One of the things that makes human beings special is our ability to get emotionally invested in completely imaginary circumstances. It's why we cry when Wilson floats away in Cast Away, or why your partner gets mad about something you did in their dream. Our imaginations are powerful things.
In improv we push this immersive ability to its limits. We (usually) have no costumes, no set, no objects, no music - none of the advantages that movies and traditional theater have to help the audience immerse themselves in the world we're presenting. We hit the stage with nothing but our bodies, minds, emotions, and the hope that if we commit hard enough to our choices the audience will fill in all the blanks. Luckily, they usually do.
But what happens when they don't?
What happens when, 5-8 minutes into our set, all the "good will" laughs have died off and the audience stops responding vocally? We find ourselves doing several scenes in a row with maybe one or two light laughs each. "Uh oh..." we think. "This is going poorly." And we're right.
And we panic.
And we need to do something - anything - to get a BIG laugh.
And we notice our scene partner's object work isn't quite as precise as it could be when we hand them a can of beer and they drink it without opening it.
And that's our chance.
"How are you even drinking that beer? You didn't even open it!"
And the audience laughs. And we relax.
We saved the show.
Or did we?
This is "going meta" - commenting on the scene you're currently inside of in a way that reminds the audience that they're a bunch of people in a room watching people make stuff up.
We do it because it works. We needed a big laugh and we got one. But at what expense? Now the audience remembers who they are. They remember they're watching a show. They remember they paid $12 to watch adults make funny faces and they're trying to decide if it was worth it. We've destroyed the immersion. We've torn down everything we spent the last 5-8 minutes building. With one line. Because we panicked. And because WE needed that laugh.
Going meta is a selfish act.
It doesn't serve the scene because it destroys the immersion. It doesn't serve the audience because it makes them self-aware. It doesn't serve your scene partner because (in this example) you're calling them out for sloppy object work. Going meta only serves the needs of the person who panics and needs the instant positive feedback of a big laugh.
This one's for metaphor fans:
Immersion functions like the Pensieve in Harry Potter.
In the Harry Potter universe, the Pensieve is an object where you can store and review memories. Here's a clip from The Goblet of Fire where Harry discovers the Pensieve for the first time:
There are two main takeaways here. 1 - David Tennant has a very active tongue. 2 - When Harry is viewing memories in the Pensieve, he's not just watching them on the surface of the water. He is literally in the room. He is experiencing the memory as if he was there himself. Though he can't interact with them, the people seem real. Though he can't influence the events of the memory, he feels involved. Look at his face as the trial unfolds. He is concerned, uncertain, and most importantly, fully invested.
A fully immersed improv audience will feel the same way. They are literally in the room, and though they can't (or shouldn't) interact directly with the performers, they will feel involved and invested as long as long as we continue to fully commit to the worlds and characters we are building.
But how would Harry's experience change if Dumbledore suddenly turned to him and reminded him that he wasn't actually in the courtroom? That he was actually face down in a bowl of magic water.
He would feel self-conscious and probably a little silly. He would be distracted, and he wouldn't be able to focus his full attention on the memory. He'd probably miss something. Going meta has the same effect on an audience. It reminds them that what they're watching isn't real, and when things aren't real they stop being important.
Now here's the part where I say "but":
It is entirely possible to comment on a scene in a way that does not destroy the immersion.
But it's tricky. When something is a little off or we blatantly break the reality of the world, the audience will recognize that. It's important for us to acknowledge that we see what they saw. They need to know that we're in total control of what's happening - or at least more than they are.
So how do we acknowledge a "mistake" without going meta?
Justify from within.
Find a way to make it work by making it a part of the reality of the scene. Pointing at it and saying "that's weird" is simply not enough and usually, for the reasons I explained above, counterproductive.
In the Pensieve scene, when Harry plops down next to Dumbledore, he is confused why his professor is ignoring him. It is not until someone passes a hand through him that he understands the rules of the memory world - he can watch but he can't interact. The Pensieve justifies from within.
In a recent Oregon Fail show I played a federal detective who was attempting to arrest someone for murder. As I did, I found myself saying the line "I'm going to place you under arrest under the order of our current President..." In that moment I recognized that I was in trouble. Oregon Fail takes place in the year 1848 and at the time I had no idea who was President in 1848 (it's James K Polk). Recognizing that my character would have to know the name of the current President, and recognizing that the audience would expect that too, I justified from within. I finished with "...whose name I will not mention out of respect." It fit my character, it fit the world, and it was a subtle wink to the audience that let them know that I saw what they saw without ruining the immersion.
The beer can example is no different. If, instead of calling out our scene partner for sloppy object work, we had made it fit the scene by saying something like "you really have to teach me how you're able to open these with your teeth", we have acknowledged the "mistake" without breaking the immersion. We've also given our scene partner the gift of a character ability instead of making them look bad. It clears all hurdles. Justify from within.
So what do we do if we find ourselves struggling at the 5-8 minute mark? We've done several scenes in a row and the audience just isn't responding. What do we do when that urge to go meta starts creeping in? Don't pull back. Don't rip Harry from the Pensieve. Push him deeper in. Commit harder. Make what's happening matter more. Believe it more. If it matters to you it will matter to the audience. If you believe it, so will they.