How Algorithms Are Changing What We Read Online

How Algorithms Are Changing What We Read Online

 

The AI of the internet determines what's relevant. One day, it decided my work wasn't

 

 

by Russell SmithRussell Smith
Illustration by Sam Island, 

The Walrus

Sep. 8, 2020

 

 

Last November, I stopped writing a regular column on art and culture for the Globe and Mail, my job for almost twenty years. Nobody noticed. I did not receive a single reader’s letter. I had a polite message from my section editor. He was sorry things didn’t work out and hoped we could stay in touch. The note contained no sense of symbolic occasion. I knew what I did was no longer important, either to the national culture or to the newspaper’s bottom line.

 

To be fair, my columns explored aesthetic topics newspapers typically avoided. I opted not to weigh in very often on big moral questions of race and gender. I didn’t cover Roman Polanski’s child-rape charges, for example, or Jian Ghomeshi’s trial for sexual assault. Instead, I steered readers toward controversies that weren’t headline news. I was drawn to the language used to discuss religion on Al Jazeera. I analyzed American Apparel’s “anti-brand” marketing campaigns. I dismissed pop music as the most conservative art form in existence. Creative tropes, I believed, were political in unpredictable ways and just as important to our intellectual landscape as the left/right punditry dailies usually traffic in. When Alice Walker published an antisemitic poem, I didn’t talk about antisemitism but instead explored the poem’s similarity—along with much new verse—to a Twitter thread.

 

For a while, it worked. My column netted me invitations to TV news shows. I was asked to speak at conferences on new music and editing. I made enemies too. When I called school spirit anti-intellectual, dozens of students at Queen’s University protested, either by letter or video. Indeed, triggering furious responses seemed my forte. The fiction writer Rebecca Rosenblum wrote, in Canadian Notes & Queries, that my columns were “about starting a fight.”

 

But, several years ago, I began to worry my readership was falling off. I would run into middle-aged people at functions and they would say, “I miss reading you in the Globe!” and I would say, “I’m still there, weekly, in the arts section,” and they would say, “Ah, I get it on my phone, and you don’t come up on the app.” Out in Halifax, my mother couldn’t read me anymore after the Globe stopped offering a paper edition in the Atlantic provinces, in 2017. I taught graduate students in writing, and I had given up on expecting them to know I wrote for a national daily. They rarely seemed to read newspapers, and certainly not that one.

 

My increasing feelings of irrelevance also came with a corresponding pressure to cover what, for me, were the less and less interesting topics that my editors felt would be better at maintaining my impact—general-interest arts stories and mainstream popular culture. That pressure, I felt, came largely from the huge moving electronic graph that now hangs over most newsrooms, tracking the articles causing the most reader interest in real time.

 

Today, I see my two-decade career as an allegory of how the digital age—and especially its omnipresent “metrics”—has changed what we read. I wanted to open the doors to private clubs. But, at a certain point, it became clear that no one wanted to walk through anymore.

 

Ipitched the column in 1999, maybe the only time something so unabashedly highbrow could have been conceived for newsprint. The National Post had just launched, and we were in the throes of a newspaper war. The Post was irreverent and colourful. Its style was reminiscent of British papers, in which wit was as valuable as investigative scoops. It loved quirky copy. It had more illustrations. Its cartoons were funnier. Suddenly, our national newspaper looked staid.

 

The owners of the Globe responded by bringing in a publisher battle-hardened in the trenches of the highly competitive UK newspaper market. He brought in a British editor-in-chief. Forty-three at the time, Richard Addis had an affable private-school charm and began recruiting writers as much for their glamour as for their journalism. Over lunch, I told him I wanted my focus to be international and intellectual. I told him I would be the bridge between suburban Canada and the violent paganism of Norwegian black metal. I made it clear I wouldn’t touch mainstream culture—no pop music, no celebrities, no Hollywood movies.

 

He smiled and told me to start in a week. I was soon joined by Leah McLaren, who began to chronicle her life as a young woman in the city—a life almost completely unrepresented in the old and male newspaper at that time. Lynn Crosbie also signed up to write a column about celebrity culture, in which she used a densely literary style to compare people like Britney Spears to heroes of classical mythology. All of a sudden, Canadian newspapers were sexier—less sober, less policy focused, less Canadian.

 

The economic crash of 2008 changed everything. Print sales were sinking. Freelancers writing for dailies feared the collapse would take them down too. I had no contract, no security. I was invoicing every week for $800. My editor at that time, Andrew Gorham, told me that I had to take a pay cut of $100 per week. I was spending a day on the column and filling the rest of the week with other freelance work to pay my rent. I could take it or leave it. I took it.

 

• Objectivity Is a Privilege Afforded to White Journalists

• Letter to a Young Indigenous Journalist

• The Role of the Media in a Time of Crisis

In 2010, the waves of departures started. In 2013, sixty Globe employees took buyout offers. In 2014, eighteen positions were cut, including nine editorial jobs. In 2016, the publisher announced he was looking for forty staffers to take voluntary buyouts. Three years later, $10 million in labour costs needed to be cut, which led to more buyouts. (Gorham took one last year.) Lucky to still have any regular pay, I stuck at it. I outlasted four editors-in-chief and nine section editors.

 

My beat probably helped me survive. Both at the Globe and across Canada, there wasn’t a lot of competition for what I was doing. I was providing a lot of content—content that helped maintain the paper’s literary brand—for very cheap. But a new arts editor, who came on board around 2016, displayed increasing concern for me. My guess, based on all his talk about “engagement,” was that he was getting pressure from management about my weak numbers. The Globe had, by then, developed Sophi, its own analytic software. Sophi tallies how much of an article is read, how many times it is shared and commented on, and most importantly, whether it being behind a paywall spurs anyone to buy a subscription.

 

Articles that show low engagement typically get sidelined in favour of pieces that show more, a measurement that, along with all of the above, takes into account the click-through rate, or CTR. “You’re looking at your analytics,” Gorham explained to me, “and you’re saying, Holy shit, this story’s got a high CTR, let’s move it forward. Surface it—share it on Facebook, put it on the home page, release a news alert, put it in the newsletter.” That support is key to keeping engagement up. “If we don’t juice it,” he said, “it just evaporates.”

 

In practice, this ensures the less read become even less read. It creates what one might call popularity polarization: a few pieces rise to the top, leaving the rest to fend for themselves. With print, this didn’t happen as much. Flipping pages, you would see every article somewhere. But, on your phone, you scroll through what’s been selected for you. And that selection likely reflects a ruthless narrowing of editorial values and priorities. “You don’t try to do everything for everyone,” is how Gorham described it. “It’s all about swinging for the fences. Don’t hit singles, don’t play small ball. You pick your one and you hit it hard.” For the Globe, it meant more resources going to major, socially relevant projects, such as Unfounded, Robyn Doolittle’s two-year investigation into unprosecuted sexual assaults.

 

My clever musings on the relevance of the opera Don Giovanni in the age of #MeToo were not exactly “big” in that sense. But I still believed that a newspaper can and should cover both sexual assault and the arts. If I wanted to keep writing, I needed to push my numbers up.

 

So, last summer, I paid several visits to the Globe’s gleaming new offices to chat with my boss about what I might do. I saw what a contemporary newsroom looks like. If you walk the full floor of the Globe, you’ll see, along the pillars, five or six big screens; even the coffee area has two of them. These are Sophi’s HUDs, the head-up displays of the huge brain. Featuring a graph with moving lines, each screen shows engagement, in real time, with the stories currently on the paper’s website. The top line is usually breaking Canadian news: the Fort McMurray fires, say, or the shooting on Parliament Hill. The more provocative political-opinion writers might constitute the second line. I was here to see if my 1,500-word feuilletons on machine art in Hamburg or new internet slang could break into those rankingsThat idea of engagement, however, made my heart race. It wasn’t at all what connecting with readers used to mean to me. If my ideas were being discussed in academic papers, if I was giving bloggers strokes, if I was annoying the powers that be at Heritage Canada or in the upper floors of the CBC, that, to me, was engagement.

 

That was how I measured my influence on the culture. But one thing that Sophi does not weight differently is readers. All readers are effectively the same: a click is a click, whether it comes from your mouse or from Margaret Atwood’s. So Sophi cannot measure engagement in my twentieth-century sense.

 

My editor suggested a new focus. Would I like to be a weekly books columnist? Of course, there was no extra money. I tried for a few months, but my reading couldn’t keep up with the deadlines. My editor relented and allow me to expand my scope and cover varied arts subjects—as long as they were Canadian. According to Sophi, Canadian subjects got the most engagement. Maybe so, but the issues I was interested in—the influence of technology on art or the echoing of long-vanished art schools—were mostly playing out on an international stage; I couldn’t find enough examples of these things from Canada alone. I feared that, if I had to write about only Canada, I would end up with the dogoody stuff the CBC is stuck with: the shortlist for a Responsible Fiction prize or a play about fighting transphobia in Edmonton, not the glamour of a rude French novelist or a noise-and-sex festival in Tokyo.

 

The week we were negotiating, an international arts scandal exploded: the Nobel Prize in Literature was awarded to Peter Handke, an apologist for Slobodan Milošević, Serbia’s former strongman who was tried for genocide in the wake of the Yugoslav Wars. Condemnation dominated arts news and blogs around the world. It was exactly the sort of thing I would have been asked to do a hot take on. But my editor confirmed that they weren’t assigning it to anyone. The outrage, I gathered, wasn’t Canadian enough to merit comment. It was at that point I realized there was no longer a role for me.

 

Most major news outlets in Canada share the conviction that their primary arts focus should be Canadian. After all, foreign papers won’t cover that play in Edmonton. Also, people no longer turn to one journal for information. If they want to read up on an international controversy, they have their pick of brilliant critics in The New Yorker, the Guardian, and hundreds of blogs and podcasts, all streaming to their phones, often for free. The web, in other words, is awash in opinion. Once I went behind a paywall, even my friends moved on. They can read Hilton Als, Jerry Saltz, and A. O. Scott any time they want.

 

One could argue that, given the easy access to international opinion, a Canadian viewpoint is even more vital. Why not stake out a role as one of a handful of commentators on homegrown art? Well, there’s a paradox in the local-only approach: Canadian artists are not isolationists. They want a show in New York as much as anyone. They readily accept residencies in Stockholm and New Zealand. Ask a Canadian composer her biggest influence and she is just as likely to drop an American or Japanese name as a Canadian one. It strikes me, then, as simpleminded to think that covering only the arts that exist within our borders is to actually cover our arts. Our arts are international. So, if a newspaper is national, it must be international as well, just as our artists are.

 

I don’t mean to propose that my disappearance should be cause for concern. I had a good run. But I worry about analytics driving editorial decisions. Most media outlets (including The Walrus) pay close attention to what generates traffic—any journal that doesn’t is likely to fail. But trusting that data to curate content is another matter: algorithms that tell us which topics are trending don’t merely reflect trends; they can also help create them. In winnowing out the slightly obscure or difficult, algorithms ensure it can never be popular. If no one is ever told that electronic music or postmodern architecture are significant topics, then those things have a reduced chance of being treated as significant. The newspaper’s role as arbiter is diminished. Simply put, if your metrics tell you the provincial is the most important thing, your journal will likely become provincial. In trying to be “bigger,” you can get smaller too.

 

It became clear to me that I hadn’t been read by my peers for quite a while. It is quite possible that interest in the intellectual had tailed off in the population at large, but I find that hard to believe. The internet has shown us that the oddest of subcultures and smallest of niches can develop followings. The reading I come across in my social media feeds now is, in fact, more cerebral than ever. The great papers of the world still seem to want to participate in the conversations academics and underground artists are having.

 

No, I don’t think readers weren’t interested. It’s that they were told not to be interested. The algorithms had already decided my subjects were not breaking news. Those algorithms then ensured that they would never be. When I took my final bow, the room was already empty.

 

Russell Smith

Russell Smith is the author of nine books and has been nominated for the Trillium Prize, the Governor General's Award, and the Giller Prize.

 

Sam Island

Sam Island (samislandart.com) has drawn for the New York Times, Monocle, The Atlantic, and Time Magazine.