This weekend we’ll be attending the a live version of the New Yorker‘s cartoon caption contest, part of the New Yorker Festival. The event has been sold out for weeks, proof that the caption contest is extremely popular.
The magazine’s weekly caption contest is the biggest of its kind in the world. Each week, thousands of readers submit potential captions for the cartoon on the last page of the magazine. The winner gets his or her name in The New Yorker and receives an autographed print of the cartoon (retail value $250). The game has even drawn in (and stumped) celebrities like Zach Galifianakis.
But with all those vying against each other, is there a surefire way to win? Is there a code behind the contest that, if not guaranteeing a win, will at least get you a better shot? Is it all about boasting the right mixture of talent, genius, and luck, or are there tricks to employ to improve your odds?
Pete got to talking to Bob Mankoff, Cartoon Editor for The New Yorker, about this during a chance meeting at a recent humor conference. Bob agreed to pass along the captions submitted for contest #281, which featured movie critic Roger Ebert submitting the winning entry, and Pete enlisted Phil Fernbach, a cognitive scientist from Brown University an expert at using quantitative techniques to conduct the analysis.
And when we were drone, we gave Mankoff a chance to
criticize analyze our results. You can find his comments at the bottom.
First place: “I’m not going to say the word I’m thinking of.”
Second place: “Why don’t you just admit it—this is the wrong desert.”
Third place: “Let me know if you spot a divorce lawyer.”
Procedure and results:
Phil categorized the data into groups: 1) a shortlist of 43 captions that were under consideration to be finalists by the editors (which included the three finalists), and 2) the rest. A research assistant then judged a subset of 86 captions (the 43 on the shortlist and 43 selected randomly from “the rest”) on abstractness (from ‘very concrete’ to ‘very abstract’) and imagery (from ‘very hard’ to imagine to ‘very easy’ to imagine; all seven-point scales).
Based on a statistical and textual analyses on the resulting database, what should you pay attention to if you want to win the contest (at least #281)?
1. Novelty: Captions that use words that are uncommon in other captions are more likely to make the shortlist.
Phil removed non-content words, like “the,” and scored the remaining words in the database by dividing the frequency of each word by the average frequency of all words, thus creating a score that indicated the frequency of each word relative to all others. For example, the word “parked” earns a high-frequency score because it shows up in many of the submitted captions. Then for each caption he calculated the average score for all the words in the caption.
When the short-listed captions were compared to the rest, the short-listed captions were much more novel. For example, a caption that read, “I told you we parked in E” didn’t make the short list, because “told”, “parked” and “E” were all common words in the database. In other words, the joke is too obvious. Here’s a caption that is very novel according to the measure: “We’ll never find enlightenment carrying all this stuff.” This is a different take on the cartoon, using words that do not appear frequently in other captions. Apparently it caught the eye of the judges because it made the shortlist.
The finalists also receive high scores for novelty, avoiding (for the most part) the use of obvious words or ideas. They were not the most novel on the short list but about average novelty for the shortlist.
A friend of Phil’s who has entered the competition for years says that he tries “to figure out what the editors are looking for.” Anecdotal evidence suggests that this is the right approach; the caption, “I want a divorce,” was submitted seven times and made the shortlist. The overall analysis, however, suggests this is the wrong approach. Instead, try to create something that other people have not thought about.
One strategy is to resist the tendency to state the obvious. Said another way, ignore your first instinct. A second strategy, which is common to comedy writing, is to create a lot of captions and then choose the more novel from the larger set.
2. Length: Captions that use fewer words were more likely to make the shortlist.
Phil counted the number of words in each caption and found that the short-listed captions were on average shorter by one word than the others (8.7 vs. 9.6, respectively). The longest caption on the shortlist was 16 words; the shortest was 4 words. The strategy, however, doesn’t seem to always work; the finalists, for example, are slightly longer than average (mean = 10.0).
In short, although “brevity is the soul of wit,” it is not necessary to make the finals.
3. Punctuation: Captions that avoid exclamation points, commas and question marks are more likely to make the short list.
An examination of punctuation is striking (especially for exclamation points), as shown below (which presents the percentages of captions in the category that contain the different forms of punctuation).
|Question Mark||Comma||Exclamation Point|
The finalists do not use exclamation points, commas, and question marks – even when they could (i.e., second place could have been phrased, “Why don’t you just admit it? This is the wrong desert!”)
Why are captions with punctuation worse? Perhaps people use punctuation to take the place of cleverness. An exclamation point says “pay attention to me”. A good caption draws attention with its wit. A different possibility is that punctuation obscures a good caption. An exclamation might draw attention away from a funny joke; a comma or exclamation might break up our flow as we read the caption making it seem inelegant. In psychology this is called a “disfluency” effect.
4. Abstractness and imaginability: Captions that are hard to visualize are more likely to make the shortlist.
Captions that cannot be experienced by the senses are more likely to make the short list (mean = 5.5) than the rest (mean =5.0; out of seven, where seven means more abstract). The finalists scored higher than average on abstractness (mean = 6.0). The imagery effect was smaller but in the same direction. Shortlist captions were harder to picture mentally.
This was somewhat surprising to us. A lot of successful stand-up comedy elicits humor by bringing to mind vivid concrete descriptions of events and anecdotes. On the other hand, the New Yorker caption contest is not a test of your stand-up comedy prowess.
Moreover, concrete caption words often refer to things that are actually visible in the picture. A lot of people mention them and thus avoiding them leads to novelty. This suggests that concrete words, when used, should not refer to things that are actually in the picture. Said another way, don’t waste words describing thing that are shown in the picture.
How to win: Try for subtlety, elegance and originality. And definitely do not use exclamation points!
First off, let me say that I am shocked–shocked, by the results. When I went to cartoon college I was taught that long, heavily punctuated, commonplace captions were the key to success.
Irony aside, I do want to thank Pete and Phil for doing this. While the results are, in fact, not shocking at all, a lot of useful science is just showing that what you think is true, is true. And sometimes, in doing that you find out as Mark Twain said that, “It ain’t what you don’t know that gets you into trouble. It’s what you know for sure that just ain’t so.”
Ignoring Twain’s admonition, I can say with certainty that we shouldn’t compromise our research ethics by generalizing the results beyond this particular caption contest to all the other 283 contests, or, for that matter, to non-caption contest New Yorker cartoons, or to humor in general. That would be a grave mistake which would only be justified if we could make a lot of money out of it.
Some other thoughts on novelty, punctuation, caption length, and, of course, how to win.
Novelty. This makes sense to a point. The best winners, for me, are the ones that are relatively uncommon without being so unusual that they don’t resonate with most viewers. Remember, the finalists were not the most novel on the short list. And there have been contests in which the identical caption that won was submitted by hundreds of entrants. This was the case for contest 147 shown here:
In that contest 37% of the captions submitted used the word “objection” and 24% the phrase “killer whale”.
Punctuation. I generally agree that shouting a joke doesn’t make it any funnier unless it’s to a deaf person. But the exclamation point in caption contest 147 goes perfectly with setting. As do these question marks in caption contest 145:
Caption length. Yes, “brevity is the soul of wit.” Also twitter. Although at 140 characters, 23 words, a maximum tweet would have minimum chance of being a short listed for the caption contest. Captions trend short for the contest because the caption is, in effect, the punch line for which the image is the set up. So most captions, good, bad and indifferent are relatively short.
My own calculations for the median value of contest 281 shows it to be 9 words. The maximum length permitted by the caption contest is 200 characters. And there was only one in contest 281 which even came close to that, at 195 characters, and which actually would have been helped by punctuation, but only marginally because of its lameness:
“What does it mean God knows our car was parked in Row B Maybe it’s a guidepost for feathered Flickers No wait better yet it means we’re f— four ways to China Did you buy the bottled water”
So definitely avoid captions with 195 characters. Also lameness.
How to win: C’mon. You wouldn’t want me to take all the fun out of it would you?
Don’t answer that, Pete.