Not all GIFs are created equal. Some, for one reason or another, make it into the pantheon of viral superstars and are shared by everyone, while others sink to the sad last page of Reddit. In one of the first studies of why humans share more pictures than others, researchers found that we tend to share vertical, animated, color images with some text. Italian researchers Marco Guerini, Jacopo Staiano (University of Trento) and Davide Albanese presented a paper on arXiv, titled "Exploring Image Virality in Google Plus." Why they chose Google+, a social network that isn't as popular over Twitter and Facebook or Reddit is a bit puzzling (and probably more manageable), but their study, as the MIT Technology Review points out, is one of the first on the virality of images.
The Italian researchers studied various things about images people shared, like orientation (horizontal vs. vertical), color (grayscale vs. color), animation, and text in 289,434 posts. And they found that pictures in color, vertical, have text and are animated were more viral than the opposite. They also found that there's a difference between someone re-sharing an image and someone "liking" it. "In particular, funny and informative images have much higher probability of being re-shared but are associated to different image features (animation and high-brightness respectively), while colored images or images containing faces have higher probability of being appreciated and commented," they found.
Those findings aren't unique to Google+. Just look at the way Reddit works. "After being sent mixed messages from my crush for a year" reads the intro to one of today's more popular posts (over 8,000 votes), an example of text. And if you click on that link, you get this color, vertical, image with some text:
Or this post (over 30,000 votes):
And if all three had some animation aspect, they might be more viral. "[T]here is clearly work ahead. Guerini and co say they next want to look at factors such as the composition of a picture and its content using scene/object recognition" the MIT Technology Review explains. Wait. Sorry, I was busy sharing that dalmatian.