Data-driven journalism in the post-truth public sphere

This essay was written for the Retreat Conference 2016 organized at Athens, Greece, on September 24th, 2016.

Once limited to the fringes of society, lying blatantly has become a fixture of the public discourse. In this post-truth public sphere, datajournalism could spearhead the effort to reimpose truth as a positive value. Or it might just be a technique to serve this small segment of readers who care about the truth. This essay argues that what datajournalism can or cannot do is not that important, because the root cause of the current epidemy of lies is to be found in academia, not journalism.

The post-truth public sphere

When asked to justify one of the lies of the Brexit campaign, Michael Gove, then secretary of state for Justice, said that “the people of this country have had enough of experts and organizations with acronyms saying they know better”1. He did not comment on the lie itself. He is not alone in using lies in political campaigning. Donald Trump, a candidate for the presidential election in the United States, very rarely says the truth. The fact-checking website Politifact found that, out of 20 statements by Trump, only 3 were true (to compare, president Obama tells the truth in one out of every two statements)2. Vladimir Putin, in Russia, uses lies and disinformation so much that some observers called his politics a “war on truth”3. These lies are not the creation of a few lone politicians. They are part of a wider framework where media outlets echo them as if they were solid statements (I’ll come back to the role of journalists later) and other politicians debate them.

Lying has always been an integral part of public affairs. The Iraq war was started on a lie, and the Vietnam war before that4. The Greek government lied about its public accounts to join the euro and the French one lied about its involvement in the Rwanda genocide5. Lying is so important for statesmen and women that Niccolo Machiavelli wrote a whole chapter on the issue in his famous treatise on government, back in 1513 (chapter 18).

Lying is not new. What is new is the perception of lies. While Machiavelli wrote that lying was a necessity, he also wrote that the Prince should keep the appearance of telling the truth. Lying, he wrote, should be done “in such a way that no one become aware of it; or, if it should be noticed, excuses must be at hand to be produced immediately.” Truth was then seen as a positive value. In 1976 still, Jimmy Carter took great care to appear as the candidate of the truth. He polished and repeated his core slogan: “I’ll never tell a lie, I’ll never make a misleading statement”6. Of course he lied later on, but at least he was elected.

The novelty of the 2010’s is the disdain of public figures for the truth. Michael Gove, quoted above, justified lying not by producing an excuse immediatly as Machiavelli advised, but by insulting people who said the truth. Not only is truth considered unimportant, it is clearly marked as a value defended by the “other”: it is an undesirable value. The prime minister of France, Manuel Valls, shared Gove’s viewpoint when he said that “to explain [terrorism] was to excuse it”7. Valls clearly said that finding the truth, that trying to go beyond prejudice, was not welcome.

At the same time, politicians who refuse the truth do not oppose it with a series of lies. Instead, they argue that any statement, be it a well-researched report or an eructation by a ignorant, are opinions of equal value. This was most obvious in the Brexit example, when Brexiters readily admitted to having lied once victory had been won8. Several regimes, chiefly Putin’s Russia and Turkey’s Erdogan, now operate entirely in a setting where the notion of truth and lies has become of secondary importance9.

This value shift, which turns truth-seeking in an undesirable trait, is what characterizes the post-truth public sphere ; much more than the use of lies in the public discourse.

What datajournalism can do

In such an environment, what can datajournalism do? Datajournalism, after all, is all about using data to infer facts. It is about using the tools of social science “on a deadline”10. What can the purpose of data-driven journalism be if it is surrounded by a world where facts have become irrelevant?

Asking this question might seem absurd, considering recent spectacular datajournalism projects. The Panama Papers, and before them the Offshore Leaks, the Swiss Leaks and the Cablegate, show that datajournalism can have worlwide impact, both with the public and with institutions. Sigmundur Davíð, prime minister of Iceland, stepped down as a result of the Panama Papers. So did José Manuel Soria, the Spanish minister of industry, and many others. By bringing factual truth to light, the datajournalists behind this investigation took the role of a counterweight to their governments - one of the roles many journalists see as their own.

Of course, the Panama Papers, because it was a one-off leak, is not representative of the daily routine of datajournalism. While no study on the impact of datajournalism has been run, in part because academics do not agree on the definition of datajournalism, the related field of fact-checking has been well explored. At least 6 experimental studies were run on fact-checking operations (all in the United States). They show that fact-checking is broadly positive in that it is popular with readers, it convinces them of factual truths and it impacts the people whose assertions are checked11.

In an experiment with readers, researchers showed that fact-checking was appreciated and effective among all groups in the population, regardless of political views and education (though it worked better among Democrats and more educated participants)12. Another study showed that politicians who were told they about their statements’ being fact-checked told less lies than others13.

The tools of datajournalism can also be used to spread untruths. Data points are not facts, they are only given elements from which facts are derived14. Some people can use data to promote lies. When they do, they give their lies credibility. A study has shown that articles that contain graphs are perceived as much more credible than the ones that do not, even if the graph are totally unrelated to the contents of the article15. Another study looked at the power of deceptive datavisualizations and found out that they greatly impact readers’ perceptions 16. The tools of datajournalism can be used to create and propagate lies as well as truths. On balance, the total impact of datajournalism on truth might not be so positive.

Truth is a niche market

In the previous paragraphs, I showed that datajournalism - as a technique - could produce factual truths and have an impact, both among readers and institutions, even if the tools of datajournalism can be used against truth. Why, then, don’t publishers invest massively in datajournalism to give their news operations credibility and impact? Simply because the role of a publisher is not to serve the truth.

Associations of journalists in Europe put the quest for the truth high in their priorities. The first article of the Munich Declaration (the cornerstone of journalism ethics in Europe) states that journalists must “respect the truth no matter what”17. However, no institution is tasked with enforcing the principles of the Munich Declaration. Journalism is a self-regulated industry and, as we know from banking, self-regulation means no regulation.

Publishers and, with them, most journalists, do not work to serve the truth. They serve three possible and not mutually exclusive objectives: cater to the interests of a donor (which can be a public institution, a private one or an oligarch), sell attention to advertisers or sell content to customers. None of these objectives necessitate the production of factual truths. Let us look at these three objectives in details.

Serve the interests of a donor. Most media outlets in Europe are financed by oligarchs or state organizations. For journalists working there, producing truths that are uncomfortable to their donors can be fatal to their careers. Cases of censorship abound in oligarch-financed news organizations. The Lagardère Group, in France, regularly censors its brands18. Mediaset in Italy, News Corp in the United Kingdom and the state-run Hungarian media, among many others, have also been linked to cases of direct censorship. Needless to say, self-censorship is what matters most in such settings. In oligarch-sponsored newsrooms, self-censorhip lingers in every journalist’s mind and hampers the production of factual truths. (It does not stop it: Some of the media outlets that published the Panama Papers, for instance, were oligarch-owned.)

Sell attention to advertisers. Almost all news organizations in Europe derive some revenue from advertising, and a few even make a business out of it. For them, pleasing their customers can take precedence over the production of factual truths. The Daily Telegraph, in the United Kingdom, did not hesitate to kill a story about the Swiss Leaks when their client, HSBC, asked them to19. More importantly, many media outlets that sell attention to advertisers do not hesitate to produce falsehoods to gain more readers. The Daily Mail, also in the United Kingdom, makes a great business of publishing lies20.

Sell content to readers. A few news organizations live from the content they sell to readers. Not unlike the relationship some newsrooms have with advertisers, they, too, must please their customers. Valeurs Actuelles, one of the top 15 French newsmagazines and the fastest-growing one, regularly features lies that fit its readers’ point of view, for instance21.

None of the business models of news outlets put the production of factual truth first and even if journalists put it high on their list of duties in the Munich Declaration, they did not take steps to enforce it.

Factual truth is only a segment of a larger market. News outlets only care about factual truth insofar as their donors, advertisers or readers do. As I wrote above, the appeal of truth as a positive value is dwindling. Therefore, the effort media outlets put in producing it diminishes and truth-driven organizations become niche operations. Mediapart, an investigative outlets in France, boasts a very respectable 118,000 paid subscribers22 - still less than Valeurs Actuelles.

Datajournalism can serve this niche market and does it very well. Many small outlets have been created that do use datajournalism intensively. In the United States, Vox, FiveThirtyEight or Quartz are just a few examples. However, none of them aggressively markets its articles to an audience that does not already consider truth a positive value.

In any case, truth itself matters little to the problems of the world. As Hannah Arendt writes in “Truth and politics”, truthfulness cannot and has not contributed to changing the world23. Humans care about concepts, stories and feelings. Factual truth is not central to the political debate. Media outlets and journalists might provide elements that help establish factual truth, but establishing the truth is not their role or their mission.

Academia is the problem

Arendt ends her “Truth and politics” essay by stressing the role academia plays in the creation and safeguard of factual truth. She sees academia as a “refuge of truth” that even governments know they should not mess with - because hampering universties means reducing the rate of technical progress. More importantly, academics produce truth using the scientific method, ensuring that their facts are not based on testimonies or documents (as journalistic facts are) but on reproducible methodologies. The practice of peer reviewing ensures the correctness of logic and of the data.

The academia that made sense in Arendt’s time is now crumbling under the weight of political interference, lack of funds, a broken publishing process and corruption.

Political interference. To ensure that academics worked on topics that are of interest to the industry, governments stopped funding scientists unconditionnaly. Instead, starting in the late 1990’s, they set up tenders on specific topics, to which scientists must apply if they want to receive funding. All major funding systems in Europe work like this, from the Framework Programme of the European Union (FP7, Horizon 2020) to the French Agence Nationale de la Recherche or the German Deutsche Forschungsgemeinschaft. The logical outcome of such processes is that scientists have to follow the trends set by politicians. A scientist specialized in a given topic might have to shift her interests wholly or in part if she wants her application to receive funding. Researchers can take decades to establish a factual truth. The current funding process ensures that research remains shallow in most domains.

Lack of funds. The combination of the massification of higher education with dwindling state resources put universities in untenable financial situations. Some of them are crumbling. Literally crumbling: The French even set up a Tumblr, Universités en Ruine, to document this development. More generally, the lack of funds pushes people with ambition, exceptional skills or who simply need a decent salary away from research and academia, ensuring a vicious circle that continuously lowers the attraction of a career in research.

The publishing process. The lower costs of digital publishing led to the creation of thousands of new academic journals. Many of them accept any paper in exchange for payment, regardless of their quality24. The trends towards publishing more content to collect more fees is not limited to open-access, newly-created journals. Once-reputable journals also accept papers that are obiously gibberish, without oversight25. Using the credibility of academic journals to spread lies is very easy. Even when results are checked, the over-reliance on p-values enables p-hacking, a way for authors to ensure that they come up with statistically significant results. John Bohannon, a science journalist, showed it when he managed to have newspapers across the world run headlines on a fictional “chocolate diet” in 201526. The problems of academic publishing have been known for decades. By refusing to address them, academia ensured that its informational backbone became unusable.

Corruption. The documents released by the cigarette industry as part of the 1998 Tobacco Master Settlement Agreement showed that tobacco companies paid scientists regularly, directly or indirectly, to influence their research27. Newly-released documents show that the sugar industry could have worked on the same scale to shift the blame for obesity on fat28. While measuring corruption in academia is hard, I have not found a single case of an academic fired or blamed after having taken money from the industry that influenced her research. Moreover, university across the world keep welcoming funds from the industry, even when conflicts of interest are clear.

It is no exaggeration to write that academia has lost its status of “refuge for truth”. By failing to address its multiple crises and de facto reliquinshing its core mission, academia has paved the way for the post-truth public sphere. It might even have created a void.

It is no wonder that datajournalists start publishing methodologies that look just like academic papers. ProPublica, a non-profit newsroom based in New-York City and specialized in datajournalism, published in 2015 the details of how it investigated the track record of each surgeon in the United States29. The document is not peer-reviewed and is authored by an editor and a journalist, not by scientists. But it has the same structure and visual aspect as an academic paper. The message is clear: Datajournalists can do research that is at least as good as the one produced in universities.

Such projects are good attempts at fixing the broken ecosystem around the production of factual truth. But let us not be fooled by their potential. There are, at most, a few hundreds datajournalists capable of producing such work in Europe, against dozens of thousands of academics. Datajournalism will not have much impact on the relationship most people have with the truth, because datajournalism is too specific a field. Training datajournalists is hard and only a few dozens graduate each year. If datajournalism one day leaves the niche it currently caters to, it will be because academia has fixed itself. Until then, it will have little impact.


In case you want to read my next essay in your e-mail inbox, type you email below and you'll be all set.


1. The lie in question was the infamous figure of £350m a week Britain was sending to Brussels. Read Brexit Campaign’s Gove Denies ‘Project Lies’ Claim in TV Special for the details.

2. Check out the Politifact website.

3. Read Emperor Putin’s War on the Truth.

4. The government of the United States and Britain faked documents to justify the Iraq war, such as the famous Iraq Dossier. The Gulf of Tonkin incident, which was used to drive Congress to war, involved a deliberate falsification of an attack by North Vietnamese boats.

5. Here’s an article about the Greek lies and here’s an article about the French lies.

6. Read the 1976 New-York Times article The selling of a candidate.

7. Find the original quote, in French, in this article. He later said that trying to understand was actually OK but explaining wasn’t - whatever that means (read here).

8. Read Nigel Farage Admits £350m Saving For NHS In EU Contributions Slogan ‘Was A Mistake’

9. Read Art of the lie at The Economist.

10. “Social science done on deadline”: Research chat with ASU’s Steve Doig on data journalism

11. The Hardest Check is the closest thing to a meta-study you’ll find.

12. Read Estimating fact checking effects.

13. Read The effects of fact-checking threat.

14. On the difference between data and facts, read my essay on the definition of datajournalism.

15. Read Blinded with science: Trivial graphs and formulas increase ad persuasiveness and belief in product efficacy.

16. Read How Deceptive are Deceptive Visualizations?: An Empirical Analysis of Common Distortion Techniques.

17. Read the Munich Declaration of the Duties and Rights of Journalists.

18. Read here a case from 2007 at Le Journal du Dimanche, here another one from 2016 at Europe 1.

19. That’s what Peter Osborne wrote when he quit. Read Telegraph’s Peter Oborne resigns, saying HSBC coverage a ‘fraud on readers’

20. Read My Year Ripping Off the Web With the Daily Mail Online

21. On the rise of Valeurs Actuelles, read Valeurs Actuelles : enquête sur une extrême droitisation.

22. The figure is not audited and Mediapart has a history of counting subscribers creatively, but still.

23. Read page 12 of the 1967 essay.

24. Science ran a study and showed that half of 300 journals surveyed accepted an obviously fake paper: Who’s Afraid of Peer Review?.

25. Read How computer-generated fake papers are flooding academia.

26. Read I Fooled Millions Into Thinking Chocolate Helps Weight Loss. Here’s How..

27. Read chapters 16 and 23 of Golden Holocaust: Origins of the Cigarette Catastrophe and the Case for Abolition by Robert Proctor.

28. Read How the Sugar Industry Shifted Blame to Fat.

29. Find it here: Assessing surgeon-level risk of patient harm during elective surgery for public reporting.


Want to raise an issue about this post? Please open a new one on Github or make a pull request directly, but make sure to read the rationale behind this blog first.