Over the past week Jack Dorsey, the chief executive of Twitter, might have felt a twinge of sympathy for countryman Mark Geiger. Geiger, you will remember, was the referee in the thrilling Colombia v England clash in the last 16 of the World Cup. If you were cheering for England, you will probably remember him as a referee who did a decent job in fraught circumstances. If you are Colombian, you might think of him as the bumbling idiot who disallowed a perfectly good goal, was hopelessly biased towards England and turned a blind eye to diving and play acting.
Search Twitter and you will see all of these beliefs, strongly held, about the role of the hapless referee. The issue of “bias” is hotly contested, even though, thanks to video footage, every single angle is recorded and analysed. If you read the Spanish version of CNN you might be forgiven for thinking you had been watching an entirely different match from the English-language CNN.
The issue of how you solve bias is very much on the minds of Twitter’s Dorsey and his counterparts at Facebook and other tech companies, as the US political cycle heads to the midterm elections in November.
News broke last week that Dorsey was one of a number of technology leaders looking to address Republican concerns that Silicon Valley companies are allowing their liberal bias to affect the way their platforms publish and edit content. By all accounts, including a report in the Washington Post, a series of detente meetings and a dinner took place. This news infuriated those who see Twitter as a main conduit through which the right coordinates attacks on critics and opponents, and as a platform enabling a far right presidency to consistently attack established democratic principles such as those of a free press.
Commentators pointed out that the destabilising of platforms such as Twitter or Facebook with incessant complaints, but little evidence of systemic bias, is consistent with the US conservative playbook.
Talking to those who are involved in both product and policy at platform companies, it is clear that there are two things happening. One is that most companies are still grappling internally with an ideological business divide about how much the respective companies should involve themselves in politics and news at all; the second is that where there is a public commitment to “solving” the problem of toxic discourse (for instance at Twitter), there is intense internal debate about how to deal with bias, or perceived bias.
Do companies adopt more overt codes and terms of use which embody company values, which are in themselves political? Do they opt, instead, for a more transparent method of complaint process and arbitration at significant internal cost? One thing, though, is fairly clear: the current state of affairs has to evolve from a push-me-pull-you of the prevailing political will.
Tech companies are seasoned veterans when it comes to lobbying political power. But they are, perhaps, less seasoned when it comes to being lobbied themselves. There are very few protocols in place that inoculate the companies against the kind of pressure media companies have withstood, or caved under, for decades.
Dorsey’s Washington dinners triggered parallel comparisons with Facebook chief executive Mark Zuckerberg’s ill-fated round table with US conservatives in May 2016, which came after disclosures that human curators suppressed ratings for far-right news sources in one particular ranking formula.
Zuckerberg famously, and perhaps rashly, promised that Facebook would be a platform for all voices. Removing human curation and replacing it with robots just ahead of the election cycle did not necessarily cause it to be flooded with Russian propaganda, Macedonian fake news, and other assorted nonsense, but it didn’t help.
Robo-moderation cannot recalibrate the political commons. There was another furore in the right-adjacent ranks last week when a passage from the American Declaration of Independence was censored from a news organisation’s Facebook page. A Texan newspaper, the Liberty County Vindicator, had part of a post containing the declaration removed, with a notification from Facebook saying it “violated hate-speech standards”. The passage refers to King George and says: “He has excited domestic insurrections amongst us, and has endeavoured to bring on the inhabitants of our frontiers, the merciless Indian savages, whose known rule of warfare is an undistinguished destruction of all ages, sexes and conditions.” It does indeed violate hate-speech standards. It should be classified as fake news but it is simultaneously an important historical document.
It is understandable that the policy guidelines and terms of use from technology companies do not tend to have much in the way of an auditing process for political balance. Quantifying bias on a platform such as Twitter is close to impossible, but explanation of contentious decision-making is a much more achievable aim. Whether it is wise or helpful for chief executives to hold closed-door meetings with complaining politicians and lobbyists is doubtful.
There is nothing less “Silicon Valley” than the idea of outsourcing complaints to bureaucratic review committees, but when a chief executive’s own political orientation and that of his board is under direct attack it might be a preferable alternative to a sorry tour of Washington restaurants.
The United States has a rich if chequered history of thinking about objectivity in relation to its daily newspapers, born partly out of the process during the late 19th and early 20th centuries when the press became separated from party affiliations. The famous journalist Walter Lippmann, who wrote many of the grounding principles for modern US journalism, noted that “honest efforts” and the “cynicism of the trade” were not enough to keep journalists from inadvertently exercising their own bias while reporting. Journalists needed “a unity of method, rather than aim; the unity of disciplined experiment”.
Lippmann’s formula for what was wrong with journalism and how to cure it has itself been put under intense pressure by its impracticality and the advent of social media. The “view from nowhere”, as the media scholar Jay Rosen puts it, stretches credibility in an online environment where we share everything from our lunch options to political preferences online. Like all media systems before them, platforms cannot eliminate accusations of bias, but they might want to think about how to train and hone their referees’ judgment and compensate for the shortcomings of VAR.
