Now You Can Listen to the Internet's Secret Algorithms

Outlier from Wesley Goatley.

Text analysis algorithms, whether corporate or state-controlled, ceaselessly and robotically scan the Internet's word horde, yet few people understand how these technologies function. In an attempt to illuminate these algorithms, sound artist ​Wesley Goatley flips text analysis tech on its head for his project Outlier, a sonification program that audio simulates sound environments—cafes, the woods, and the beach—for people reading articles or social media posts.

Goatley told me that the project came about through his ongoing exploration of the "hidden elements of the Internet", including his ​Wireless-Fidelity piece, which digs into the sounds of corporate wireless network operators. ​Available for free via Mediafire, users can have Outlier up and sonifying their text in no time.

Built on Max/MSP, and inspired by commercial and state text analysis technologies, Outlier assigns "readable" sounds to the results of an algorithmic sentiment analysis—Positive, Neutral, Negative—of any inputted text, as well as any keywords that the reader desires to search. The results are represented in a binaural sonic environment (simulated 3-D sound); becoming, according to Goatley, a form of narration, playing back at the speed of reading so that data about individual words and sections is relevant to what is being read in each moment.

"In Outlier, [text analysis] technologies are repurposed to analyze the text we're reading and give us new insights that computational text analysis allows," Goatley said, "such as how the logic of the language used implies a positive or negative attitude expressed by the writer, or how frequently and in what patterns certain words are used, analyses that are performed near-instantly."

Goatley calls the sonified results a form of "hyperreal background sound". He also variously calls Outlier "augmented reading," a "cyborg prosthesis," and "computer-assisted reading" since humans can read the text while the text analysis sonification plays out. Whatever it is, Outlier puts the text analysis algorithm at the disposal of internet users, not just NSA crypto-spooks.

"It seems that we have cause to speak of algorithms when we discuss almost every aspect of our online activities, though rarely do we have the chance to explore how these opaque computational processes interpret our data," Goatley writes on the Outlier website. "This is reflected in ​the ​language we use to describe them, which is often analogous to 'an algorithm ate my homework and monitored all of my textual communication.' With a lack of experience in their capacities and limitations, informed critique of the culture which deploys these processes becomes more difficult."

While Goatley envisions inputted sonified keywords becoming "embedded disruptions" in the reader's hyperrreality, with a banality that becomes unsettling through its hidden presence and repetition, he also hopes that Outlier works toward "unconcealing" text analysis algorithms, and increasing our understanding of their function online and in society. And, indeed, it simultaneously works as data-infused, cyborg sound art and fulfills Goatley's goal of shining light on the background algorithms of the Internet.