Computing Veracity – the Fourth Challenge of Big Data

Journalism of the tools

by Christophe Bruttin

Launching the Pheme project in January resulted in global media coverage, mostly through the catchy (and slightly erroneous) echo of a “Lie detector for Social Media”. When it came to press interviews, two questions kept coming: the danger of getting a wrong result and the future of the profession if a robot could perform one of its core tasks. In short, a (mostly) tongue-in-cheek defence of human journalism at a time when automated journalism keeps making the headlines. This is unfortunate because attempts like Pheme may be exactly what is needed for journalists to be able to focus on their fundamental role.

We are getting out of an age of denial in the newsroom. Ten years (!) after “my readers know more than I do”, most media have started to listen and spend time and money digging into the complementary content trove growing under the “social media” umbrella. The evolution was helped by the arrival of convincing young journalists and successful pure players, but also by the survival of the old guard (yes, it is still possible to do great journalism without Twitter). Less confrontation led to more natural interest, which was bound to happen since the kind of information we get through these new channels is not so different than before.

We just get a whole lot more of it.

Ever expanding connectedness and technology, especially among the less privileged, means we are only going to get more and more information coming directly from the people involved in the news we are covering. “Social Media” – which represents a total democratization of practices rather than a specific set of tools – means that this information is coming fast, easily, and can become a conversation or even a dialogue. Increasingly, it also means that whatever the topic, it is possible to get in touch with a specialist who can help to better understand the facts. One may already be present in the conversation, defending or rejecting a specific statement.

This is where Pheme and similar tools come into play. We need all the help we can get to assist us: an army of customized Wall-Es to fetch the relevant bits, highlight potential frauds, make patterns visible and reduce the noise. Any journalist involved with online content already possesses a complex ecosystem of platforms and tools to manually do what Pheme will try to achieve by itself, from tracking the original tweet to guessing an account’s trustworthiness. This is not about replacing journalists, nor about defining what matters: this is about getting better data ready to be analyzed and spend more time where it’s required. Pheme won’t be 100% correct – and it is not one of the demands placed upon it. Even if the algorithm is wrong in its final verdict, the analysis of how the rumor was shaped and how it evolved will prove useful to journalists in finding new leads, protagonists or questions.

Sure, some conclusions stemming from the tool may be wrongly applied in editorial production – which hardly makes it original at a time when a lot of the newsroom discussion is centered on the implications of live measurements tools, trending tools, or on how many embedded gimmicks can be invented before performing storytelling seppuku. “Of the good use of tools” is already on the agenda to ensure that the final offer still is (good) journalism.

The case of Pheme is rather straightforward: a person tweets that the police is pepper-spraying students. Advocacy groups retweet it. Neighbours report an assault by a young mob. More people contribute with pictures and comments, and no one is doing journalism yet. If you are trying to piece together a clear, fair and accurate report of what is happening and why, adding context to the situation, you are, with all the support you can get.Nobody is asking you to trust the algorithm more than any of the participants, and artificial intelligence is quite far from asking that crucial question that had been missing from the conversation – and insisting on getting an answer.

P.S.: The grumpy old journalist, sick of constant changes, social media and collaborative possibilities, may want to look at it through another lens. By allowing journalists to comb faster through much more data, tools like PHEME may make it possible again to work within a new reality in the old-fashioned way : alone.

Christophe Bruttin is a member of the editorial board in charge of new editorial formats at, the international service of the Swiss Broadcasting Corporation and the journalistic case study for the PHEME project

Be Sociable, Share!

Comments are currently closed.