I went to the excellent Book Passage travel writing conference two weeks ago, and since then I’ve been thinking a lot about Good Writing and What’s Happening With It.
I left my last job in news because it was becoming more and more thoughtless: skim copy for obvious mistakes, optimise for SEO, push publish button, repeat. Not only was it numbingly tedious, but upper management were always looking to ‘cut editorial costs’ – by which they meant get rid of editorial jobs by automating editorial work as much as possible.
Yesterday in the Colombia Journalism Review I read about a similar move from Gannett, a newspaper group in the south-east of the US, who are eliminating editorial jobs and basing editorial decisions on metrics:
“As the reporters become more attuned to their metrics and what readers are telling them, and become more expert at analyzing that data,” [Gannett's vice president for news, Kate] Marymont says, “the link between reporter and conventional assignment editor isn’t as necessary. Readers become the assignment editor instead of the more conventional assignment editor of the past.”
- “Gannett cribs from Advance Publications playbook”, August 18, 2014
Actually, Marymont is wrong when she says “readers become the assignment editor”. The readers don’t become the assignment editor, the algorithm does. No human editorial judgment is involved – it’s all based on data. Readers being assignment editors would look something more like the Guardian’s daily Comment is Free suggestion posts, where readers suggest columns they’d like to see and the editors choose from those ideas to commission pieces from their staff writers.
The idea of journalists writing stories completely based on metrics reminds me of the hilarious mishap a few years ago, when two algorithms aimed at gaming Amazon Marketplace pricing unintentionally led to a book on developmental biology in flies being priced at nearly $24m. Two booksellers separately set their pricing mechanisms in such a way that they both kept adjusting up in response to the other. Eventually one of them hit $23,698,655.93, at which point some agent (possibly a human manager) stepped in and reset the price to a more reasonable $106.23.
It seems to me that the logical endpoint of automated editorial decisions is that the whole process of content production (what Gannett calls “journalism”) will become completely automated. As computers get better at sounding like people, the same algorithms that are currently being used to figure out what to write will also be able to generate that content and publish it. Basically, a parallel news world will be created, filled with computer-generated content that doesn’t involve humans at all: an online publishing industry populated by algorithms, for algorithms.
I do think algorithms have a place in people-driven online journalism: things like Twitter trending are useful and important. Ferguson became a story because thousands of people were talking about it, and might not have become one if it had been left up to editorial judgment. Media institutions have biases too, and in the case of Ferguson, algorithms were a corrective to those biases.
Fully automated content production would also obviously mean a large loss of writing jobs. According to that CJR article above, after currently planned layoffs, staffing at Gannett papers will have dropped 62% since 2001. If I were one of their employees left standing, I wouldn’t have high expectations for my five-year job security. At content farms, a lot of people are being paid to write articles aimed at search engines; if those sites begin completely automating content production, those jobs will go.
But I think the increased automation of mindless content will mean the people who are being paid for writing will be paid for good writing. Buzzfeed uses its clickbait to generate income to pay for longform investigative journalism, as does eg the Daily Mail (yes I know). Fewer people will be able to make a living as full-time writers, but actually I think there was an unusually high number of full-time writing jobs in the past 70 years in US journalism. So maybe we’ll be moving away from there being a lot of bad writing jobs to there being fewer, but better ones. Yay?