Should data scientists <3 emojis?

Within a sea of New York start-ups hides a data journalism lab that treats the heart tacked on at the end of your tweet with the same seriousness as its 140 characters. PRISMOJI, founded by ex-Facebook data scientist Hamdan Azhar, hunts and reports on patterns in tweet trends. From the Swift/Kanye debacle to Brexit, the lab has substantially weighed in on our ever-growing emoji footprint with an ardent sureness in its importance. Huffington Post featured Azhar’s story on the most popular emojis in anti-Trump tweets, where he suggested the “fist” emoji might be a similar rallying label as the anti-Trumpist hashtags. And Azhar clearly believes in this work whole-heartedly. He shared with Vice that he hopes, “people start taking this seriously. Just as with any new language, learning the alphabet, we can eventually do more complicated, nuanced things.”

And people are: in 2015, the “crying smiling face emoji” was Merriam Webster’s word of the year. And, technically, emoji death threats are a legitimate claim – though a case has yet to hit the media, a person can actually be charged for threatening someone by way of emojis. Foreign as the thought might be, somehow emojis have caught up as a perfectly valid form of communication. Even sources like Hillary Clinton have asked people to “respond with three emojis describing student debt.” Is this all just a way to reach younger people? Maybe, but most of us aren’t strangers to tacking on a smiley face to make sure a friend gets the right idea from a quick message. Emojis are a widely accepted form to express precisely the way someone feels, sometimes managing to communicate emotion lost in rote text.

This is precisely what makes emojis so powerful. The ambiguity of a typical message can be risky when sensitive information is on the line. The distinction between sincerity and sarcasm is an extremely fragile one, yet a single picture can make the difference immediately obvious. A study from 2014 found that emoticon (the keyboard equivalent to an emoji, where ‘😊’ is ‘:-)’) actually is perceived by people in the same way as an actual person’s face would be. Maybe seeing an emoji communicates just as much as seeing a split-second glance from another person. A new form of image based language has been created and, without much effort on our part, has trained us to think in a completely different way.

It does all sound incredibly appealing, but it just might be a little too good to be true. The same article on Huffington Post talking about Azhar mentions that the smiling crying emoji could be used both as “tears of joy” and “tears of ridicule.” That has a huge impact on whatever sentiment analysis leaning on this tool. If data scientists turn to emojis for clarification when encountering potentially sarcastic text, for instance, the images can’t be used to say much about the message when their own meaning is unsure. In reality, these images may carry a comparable level of uncertainty to the text. And, while emoji usage is heavily skewed to American users, the meaning of a winky face is further complicated by the international understanding of what each picture means. In most places outside the U.S., a peach is often just a peach.

Though cheesy, it’s becoming more and more true: a picture is worth a thousand words. The problem is, are we any closer to understanding what those words actually might be?