Clever #Hashtags Confuse Twitter Computers

Social media giant Twitter relies on human workers to make sense of odd hashtags. CREDIT: Twitter image via Shutterstock View full size image

Twitter, the darling of the international social-media set, has blazed a path 140 characters wide across politics, entertainment, sports, media, academics and just about every other industry in the modern world.

Peek behind the curtain of the Twitter dynasty, however, and a surprisingly old-fashioned way of working is revealed: low-paid workers toiling at all hours of the day and night, according to the Daily Telegraph.

Hashtags, the Twitter search terms preceded by a # sign, aren't always easy for computer algorithms to understand — especially since many of them have evolved into exercises in creative writing and are influenced on a minute-by-minute basis by pop culture. In fact, "hashtag" was voted word of the year for 2012 by the 124-year-old American Dialect Society, which includes linguists, lexicographers, historians, writers, editors and others.

Consider the hashtag #bindersfullofwomen, which was born in an instant when presidential candidate Mitt Romney casually uttered the phrase during the presidential debates of 2012.

In a flash, the hashtag rocketed around the globe as fast as political pundits' fingers could type it, making it a fast-trending topic among the Twitterati. But without any previous context, Twitter algorithms couldn't make sense of the phrase.

This hashtag confusion presents a challenge when organizing search terms on Twitter and, perhaps more importantly, selling advertising that's targeted to specific demographics.

"How would you know that #bindersfullofwomen refers to politics, and not office accessories?" wrote Edwin Chen, a Twitter data scientist, and Alpa Jain, an advertising specialist for the firm, on a Twitter engineering blog. "Since these spikes in search queries are so short-lived, there's only a small window of opportunity to learn what they mean."

This is where "low-tech" human beings enter the picture: Twitter relies on what they enigmatically call a "real-time human computation engine" to decipher popular but puzzling hashtags like #bindersfullofwomen. [10 Things That Make Humans Special]

That "computational engine," according to the Telegraph, is a small army of workers around the world who are paid less than $4 per hour to make sense of odd hashtags. These workers are selected from an Amazon service known as Mechanical Turk, reports the Telegraph.

Employees of Mechanical Turk are typically home workers who are wired to the Internet and able to perform mundane tasks that confuse computers, like transcribing human speech.

Though Twitter did not reveal its precise pay scale, the rates on Mechanical Turk are typically low, according to the Telegraph. Transcribing 135 minutes of video, for example, would take over eight hours and pay $33.57 — that breaks down to about $3.73 per hour.

This isn't the first time social media has been accused of being somewhat elitist: Researchers from North Carolina State University found that informal social networks such as Facebook and LinkedIn play important roles in finding jobs everywhere. But in the United States, those networks are much more important for high-paying jobs than for lower-wage jobs, which may contribute to economic inequality.

Follow LiveScience on Twitter @livescience. We're also on Facebook& Google+.