E.B. White once likened analyzing humor to dissecting a frog, in that the frog dies and that most observers are grossed out by the process. But who will object if we dissect Dane Cook? Here's a statistical look at how two very different comedians elicit reactions from their audience.
Given the risk of taking the fun out of fun, I made sure to choose two performers with unshakeable comedic consensuses: Louis C.K. and Dane Cook. A little math won't spoil C.K., one of the best ever, while most readers won't be able to like Cook any less than they already do. The pair made headlines recently after Cook went on Louie and discussed accusations of joke theft. Though the two differ in style and quality, they share a similar, impressive command of their audiences. C.K. handles a crowd through impeccable rhythm and timing; Cook does it by apparently busing in his friends and extended family. Even the most appreciative audience shouldn't have unconditional love for the comedian, and as a result he's the only stand-up act who'd be more palatable delivering his jokes to silence. If you need proof or are out of ipecac, watch him at Madison Square Garden.
Curious to see how this difference would appear numerically, I took a stopwatch and charted the durations of audience laughter, as well as the duration between the laughter, on two of their live stand-up albums: Cook's "Isolated Incident" and C.K.'s "Hilarious." There are too many other variables (crowd size, availability of alcohol, editing, etc.) to completely isolate the comedians themselves, but we might get some valuable clues. Here are the basic stats:
Louis C.K.: 426 instances of laughter (4.7 per minute) with an average duration of 5.05 seconds. The audience was laughing 40.1 percent of the time. Gaps in laughter lasted an average of 5.24 seconds.
Dane Cook: 300 instances of laughter (5.7 per minute) with an average duration of 4.26 seconds. The audience was laughing 40.4 percent of the time. Gaps in laughter lasted lasted an average of 6.15 seconds.
It's striking how close the percentages are: despite different frequencies and average durations of laughter, the total duration is nearly identical. It's impossible to figure out from just two acts, but this might suggest a universal ceiling: can stand-up comedians get their audience to laugh significantly more than 40 percent of the time, and would the performers even want them to? Regardless, the difference in the frequency and duration of is worth looking into some more. The following graphs show the distribution in the duration of laughter and audience silence (click to expand).
In all four cases, a majority of the instances occurred on the left side of the graph. These distributions are all "log-normal" — that is, they look like a bell curve when distributed by orders of magnitude instead of their raw value. It comes up a lot in economics. Laughter is distributed a lot like income in that most laughs are middle-class, lasting only a few seconds, while the punchline-induced roars are the billionaires, enough to move the average way up despite their low frequency. The gaps in laughter work in reverse: long pauses occur only at the beginning of a routine as the comedian introduces the joke.
Just because the durations follow the same type of distribution, however, doesn't mean that they're the same. Does C.K.'s laughter follow the same pattern as Cook's? Are laughter and gap duration significantly different? Using a technique called the K-S test, I found the answers.
When the audience isn't laughing, Louis C.K. and Dane Cook look the same. The gaps between laughter have a similar distribution for both comedians.
When the audience does laugh, the two performers diverge. Louis C.K.'s audience laughs with a different frequency and duration than Dane Cook's. And, specifically, the audience laughs for Louis C.K. at different intervals than it listens quietly. When he gets the audience laughing, he appears to move it into a different state of behavior.
Dane Cook's audience shows no such change. The statistical profile of the crowd's laughter is the same as the statistical profile of its waiting to laugh. The difference between setup and punchline is like flipping a coin.
There's not a lot of information to get from just two performances (especially without an easy way to measure decibel level), but the above data raises questions about the relationship between successful comedians and their audience. Are critically acclaimed comics more similar? What about acts in the same genre? It might be possible to establish "footprints" of comics, venues, and crowds with more research. But that would be both reductive and difficult, and Comedy Week is almost over. So fuck it. Statistics says Dane Cook is a robot.