculating the arc created when the Tilt disrupts the Platform.

And then comes Point C: the New Normal. The Tilt doesn’t just disrupt—it resolves. The characters must find their footing in a world that has shifted. This is where the geometry completes itself: the vector from A to B creates momentum, and the landing at C gives it meaning. Without the New Normal, you have chaos. With it, you have a story.

Hacking the Hardware #

To teach this, I used constraints that bypassed the intellect. I didn’t want actors thinking about “building a character.” I wanted them to arrive in a body and a voice with a feeling.

I would push them to quickly spin around 360 degrees and land immediately with a new Physicality, Vocalization, and Emotion. And we would do it over and over again.

Then, after a while of solo work, we would add a second person. They have to make a choice. Right now. Are they a Joiner? An Opposer? Or an Enabler? They have to make this choice instantly, and with their own physicality, vocalization, and emotion. These two characters would then exist together.

They exist in this state until the Tilt arrives. A tilt always starts small, but eventually, it won’t leave you alone. The universe shifts. We see the two characters we know affected by this new reality, and they must work to establish a New Normal.

The Grandmaster Effect #

One metaphor that holds up 12 years later is where I compared expertise in improvisation to expertise in chess or Brazilian jiu-jitsu. Chess grandmasters and martial arts black belts don’t memorize every possible move; the search space is infinite.

Instead, they recognize shapes. They see patterns. A chess player sees the board is “Queen-side heavy”. A fighter feels the opponent’s momentum leaning forward. They aren’t reacting to a specific move; they are reacting to a probability distribution.

The trained improvisational performer is one who can recognize these shapes. They appreciate the danger of nothing. They can mine the gold from any suggestion, just by recognizing the patterns, adapting, and justifying the new reality.

The Pivot to 2025 #

When I look back at my manifesto from 12 years ago, knowing what I know about AI now, and seeing how the field has developed and evolved, the connection isn’t subtle. These weren’t just stage rules; they were algorithms for navigating complexity.

I look at generative AI now, and I see an improviser. A large language model doesn’t “know” facts; it recognizes the shape of the prompt and predicts the trajectory of the completion. It treats a prompt exactly like an improv offer: it analyzes the geometry of the preceding tokens and predicts the most statistically probable “yes, and…”.

Day to day, as a researcher I see more clearly than ever the “Alley-Oop.” “enabling” has some negative connotations, but I think that in research, and in many creative pursuits, an enabler is critical.

If I fight to be the most interesting person in the room, I have failed. My job is to be the most interested. My job is to validate the “tilt” offered by a colleague so the project can evolve in ways I couldn’t have predicted. If we can all work this way, then comllective is greater than the sum of the parts.

And as I look at new products and innovations, and I see the A-B-C vector. If you cannot define the “tilt”—–the precise variable you are disrupting in the market—–you don’t have a product. You just have a slightly noisier normal.

The Amalgamation #

I ended those notes on the train with a sentiment that has become my anchor. I wrote: “Your creativity is a beautiful amalgamation of everything you’ve experienced”.

This is our moat. This is what makes us unique. Generative AI has plenty of training data, plenty of experience, but it doesn’t have your specific amalgamation of train rides, failures, and moments of joy. It doesn’t carry your baggage.

No need to be perfect. Just look for the patterns in the chaos. Accept and expand. Find the gold hidden inside the coal.