“Data journalism is so much more than beautiful visualizations”
In this year’s US election primaries, Nate Silver’s Fivethirtyeight is providing a very fresh perspective on journalistic reporting. Silver’s team not only developed new polling models for the respective state’s primaries, but also contributes a data-driven perspective in numerous well-researched articles. In Germany, such a journalistic voice is currently still absent. What does that mean for journalism in Germany? We talked to Eva Wolfangel (@evawolfangel), who is currently doing research on new tools to empower investigative journalism, about data driven journalism, algorithms and the road ahead for the journalistic profession.
What’s wrong with current journalism?
Journalism has been massively transformed by the internet. With a few exceptions, print circulations has been decreasing, while online-only news formats are on the rise. A vast share of the younger population is getting the news primarily from newsfeeds, be it Facebook or Twitter. This new era of journalism is dominated by traffic-driven clickbait business models, specifically designed for the short attention spans of consumers. “Very few publishers can actually afford to employ an investigative team, which can research for days or weeks”, explains Eva Wolfangel.
Journalists are measured in terms of output, sometimes scheduled to work on “live blogs”; being the first to report is oftentimes more important than being accurate. “One of the main problems why investigative journalism has been on the decline is that there are barely any adequate tools, which fully utilize modern technology and could help journalists to conduct investigative research faster”, elaborates Wolfangel. If investigative journalism is too expensive, then one key strategy would be to focus on bringing its costs down. Together with Jonas Kuhn and Andre Blessing, researchers from the University of Stuttgart, Wolfangel now works on a tool which could eventually assist journalists to track down the influence of lobbyists in the legislative process, thus contributing to more transparency and democracy.
DebateExplorer – who is actually writing our laws?
Lobbyism, the effective voice of special interest groups, is a well known component of the political process. In Germany, multiple hundred lobbyists are officially accredited at the German Bundestag. Their real influence on the legislative process, though, remains shady. “Currently, the only way to track down lobbyism, is to manually read through all parliamentary debates and compare that to special interest group positions. Obviously that is not really feasible”, describes Wolfangel the current state of journalism. “So the only way to actually uncover direct special interest influence, is to get a direct hint from someone within – and that does rarely happen.”
Given this setting, any kind of automation would greatly benefit the journalistic profession. The challenge, however, is not trivial at all. Wolfangel and her team essentially need to come up with an algorithmic framework, which captures and traces the similarity of arguments, meaning and supportive attitude in speeches, documents and newspaper articles. How does this look like in practice? “In a first step, we are only looking at transcribed Bundestag debates, which we cluster around certain topics”, explains Wolfangel. Wolfangel then annotates sentences within those clusters to help the computer understand, whether a certain sentence is, for example, supportive of a tax increase or not. “That is actually quite a tough endeavor”, reflects Wolfangel, “because political rhetoric has its own kind of tweaks. For someone with regular exposure to political debate, it might be easy to determine the tendency of an argument. But it is sometimes extremely tough to figure out, whether I should tell the computer that someone supports an issue, even though large parts of the paragraph are dedicated to the repetition of the counter-arguements.” Double negation, repeating arguments just to deny them afterwards – politicians are rhetorical experts, which increases the difficulty of the project significantly.
In a first pilot, Wolfangel’s team – which is supported by the Volkswagen Stiftung – focusses on the hotel tax, a well-known law in Germany which was always perceived to serve primarily a certain special interest group. “We are not yet at the stage, where we can present something that actually works, but already the process has proven to be quite enlightening”, says Wolfangel. At a recent talk at re:publica, Wolfangel received positive feedback regarding her approach and its potential upsides for journalism.
Data-driven journalism – quo vadis?
Building the tools to enable investigative journalism is one thing, but the presentation of results is the other side of the evolution of journalism. Data-driven journalism has become a buzzword, which is used by many to capture the trend towards increased usage of data points, data sets and visualizations to support arguments. Nate Silver’s Fivethirtyeight in the US is one of the prime examples of data-driven journalism. “In Germany, however, there seems to be a lot of skepticism towards the increased usage of algorithms and big data in journalism”, says Wolfangel. “However, that is still not an argument not to experiment with it”, she adds.
In Germany, data-driven journalism very rarely describes innovative data-driven research, but is rather a term used for (interactive) visualizations. “Sometimes I have the feeling that a nice bar plot is already called data journalism”, says Wolfangel. The innovative potential of data journalism (even just as a supportive element) still has not arrived in the minds of the publishing executives and editors. “We need to develop a better culture for work and analysis with data and predictions”, formulates Wolfangel. While it is essential to praise the benefits, a solid skepticism is certainly always appreciated and will only contribute to a more educated and holistic discussion about data journalism. What’s the road ahead? “Well, data journalism is so much more than beautiful visualizations. I hope we will fully tap its potential in Germany as well.”