Last September, the French government appointed Henri Verdier as its first “Chief Data Officer”. Four months after his appointment, LSE student Marine Gossa looks back at the debates that Verdier’s new function sparked in France and what a Chief Data Officer could mean for French democracy.
Working under the Prime Minster’s direct orders and within the Ministry for Public Action Modernization, Verdier has two main challenges to face in coordinating public data collection and use: enhancing the State’s accountability (open data) and efficiency (data governance).
Open data and the risks of a ‘one-way’ accountability
‘Open data’ refers to the idea that institutions should be sharing the data they have collected with the public so as to ensure they are accountable for their actions. In France, the government has been working on a new platform, www.data.gouv.fr, where it is publishing several data sets on topics as diverse as employment rates for master’s students, European subsidies obtained by French farmers, and political parties’ finances. This platform celebrated its first anniversary on the 18 December. Proof of the interest it has raised abroad, data.gouv.fr has been featured in the Open Government Partnership Awards and ranked first in Europe by the 2014 United Nations E-government Survey.
However, such a huge amount of data needs to come with the right format and the right tools to be understood by citizens and thus, make politicians accountable. Last July, the release of French MPs declarations on their finances -one of President Hollande’s key proposals for French administration transparency – raised a huge debate. These documents are filled out by the MPs by hand, making some of them almost illegible, according to Le Figaro. A group of citizens, Regard Citoyens, decided to transform these data through crowdsourcing methods to make them useful for the general public. 8,000 people helped to release a computerized version in just a week, evidence of the interest among citizens for such information and transparency.
As Tech Crunch also pointed out when the data.gouv.fr platform was created, “raw data is not enough if you don’t know how people are using it”. Verdier must make data accessible to the public – through an easy-to-use website proposing access to data under open license – and promote its use among French administrations through cooperation in between Ministries.
Data governance and its implications for democracy in France and elsewhere
The second challenge for French Chief Data Officer is the State’s effectiveness through a better approach to data governance. As Verdier pointed out in an interview to NextInpact.fr, data is useful to “make better decisions, design better public policies and be more efficient”. The idea behind this claim is governance by algorithm. Best known for its use by web companies such as Google and Facebook to tailor their services according to data collected on their users, ‘public’ algorithmic regulation uses citizen data collected on the benefits they receive, the taxes they pay, etc. to ensure more efficient, rational public policies.
Some welcome this idea of ‘regulation through algorithms’. This could eventually lead to more efficient resource allocation, for instance through the so-called smart cities as experimented with in Amsterdam for parking management. In France, there has been a long tradition of using data for improving public policies. The creation of the French National Institute of Statistics (INSEE) in 1946 helped the government to plan economic recovery after World War II by collecting data on the country’s growth, (un)employment rates and other important demographic data to improve economic panning. Thus, governance by algorithm only differs by the amount of data collected and the new possibilities offered by new technology.
But opponents have serious points to raise: in the United Kingdom, data sociologist Evelyn Ruppert has pointed out that data is not natural but constructed by collection methods and design. Thus, biases exist according to what public bodies record – and what they put aside. Essayist Evgeny Morozoz illustrated another concern in an article published by The Guardian. To him, algorithmic regulation could be used to prevent certain behaviours, based on predictions. So he asks, do we need law anymore? And we can go even further: in a world where our data is used by governments, polices and corporations so as to ‘best’ suit our needs, is there any room for individual choice?
At the crossroad of future democracy
In response to these opposing visions, Verdier published an article on his blog fueling the debate further. The question, he writes, is not whether we want algorithms or not, but how do we want to organize their use? What government roles are we willing to let algorithms be in charge of? Some journalists in France echoed these questions. Political daily Liberation published an article exploring how the administration could restore citizens’ trust through data sharing and efficient use, and business newspaper Les Echos listed the different ways data could help reduce spending through tailored and more efficient services.
Other questions need to be taken into account. As Joris van Hoboken and Natali Helberger have pointed out, use of algorithms should be preceded by asking who is liable for the automation and data collected, and how do we hold them to account. The first step taken by a government willing to use big data to improve its public policies should be to explore how it can be ethical and respect transparency principles.
A line has to be drawn between algorithms used as facilitators for public policy purposes, and algorithmic automation of decisions. This could go in two different ways, and as the first European state appointing a Chief Data Officer, France could become a positive or negative case study depending on the outcomes of its experiments. The key to success lies in the individual’s involvement and the state’s. It is as much citizens’ as governments’ responsibility to work towards an open, well-governed ‘data democracy’, for the greater good.
This article gives the views of the author, and does not represent the position of the LSE Media Policy Project blog, nor of the London School of Economics and Political Science.