Infinite Improvisers Theorem
What you can learn from watching absurd amounts of online improv
The infinite monkey theorem states that, given an infinite amount of time and infinite number of monkeys banging on an infinite number of typewriter, one of them will accidentally create the complete works of William Shakespeare (or, for Eastern reference, Mahabharata.)
This is why I enjoy watching online improv.
Monkeys, in this poor analogy, being beginner improvisers.
The infinite improvisers theorem, I define, states that given an infinite number of beginner players broadcasting online shows, one of them will create an artful piece which will remain in my mind for a long, long time.
And they don’t have to be expert improvisers. In fact, I learn more from watching beginner improvisers.
So… in the beginning of the pandemic, I strictly watch the “expert” improvisers and it was super helpful. You know, when you watch pros at work, you try to mimic what they’re doing. After a while, you kinda figure out what techniques they are using at the moment, like “Oh she’s framing the Game now”… “They is making a strong character declaration,”… “He is adding specificity to create metaphors.” All good behaviors. All good improv habits. And all the moves are super-economical.
Then, after a while, I enjoy more watching beginners.
When beginner improvisers play, they can be flailing in many directions. It’s like their arms go here and there and legs kicking and breaking stuff. It can be WILD. Imagine an unsafe yoga posture — while pros know how far their limbs can bend and stretch, beginners might just injure themselves from very awkward postures.
But it’s the awkward postures that is educational for me. Because sometimes they create an injury and sometimes not. Sometimes the scene crashes and burns, sometimes they float. Sometimes it reaches places where it shouldn’t work, but it did. This makes me do introspection of the improv ways that I know, and challenge my beliefs.
I don’t get this as much from teaching / coaching. During active role as teacher I am constantly thinking of good things and bad things I see, from what I know, in preparation of notes. Maybe a few times sidecoaching. Whereas, watching online scenes I can do nothing but follow the scene and see what pops out in the aftermath. When I’m watching with no teaching obligations, I don’t have to respond or make judgement. I can purely listen.
About machine learning, and Alpha Zero
The best thing about online improv is how much scenes there are. Before, I could get 8 episodes of TJ & Dave on Vimeo, and dissect it 50 times. These days there is massive amounts of scenes, from good to horrible. And I can skim through bad shows. In the old in-person days, when I go to a theater and hate a bad show, I have to sit there for 20 minutes til it’s over and amuse myself by pondering how the theatre’s color spotlights work. Here I’ll watch a bad show for a bit until I get the idea, then I can just cut my losses, move to another channel and find more shows to watch. Get more data. This is the era of Big Data in improv.
Time for a little detour to my field of science, Machine Learning (ML) or Artificial Intelligence (AI). So in a nutshell, ML/AI is a field where you chuck in big data and let the computer make sense of it. In early computing science, you give a set of rules to the computer and they will compute for you. In ML/AI you say, fuck it, figure out the rules yourself.
And this they found out, when they first taught the computer how to defeat humans in games of Chess or Go (starting with Deep Blue and Garry Kasparov, leading to today’s Alpha Zero). Deep Blue in 1996 was the first time a computer narrowly edged the world’s best human player. Today, Alpha Zero pretty much slaughters humans like insects.
The first generation, they code in expert knowledge, as guideline for the student (=the computer) for what is good Chess or Go wisdom. But then the next generation of Alpha Zero said, fuck it, we are not even giving you a chess opening book, figure it out yourself. And it turns out to be light years better than the predecessor! Because they are now processing every possible scenario, even the most unlikely according to expert analysts. They learn from good games and bad games.
Don’t be a snob
What’s moral message here? Well maybe there isn’t any. But like the Artificial Intelligence story, to go beyond a certain point you need to watch the bad games as well. Not just the veteran improvisers but also the beginner improvisers. Because that’s when you start reflecting about what you know, and whether that’s valid or not. I guess the moral message is, don’t be a snob and only learn from the very best improvisers.
But the price? TIME. Big data takes an absurd amount of computing time. The reason Alpha Zero was not invented earlier was they didn’t have enough computing power. Likewise, you would need a lot of computing power watching infinite scenes to get Shakespeare/Mahabharata. When you watch the best you can progress very very fast at first — like you do, if you follow specific programs at improv schools. When you open up to the rest of the world, you need infinite monkeys and infinite time in your brain. It’s your choice neither is right or wrong.
I guess the moral message is, when you’re learning, give yourself checks to deconstruct your beliefs and rebuild your knowledge again.