The Politics of Bots in Politics
Journalists employing bots on social media to cover political spectacles — what could possibly go wrong?
The 2016 US presidential election introduced new tools, or saw the expansion of existing tools, in political reporting — live-streamed debates, Google Trends interactive data charts, and bots. Bots everywhere. News organizations like the New York Times, ProPublica, Quartz, along with others, developed bots to deliver election updates and related news to an inconvenienced public. It remains to be seen if 2016’s bot revolution helped, hindered, or didn’t affect at all the results of the presidential election — what is clear: bots are only expanding in today’s newsroom. Will their proliferation reach the masses, or remain within a digital media bubble?
To understand the impact of bots in reporting on the 2016 election, let’s chronicle the year’s highs and lows for news bots:
Bots and the Presidential Primaries
The New York Times experimented with digital innovations throughout the presidential primaries, with perhaps their most ambitious endeavor being a Slack chatbot which allowed users to ask the NYT newsroom questions with the prompt: /asknytelection. The Times’ chatbot was an attempt to offer greater accessibility to readers, while conveniently informing them through an app they already have — or do they? Discussing the Times’ election Slack bot with Nieman Lab, Mark Lavallee, the NYT editor of interactive news, initially forecasted the limiting potential of Slack bots, saying: “The people who are willing to even be chatting about something like a primary or caucus night on Slack are a little more of an insider audience.” Lavallee sees the NYT election bot as a predictor of innovations to come:
“We haven’t crystallized what we’re committing to, and we haven’t decided how to present the value of this to readers. We’re just practicing a bit in public.” — Lavallee
Clinton v. Trump and Twitter Bots
The Slack bot was not the Times’ only foray into election bots — the news giant also created a Facebook Messenger bot, this time emphasizing the experimental purpose of the bot.
Once again, Lavallee explained the reasoning behind the bot as an experiment in engaging the public:
“What we’re trying to do is figure out if there a space between broadcast, where everyone gets the same exact thing, and a total human-powered one-on-one interaction — which obviously doesn’t scale easily — to see if we can have the best of both worlds.” — Lavallee
The human-powered quality of bots, as Lavallee asserts, is perhaps why implementing them requires so much experimentation — there is never any clear way to engage the public, especially when journalists are asking for interactions beyond the passive tasks of watching broadcast news, or scrolling through Twitter feeds.
Speaking of Twitter, there was another sort of bot crashing the political party this year: fake profile bots. According to research from Oxford University, a third of pro-Trump tweets and nearly a fifth of pro-Clinton tweets, during the first and second presidential debates, came from bot accounts. Fake Trump and Clinton supporters polluting the legitimacy of post-debate Twitter polls, and potentially throwing a wrench into the “human-powered” interactiveness of bots.
What does it mean for news organizations when the public develops a perception of bots as being, essentially, robot trolls?
Bots After President-Elect Trump
The Times’ Facebook Messenger bot is currently defunct, but not every news organization abandoned their election bots post-Election Day. ProPublica still has a live election data bot, updating readers on people and topics connected to the presidential candidates. ProPublica’s data bot looks specifically at campaign finances, Google Trends, and 538 election forecasts. And the ProPublica Election Databot tagline: “There are a thousand stories in every political campaign. Use this to find them.”
Finding stories might imply the journalistic value of ProPublica’s Databot, which may also support Lavallee’s concerns for the insider nature of news bots. Who is meant to utilize bots like the NYT election bot or ProPublica’s Databot? Is the proliferation of bots in digital media adding to the argument that journalists are reporting within media echo chambers?
Next-Gen Bots
If journalists truly consider the proposition of meeting readers where they are, developing news bots on Facebook, What’s App, and Twitter could potentially be the next wave. In 2015, users on the popular youth-driven messaging app Kik, exchanged 350 million messages with bots — bots mostly connected to entertainment brands like Funny or Die, Moviefone, and Massively. So, how do journalists balance the development of personable bots, like a Slack bot that takes you into a New York Times newsroom, with the practicality of a Moviefone Kik bot that gives you new movie showtimes?
The amazing thing about bots is the exploration of artificial intelligence — ‘intelligence’ that should be live even when the journalist is asleep. The temporality of the Times’ election bot experiments sort of defeats the point of bots. What about New York City’s 2017 mayoral election, or the 2018 Senate elections? Lavallee said the Times was “practicing in public,” but does that assume the public isn’t entitled to some sort of reliability?
Consider Microsoft’s much-despised Clippy — an early manifestation of bot technology — no matter how often Clippy was discarded, he just kept popping up.
The human-powered goal of news bots, like Lavallee’s election Slack bot, is likely a fever dream in a world of busy journalists and an expanding global population — what is real, is the potential of A.I. bots like the entertainment bots Kik users love chatting up, or assist-bots like Cortana and Alexa.
Chatbots like Politibot (Spain) and Purple have made strides in covering single issues, like political trends. Purple has a choose your own adventure thread, as well as live chats with the editors and prominent people — Purple once had the head of the Federal Reserve answer questions about interest rates. Chatbots can also be an effective way to cover elections, by using simple prompts to explain issues candidates are focusing on, along with bots’ overall demystifying ability in informing the public on the election process.
By getting over the desire to make bots human-powered, journalists can develop bots that assist when and where journalists can’t. Experimentation should move beyond the platform in which the bot lives (although, that’s important too), and should focus on the assistive purpose of bots. Politics isn’t only confusing during presidential elections, so will the next New York Times’ political bot live on post-November?
