Every definition on Urban Dictionary was written by a user like you. About a thousand definitions are sent to Urban Dictionary every day, but only half enter the hallowed ranks beside ratchet, Harlem Shake, and Cleveland Steamer. People often email and ask, why didn’t my definition get published? Well, we’d like to know too.
The truth is, we don’t decide. Sometimes we’re just as puzzled. Some rejected definitions seem perfectly good to us. Sometimes we wouldn’t vote for a definition, but we can’t really say why. Urban Dictionary’s users get to edit the content, so they get to decide. However, since this is a frequent question, we thought we’d look into how users make the decision to publish or not publish a particular definition.
We wanted to understand what matters in a definition, what helps it get accepted. To do this, we started by asking users to just describe a list of eight random submissions. We gave people one line of text to describe the submission. For example, one of the words was “phil collins.”
People described this definition as “Somewhat biased”, “opinionated”, or “NO! PHIL COLLINS IS THE SHIT”. We collected 69 responses for each word, grouped them together and tried to find the main idea in the description. We continued to group things together until we ended up with eight categories.
clearly and correctly written
complicated/too long
bad grammar
capitalized/formatting
confused
specificity
personal
too specific
about a celebrity
slander
biased
opinionated
offensive
degrading
slander
sexist
racist
</td> <td> usage
redundant
original
makes sense
helpful
lame
strange
weird
quality of the definition
informative
true
wrong
accurate
stupid/dumb
graphic
disgusting
visceral
dirty
provocative
sexual
improper
</td> </tr></table>
We took these categories and turned them into possible descriptions of definitions. We edited some of the category names so that they roughly corresponded to a fifth grade reading level, for example, “graphic” became “nasty”. We then asked people editing Urban Dictionary to rate fifty submissions (25 that had been accepted, 25 rejected) on these descriptors. Each person got to rate only one word, one time. We also recorded the length of the definition as a possible explanation for why a word got published.
Using these data, we ran a logistic regression using each descriptor and definition length as predictors and whether the definition was accepted or not as the dependent variable. This let us know how much each descriptor mattered relative to the others for users deciding to publish or not publish a particular submission. To do this, we looked at the odds-ratio, or the likelihood that a submission will be published given that it is, say, funny.
(In the graph above, we subtracted 1 from each odds-ratio so a value of 0 would mean that the descriptor didn’t matter. A high positive value means a definition is more likely to be accepted.)
People love a good joke, and Urban Dictionary’s editors are not an exception. For every point of funny, a submission is 1.68 times more likely to be published. They also want to read definitions that are useful – things that might be spoken in actual conversation – and that mean something. They don’t like definitions that are too specific, even if they are about prog rock drummers, or ones that are too nasty. For every point of nasty, a submission is 0.80 times as likely to get published. Urban Dictionary’s volunteer editors are a worldly bunch, unlikely to be prudish about sexual or offensive definitions. So when you are writing your next submission, keep the audience in mind – this is how they decide whether to publish your definition.