Does funding article-writing robots really support high-quality journalism in the digital age?

Here’s a riddle for you:

If a programme set up to support high-quality journalism in the digital age funds the creation of article-writing robots, is it really supporting journalism?

The answer might depend on how you define “journalism” to begin with.

A few days ago, Google announced the recipients of the latest round of funding from its Digital News Initiative. The Digital News Initiative, or DNI, is “a partnership between Google and news publishers in Europe to support high-quality journalism through technology and innovation.” The latest round of funding sees almost 22 million € worth of funding allocated to 107 projects in 27 countries across Europe.

One successfully funded project in the UK is causing a bit of a stir. Google has awarded 706,000€ to the Press Association to fund RADAR, a project that will see 30,000 new stories a month created for local news media – by computers.

RADAR, or to give the project its full name, Robots And Data And Reporters, aims to use open government and local authority databases to create automatic stories about health, crime, employment and so on. It will be overseen by a team of five (human) journalists, but the bulk of the work will be done by AI.

Canadian news outlet Global News writes that,

“The envisioned workflow would begin with human journalists identifying open data sets and “creating detailed story templates across a range of topics including crime, health and employment.” The robotic reporter would then take over and scan the data, use language generation software to craft together story text and automatically generate relevant photos and video.”

According to Press Association editor Pete Clifton, “At a time when many media outlets are experiencing commercial pressures, RADAR will provide the news ecosystem with a cost-effective way to provide incisive local stories, enabling audiences to hold democratic bodies to account.

“Skilled human journalists will still be vital in the process, but RADAR allows us to harness artificial intelligence to scale up to a volume of local stories that would be impossible to provide manually.”

Okay, there’s a lot to unpack here. It’s tempting to give a knee-jerk response and declare that news bots could never replace human reporters, but there are cases in which that’s true and cases in which that is patently false. Artificial intelligence and natural language processing are getting more and more advanced. We already have software that can generate email marketing copy that “beats humans more than 95% of the time” (according to its creators) and an AI personal assistant who can convincingly schedule meetings and reply to your emails. Add to that the fact that much of news reporting has already become about rewording press releases, and it’s not hard to see how a robot could do that job perfectly well.

But that doesn’t mean that RADAR will do exactly what Clifton claims it will for local journalism. I’m stuck on the part where he says that automated news articles can help “audiences to hold democratic bodies to account”. Holding elected bodies to account is indeed one of the most important roles of local journalism, and one which has suffered the most in the wake of so many local newspapers closing down. But it’s this which typically requires the touch of a human reporter: digging into possible leads, making connections between different pieces of evidence, interpreting the data, and weaving a story from it all that local readers can follow along with.

Can automated data bots do any of this? Without having read a sample of the articles that the Press Association plans to produce with RADAR, it’s hard to say; my guess is that automated news articles would at best be able to create a fairly basic narrative around the data in question, leaving it to the audience to identify any suspect details or spot any interesting connections – a job that, normally, would be done by journalists.

I would define journalism, broadly speaking, as the act of rendering information more accessible to an audience. So once the onus is on the audience to interpret the data themselves, can it really be called journalism?

A lot of details of how the Press Association plans to implement this project are still unknown. 30,000 news stories a month sounds like a lot (The Register‘s coverage of the story ran with the headline ‘Google coughs up $800k to build news bots that will flood UK with 30,000 ‘articles’ a month‘), but divided up among the entirety of the UK’s regional and local news outlets, it may not amount to that many articles per publication. Ultimately I imagine it’ll be up to the editors to decide how much of their news coverage they want to be made up of robot reporting.

There are also many kinds of local stories that can’t be told through data. Human interest stories, local event coverage, reviews and other features – these are all types of articles that can’t be assembled using a template and open databases. So there is plenty of need still for the human touch in local journalism. I’m guessing that the plan is for the ‘robo-reporters’ to take some of the weight off the human journalists, by automating the mindless stuff and freeing them up to write the stories that really require a person to tell them.

In an article for The Guardian about the future of local news reporting last December, Sam Blackledge wrote that Britain’s local newspapers will need an “army of reporters” if they are to survive in the digital age. He wasn’t talking about an army of automated news copy-writers that would pad the pages of the local papers, but an army of determined, inquisitive local journalists who will stop at nothing to uncover the truth. However, maybe the existence of the former can free up the latter to spend more time doing what they do best.

Still, in a world where journalism of all kinds struggles to turn a profit, I can’t help but feel a little rueful seeing hundreds of thousands of Euros go towards funding short-form, automated news articles instead of supporting well-crafted, in-depth, long-form journalism, of the kind that’s much harder to come by nowadays.

But there is a fantastic diversity of projects being funded with the money from Google’s Digital News Initiative, of which RADAR is only one. In the UK alone, other successfully funded projects include ‘News Immersified‘, a project by AJ Labs which aims to deliver journalism in interactive form via a messaging app; ‘Watch Together‘, which combines live streaming the news with collaborative viewing and real-time discussion; and ‘DMINR‘, a project by my alma mater, City University London, which will create a “research and verification tool to help journalists work with big data and conduct investigations in the digital era.”

385,000€ has also been awarded to help “scale up” Jimmy Wales’ collaborative news endeavour WikiTribune, which I intend to do a separate post on at some point.

The DNI’s third round of funding also coincided with the publication of the first DNI Annual Report, which contains some fascinating stats about the different kinds of projects funded and their early impact, and which I will definitely also be digging into in a separate article.

In the meantime, I’ll wait to see the impact of this “flood” of automated news content into our local media. Best-case scenario, we won’t notice anything, but some intrepid reporters might be able to publish five-page exposés of local government corruption where they would have otherwise been stuck in the newsroom churning out rewritten press releases.

Worst-case scenario, the future of local journalism in the digital age will remain murky, and we’ll go looking for another way to fix the problem.

On “emoji translation” and news in emoji

This week, the BBC reported on a news story that is likely to raise a few eyebrows: a London-based firm has advertised a job opening for an “emoji translator”, thought to be the first job of its kind worldwide.

The story becomes a little less bizarre when you know that the recruiting company is a translation firm, ‘Today Translation’. But this then raises a number of other questions: namely, what kind of jobs would require someone to translate into – or from – emoji? And does this mean that emoji is considered a language equal to the likes of French, Spanish and English?

In the same news article, the BBC referenced a claim by Bangor University professor Vyv Evans that emoji  is “the fastest-growing form of language in history“. But is it, though? Or are commentators just getting carried away about what they see as a hip and youthful new form of communication?

Obama's State of the Union address in emoji

In journalism, in particular, there’ve been a number of fairly cringeworthy attempts to convey or to jazz up the news using emoji; whether by looking at current news stories through the lens of emoji, or through “emoji translations” of news, such as Vox‘s 2014: The Year in Emoji or the Guardian rendering Obama’s State of the Union speech (mostly) into emoji.

If you can successfully convey a news story in emoji without losing any of its key meaning, then that lends credibility to the idea of emoji being, or having the potential to be, a language in its own right. At the same time, though, I can’t help but wonder: What’s the point of telling news stories in emoji? What does it add to the story?

Emoji as visual language

Emoji might just now be gaining mainstream attention, but those of us who’ve been around the internet know the concept is ages old. And not just with simple text emoticons like :), :( and :D – communities like deviantART boast a huge range of different emotes, while MSN custom emoticons let you express pretty much whatever you wanted (I remember the agony of getting a new computer and losing a huge library of custom emotes, and have to get them all back from your friends).

There’s also kaomoji, the Japanese text-based emoticons which use a dizzying array of different symbols and punctuation to convey much more sophisticated emotions and gestures. o(≧▽≦)o (ಥ﹏ಥ)

What makes emoji different is the level of standardisation and universality: the same Unicode-approved library of emoji appears across Apple devices, Android, Microsoft, Twitter and more. But the very fact of Unicode’s standardisation (the entire process of submitting a new emoji concept for approval takes about 18 months, and only around half of public emoji proposals make the cut) is also a major barrier to emoji becoming a language in its own right. As the BBC’s Neil Cohn pointed out in a thoughtful consideration of whether or not emoji could become a new language,

“Emoji do not allow this building of units from parts … [T]hese limits also prevent users from creating novel signs – a requisite for all languages, especially emerging ones. Users have no control over the development of the vocabulary. As the “vocab list” for emoji grows, it will become increasingly unwieldy: using them will require a conscious search process through an external list, not an easy generation from our own mental vocabulary, like the way we naturally speak or draw. This is a key point – it means that emoji lack the flexibility needed to create a new language.”

New York Times article about the world’s first emoji convention, Emojicon, describes some of the ongoing efforts to liberate emoji from the control of Unicode’s “shadowy” committees, and open up the process of adding new emoji to the wider public. Even then, as Cohn also points out, emoji would still lack another key component of language, which is grammar.

“A grammatical system is a set of constraints that governs how the meaning of an utterance is packaged in a coherent way. Natural language grammars have certain traits that distinguish them. For one, they have individual units that play different roles in the sequence – like nouns and verbs in a sentence.

“When emoji are isolated, they are primarily governed by simple rules related to meaning alone, without these more complex rules. … I would hypothesise that emoji can use a basic narrative structure to organise short stories (likely made up of agent-action sequences), but I highly doubt that they would be able to create embedded clauses.”

In their current form, emoji serve a purpose similar to a gesture or a facial expression: they accompany verbal expression, clarifying, punctuating or adding nuance. Proponents of the ‘power’ of emoji like to quote statistics such as “72% of Brits aged 18-25 find it easier to express their feelings in emoji, rather than words“. But a statistic claiming that 72% of young people find it easier to express their feelings with a facial expression, rather than words, wouldn’t be considered newsworthy at all.

Rolling eyes emoji

If someone told me that the Guardian translated Obama’s State of the Union address into emoji, my reaction would probably be to roll my eyes, rather than to express my exact feelings in a sentence.

If someone told me that the Guardian translated Obama’s State of the Union address into emoji, I might find it more efficient to express my feelings by rolling my eyes, rather than trying to pinpoint the exact words to best convey my exasperation. But no-one goes around proclaiming that facial expressions are the UK’s fastest-growing language, or that brands need to try and connect with young people by using facial expressions.

Emojifying the news

I think that a lot of the hype surrounding emoji has come not from the fact that they’re anything new or revolutionary, because they’re not, but from commentators who are largely of an older generation, who view emoji as some kind of mystical, hyper-youthful mode of communication that, if they can just tap into it, will allow them to communicate with young people on ‘their level’. On his website, Professor Vyv Evans writes that, “the inclusion of emoji to help convey meaning in abridged versions of Shakespeare could help bring those great stories to life for a whole new generation.”

I can’t fathom what possible benefit could be gained by adding emoji into the works of Shakespeare. Should the lines about sex be punctuated with aubergine emoji? Do they think young people need a heart emoji in order to grasp the meaning of a particularly romantic line? Do we need an face-with-xs-for-eyes emoji to depict Gloucester having his eyes gouged out in King Lear?

Two book covers side by side, one entitled Macbeth #killingit, the other A Midsummer Night #nofilter. Both depict scenes from the plays on the front cover, with the characters' faces replaced by emoji.

…Of course this is already a thing. (Images by Penguin Random House)

As I said in the beginning of this post, a number of news organisations have attempted to use emoji to make news stories somehow funnier, snappier, or easier to understand. Well, as a ‘millennial’ who is supposedly part of the generation that best understands emoji, I find news headlines in emoji completely indecipherable. @newsinemoji, for example, was a fairly short-lived Twitter account which would tweet out news headlines in a series of emoji, together with a link to the story on German news site Spiegel Online. But without the link to give context, the tweets just become a string of nonsense.

Billy Penn, a mobile local news platform for Philadelphia, USA, made a similar attempt when it live-tweeted updates from the final Pennsylvania Governor’s debates in October 2014 using emoji. Billy Penn wrote on its Storify of the live-tweets that, “[O]ur goals revolve around getting you interested in things that matter. Do you really want to hear gubernatorial candidates drone on about the pension system? Probably not. It’s complicated, but it’s important.”

The question is, can complicated and important gubernatorial issues really be conveyed with emoji? A couple of the tweets were genuinely punchy and funny, such as this one, which humorously summed up Democratic candidate Tom Wolf’s opinion without using rude language:

But others were a complete mystery:

If news organisations and others are set on using emoji to communicate their message, then maybe we do need “emoji translators” – or, as Dr. Rob Drummond suggests, emoji specialists – to at least make sure that they’re used in a way that doesn’t completely fall flat, or turn the organisation into a laughing stock.

For a local news outlet, it’s pretty harmless, but the backlash that arose when the Clinton campaign asked young people to describe how their student loan debt makes them feel “in 3 emojis or less” probably left the campaign wishing they’d run the idea past a specialist. (Or a young person. Any young person).

I don’t want to seem like a killjoy for suggesting that news organisations and journalists can’t experiment with new ways for communicating the news – of course they can, and should. But it’s probably worth considering first what you think using emoji is going to add to the conversation. Just because emoji play an increasingly important role in person-to-person communication doesn’t make them suited for conveying more complex messages, any more than hand gestures or facial expressions are.

Emoji can be brilliant for adding nuance, humour or emphasis to text when used in the right way, but they’re a long way from being a language in their own right. And if you’re hoping that using emoji will magically attract and engage hordes of young people who are overjoyed that you’re finally ‘speaking their language’, well…

Unamused face emoji

Yeah.

The death of Vine: Did Vine ever live up to its potential as a journalistic tool?

At the end of last month, Twitter dropped a bombshell: it would be killing off Vine, the app for creating and posting six-second looping videos.

Vine was a big deal back in the days of 2012 and 2013 when short-form video on the internet wasn’t all that common. This was before Twitter introduced the ability to add 30-second video clips to your tweets, so Vine provided a way for people to share video content on Twitter for the first time. Snapchat and Instagram video also had yet to be introduced or catch on much by this point.

Vine came into its own mostly as a comedy platform and social media’s answer to ‘You’ve Been Framed’. But during its initial surge in popularity there was also – as there generally is with any new media fad – a raft of commentators waxing lyrical about its potential as a journalistic tool.

So what happened to that? Did Vine ever come close to living up to that potential?

Vine as a news tool

Journalists started talking about the potential for Vine to be used as a news tool in 2013, during its first year of life. Pando called Vine a “killer news tool“, citing its speed and ease of use and the ability to give audio and visual dimensions both to journalists and their stories. The short duration of Vine clips was also seen as an advantage in our world of short attention spans, as most believed that people would rather consume news in ‘bite-size chunks’ on the go rather than sit through a long video.

Francis Perraudin of the Guardian also pointed out that Vines could be uploaded using a relatively weak connection, which made it a huge advantage for journalists covering stories overseas in areas with poor internet. Some notable examples of this were the Vines shot by Channel 4 News chief correspondent Alex Thomson whilst covering the Ebola outbreak in Sierra Leone in 2014.

Perraudin wrote that:

Thomson’s Vines from Sierra Leone caused a stir on Twitter. People weren’t used to seeing six-second snatches of video, usually reserved for comedy, being used to document such serious subject matter. But his use of Vines to record snapshots of the situation seemed to work – “Your coverage of the Ebola outbreak has been fantastic. I understand the situation much better because of it”, one Twitter user tweeted to Thomson.

Journalists saw Vines as the potential missing piece for text-based news online, contextualising and humanising the stories reported; or even as a new format for visual news coverage. And at a time when Twitter was developing its potential as a centre for breaking news, Vine took it the extra mile, giving consumers of news on Twitter a way to experience stories as they happened, with the sights and sounds of the event as it unfolded.

In October 2013, Twitter’s New York City office hosted the inaugural Vine Journalist Awards, honouring the best examples of six-second video journalism across four categories: Behind the Scenes, Sports, Breaking News and Inspirational. The categories encompassed the breadth of what journalists were using Vines to portray, and the winners included a breathtaking video of a paraglider taking off in Germany, an 18-year-old Polio survivor riding a bicycle in Moradabad, India after reconstructive surgery, and the moment that bombs exploded as runners crossed the line at the Boston marathon.

Yet the first year of the Awards was also its only year, as Twitter appears not to have hosted the Vine Journalist Awards in 2014 or any year afterwards.

So what changed?

I think it’s safe to say that few people have used Vine as a journalistic tool over the past couple of years. And while there are numerous reasons why Twitter is killing off (or possibly selling off) Vine as a service in general, here are the reasons why I believe its usage in journalism died out:

Twitter video. In 2015 Twitter introduced the ability to upload video clips of up to 30 seconds directly to Twitter, as well as to shoot and edit videos within its iOS and Android apps. While Vine’s punchy, short-form six second offerings were undeniably an advantage, 30 seconds is still short enough to hold the user’s attention with a good video, while leaving much more room to manoeuvre.

Journalists could still shoot videos of under 30 seconds if they wanted to, but they weren’t as restricted as they would have been with Vine, and there was no need to create and edit in a separate platfrom – giving all of the advantages of a Vine video with few of its disadvantages.

Instagram and Snapchat. In 2013 Instagram introduced its own video sharing feature which allowed users to upload videos of up to 15 seconds. In 2016 it then expanded that limit to 30 seconds. While Instagram is not a notorious platform for breaking news, news organisations such as NBC News and the Washington Post have still found it a strong platform for sharing short-form video and promoting brand awareness.

The size of Instagram’s userbase far outstripped that of Vine (with 500 million monthly active users in 2015 versus roughly 200 million on Vine), giving media organisations access to a much bigger potential audience on a platform where they could upload both photos and videos.

A mobile screenshot taken from CBS News' Instagram page, showing a cartoon image of Donald Trump and Hillary Clinton at neighbouring election podiums. The image is a grid made up of 16 video thumbnails. Across the bottom, over the podiums, are quotes from each candidate about their policies.

CBS News uses video thumbnails to great effect on Instagram ahead of the 2016 US Election

Or for truly ephemeral photo and video content, Snapchat’s rise to huge popularity since 2013 has offered an attractive alternative for journalists and media brands. Just like Vine, Snapchat promised an up-and-coming, trendy social platform for sharing news in real time, with a young and highly engaged userbase. Snapchat’s Stories make it easy for users to collect photos and videos about the same event in one place, and the Discover feature has given media brands a permanent home. (While they might not always know what to do with it, the same could arguably be said of Vine).

Periscope. Twitter’s own live video streaming service was launched in 2015 as a rival to the newly-popular Meerkat (which found itself out-competed and discontinued in 2016). Periscope also has the potential to immerse users in the sights and sounds of a breaking news story, but in complete real-time, interacting with and commenting on the live video stream. Periscope has so far been used to broadcast ongoing and behind-the-scenes coverage of events such as the UK General Election debates in 2015, a building collapse in New York’s East Side, and the Democrat House sit-in protest over gun control.

periscope-building-collapse

A building collapse is broadcast live on Periscope (via The Verge)

The future of Vine still isn’t certain, as Twitter is reportedly in talks to possibly sell the service, and is entertaining bids from five different companies (TechCrunch). But even if Vine survives as a comedy and entertainment platform, it’s pretty clear that the ‘journalism tool’ ship has sailed, with other video services – including those owned by Twitter itself – offering more flexibility and fewer constraints.

Vine was ahead of the pack in offering short-form video before most other platforms hopped on the bandwagon, and I think that it did live up to its potential as a tool for breaking news in its time. It would have been nice to see the Vine Journalist Awards recur for another year or two, but Twitter had obviously moved its focus onto other areas already by then. Maybe next year will see the debut of the Periscope Journalist Awards?

Separating citizen journalism from social media hype

The other day I read a piece by Jesse Singal on New York Magazine about why “‘Citizen Journalism’ Is A Catastrophe Right Now, And It’ll Only Get Worse“.

In the third paragraph of the article, Singal writes,

Every time WikiLeaks drops a new trove of Hillary Clinton or Democratic National Committee emails, a torrent of bullshit is uncorked. That’s because countless citizen journalists rush to pore over the documents, posting j’accuse screen-grabs ripped from context that are quickly retweeted through huge, hyperactive networks of anti-Clinton Twitter denizens.

Singal then goes on to contrast the methods of these unscrupulous ‘citizen journalists’ against knowledgeable and trained professional journalists, who would know, for example, that there is nothing noteworthy about the New York Times contacting the Clinton campaign ahead of an article about to drop.

“To any working — that is to say, professional — journalist, there’s nothing here. Not because we’re trying to cover up collusion, but because this is a perfectly conventional interaction between a writer and subject,” writes Singal. “The headline description of this email is “Journalist Does Job.” And yet it, and countless other nothing snippets of nothing emails, has been crackling energetically across gonzo social media.”

The whole article, but particularly this section, made me wonder where we should be drawing the line between ‘citizen journalism’ and general social media hype. And it’s not a question that I’m claiming to know the answer to, but nor do I necessarily agree with the way that Singal frames the issue, either.

What makes someone a “working journalist” as opposed to a credulous “citizen journalist”? Is it that they get paid? In the article, Singal names right-wing political commentators Matt Drudge and Rush Limbaugh as having fallen for a troll tweet about ripping up absentee ballots in the Post Office. While they’re not news reporters, both Drudge and Limbaugh are established media figures who get paid for what they do. What about that makes them citizen journalists? I am a “working journalist” who is paid to write, but I write for B2B publishers and have never been in the position of having to contact a source for comment ahead of an article I’m publishing. Would I be classed as a citizen journalist if I fell for false information on social media?

Last weekend, I went along to the London Student Weekender, an afternoon of talks by great journalists like Buzzfeed’s James Ball and The Times’ Emma Tucker. Alistair Reid of First Draft News gave my favourite talk of the afternoon, which was all about verifying false information on social media.

Surprise surprise, any number of established media organisations have fallen for social media fakes and treated them as fact. These range from the Evening Standard and others reporting a story about a ‘killer clown’ being chased into a Cambodian minefield from the fake local news site The Times of Cambodia; to the BBC, New York Magazine (oh hi there) and Sydney Morning Herald reporting on an Australian man who had supposedly been banned from Facebook for using his legal name, Phuc Dat Bich – his real name was revealed to be Thien Nguyen.

A screenshot of an Evening Standard news article headlined 'Killer clown' dies after stepping on landmine in Cambodia, with a picture of a sinister-looking clown beneath it.

Having misinformation gain a wide platform and be spread quickly and uncritically is a hallmark of the social media age, but it isn’t just the fault of ‘citizen journalists’, and nor does it mean that citizen journalism in the social media age is bound to be a catastrophe. And it’s not that, as Singal writes, the “members of the community spreading these falsehoods” (which, as we’ve established, also includes professional journalists) “forcefully reject or are unfamiliar with the values of skepticism and fact-checking”. It’s more that these values aren’t always applied as consistently as they should be in the rush to be first; or that more training is needed with verification techniques such as reverse image search, using EXIF data and checking domain registration details.

As Reid pointed out in his talk, we rarely remember exactly who was the first to break a story, but we do remember the times when they got it really wrong.

Going back to the ‘Hillary emails’, Zeynep Tufekci has a great opinion piece on the New York Times about why Wikileaks isn’t whistleblowing, which talks about the dangers of the way that Wikileaks (and other data hacks) have dumped reams of information, unfiltered and uncurated, into the public domain with no regard to the consequences.

To me, the article just speaks to the need for journalists – citizen or otherwise – to be the ones who handle whistleblower information with consideration and discretion, making a call about what to release and what sensitive information to censor, just as when Edward Snowden released NSA documents to journalists like the Guardian’s Glenn Greenwald and documentary filmmaker Laura Pointras. Do that, and the stories get reported effectively.

I suppose the point that I’m making here is twofold; one, that if you release reams of sensitive data to the general public with no discretion, then of course the result is going to be a mess of speculation, hysteria and misinformation. And two, just because social media is often used to spread misinformation doesn’t mean that citizen journalism is doomed.

Putting across a reasoned, balanced point of view in the online hubbub is always going to be a challenge; everyone has a platform, and often the most sensational, exaggerated or baseless ideas are the ones that get all the attention. This is a problem faced by the journalism industry as a whole, as well as by politics, and any number of other sectors. It’s not limited to citizen journalists.

That’s not to say that citizen journalists can’t be responsible for social media hype or for the general spread of misinformation online, but it’s an internet-wide problem. It doesn’t mean the potential of citizen journalism is being ‘squandered’ any more than the potential of social media or the internet as a whole is being.