Using crowdsourcing to make lobbying more transparent in the French National Assembly

The partnership of Regards Citoyens and Transparency International France attempted to create a larger list of those seeking to influence by extracting information from the 1,174 official reports that Assembly members had published between July 2007 and July 2010 in relation to the preparation of legislation and government evaluation work. In some of these documents, Assembly members provided an appendix outlining all the formal hearings and meetings that took place in the preparation phase. However, such appendices were included in only 38% of reports. Nevertheless, the research team were able to create a dataset of 15,451 names from these documents. This already demonstrated a 160-fold increase in the number of identified individuals, compared to those in the official register.

The team then attempted to identify the organisation (and interests) that these individuals represented.  Sometimes this was actually recorded along with the individual’s name. However, many of the others were more difficult to trace – it would have been very time consuming and beyond the resources of the team to piece this information together.

Consequently, they decided to take a very different approach, mobilising the resources of the (online) community in a process of ‘crowdsourcing’, in which anyone with information could participate and link the identified names with their corresponding organisations more quickly.

What is crowdsourcing?

Crowdsourcing is the act of delegating tasks traditionally performed by specific individuals to an undefined large group of people or community (crowd) through an open call.

Watch this video explaining the process:

Using free software, the team created a user-friendly application which enabled crowdsourcers to link each of the names to an organisation. Once three users had validated a named lobbyist as representing a particular organisation, this would be added to the team’s records. The project team originally envisaged that piecing the information together using crowdsourcing would take several months. However, in reality the process was so successful that it took only 10 days, thanks to the contributions of 3,214 citizens.

Success factors

Simplicity – The layout of the online tool made it quick and easy to use, so that potential contributors were not put off by a technical interface. Also, the tool didn’t require registration, which enabled fast and open participation.

The incentive of competition (gamification) – For those prepared to register, the top 50 contributors to the process were identified in a ‘roll of honour’ and this competitive element ‘nudged’ citizens to try and link as many names as possible – with obvious benefits for the project

Keeping it fun and positive – The simplicity of the tool turned the hunt for names into a kind of ‘Easter egg hunt’, which was enjoyable for the participants. Also, the public-spirited nature of the activity, helping to shed light on the murky world of lobbying, was itself seen as highly rewarding.

The project therefore created a database of the 15,451 individuals, including their names, gender, function, and organisations. Although they had gathered these names from public records, the project team decided to publish only the names of the organisations, not the individuals, due to concerns about potential infringement of French privacy rights. In any case, the objective of the project was to create greater understanding of which organisations and interests were seeking to influence democratic processes.

The team then used software to refine the data and to categorise each organisation according to the interests they represented (e.g. corporation, union, NGO, and so on).  Initially they tried to make use of the EU’s transparency registry but soon found that this was an inadequate, due to limitations in the Commission’s categorisation system. They therefore created their own typology, distinguishing 8 categories, divided into 28 subcategories, as shown in the following table, and then analysed the data.

This project lifted the lid on lobbying within the French National Assembly. The team created a visualisation that presented the data in a way users could easily understand, as shown below. 

The project highlighted the inadequacy of the so-called ‘transparency safeguards’ that existed in the National Assembly. It showed that in 62% of reports to the Assembly there was no list of persons who had been met or interviewed in the formal hearings.  This means that the footprint of influence on the legislative process is unclear.

The number of organisations sending representatives to represent their interests, uncovered by the work of Regards Citoyens and Transparency International France, was 4635, 40 times greater than the 124 found in the official register. Since this information came from only 38% of reports with an appendix that gave names of individuals, the true figure is clearly even higher.

Moreover, the results indicate a gender imbalance amongst individuals that meet with Assembly members. Of the 15,447 individuals recorded in the formal hearings, only 24% were women (compared to 51.4% of the population). The only policy topic which provided an exception to this was (perhaps unsurprisingly!) gender issues, where women constituted 57% of those participating in the hearings.

Another valuable outcome of the project was that it reveals who is most listened to by National Assembly members overall. The data, in Figure 1 below, indicated that public sector organisations accounted for 48.3% of those participating in formal meetings held by Assembly members. (These public sector organisations only represent 20% of the organisations in the Register of the National Assembly).  Member organisations had the second highest frequency of formal meetings with Assembly members (20.9% of meetings), then private sector organisations (16.4%) and then civil society organisations (with only 7% of formal meetings).

Figure 1 - Distribution of the 9,302 meetings with National Assembly Members by organisation

The results show a very low number of formal meetings with professional lobbyists and think-tanks, both accounting for only 1% of known formal meetings. Nevertheless, these organisations are believed to exercise a significant influence in the National Assembly. This indicates that influence comes from far more sources than simply formal meetings and hearings. This suggests a need for improved recording. Members’ records should include a list of people with which they have meet informally as well as the traditional formal hearings – without this, the records of the lobbying process will be incomplete and suspicions of the process are likely to remain.

Figure 2 - Distribution within environment sector from 652 meetings