CoCoRICo-CoDec
CoCoRICo

WP4

WP4: Development of a Collective Decision Making Platform, and Promotion of its Uses

Coordinators: Sylvain Bouveret and Vincent Merlin

The goal of this work package is to implement and test the algorithms and protocols worked out in WP1, WP2 and WP3 and make them accessible to the users (citizens and decision makers). It aims at promoting new forms of democratic participation in low-stake situations, by enabling citizens to express precisely their preferences, and by giving them tools that should help them make better common decisions. The platform aims at being used as a pedagogical tool in events such as Fête de la science, and used to perform research experiments (in laboratories or in decentralized contexts).

This work package has three main objectives:

  • Developing and testing the software platform

As a result of the advent of computer networks and connected interfaces, people have the opportunity to participate in an increasing number of collective decision making situations. Social choice theory provides a set of principles, rules and methods in this context. However, rather surprisingly, there is an obvious lack of practical tools to help people make collective decisions in everyday low-stake situations.

One of the key outcomes of our project is to bridge this gap by developing a large-scale platform for various tasks of collective decision making. Currently, the most well-known application for everyday life collective decision making is probably the poll application DoodleTM, which allows people to express their preferences, and allows the decision maker to have a survey of these preferences. However, these systems soon become unsatisfactory as soon as the individual preferences are more complex and subtle than just approving/disapproving some options (or remaining neutral), and hence fail to give satisfactory answers in most collective decision making situations. To overcome these limitations, the part of the team located in Grenoble has developed an online voting platform, Whale. With Whale, users can currently create open or sealed ballot polls, cast their votes using expressive preference modes (orders, utility functions,...), and use classical voting rules to figure out which issues are the best ones among the candidate issues. Although Whale improves upon DoodleTM, many important decision making mechanisms are not currently implemented.

During our project, the Whale platform has been completely rewritten into a new version, Whale 4 (https://whale.imag.fr), with many new features:

  • Integration of new voting rules (such as Range Voting and Majority Judgement) and settings.
  • Adaptability to various kinds of devices like smartphones and tablets.
  • Possibility to provide data sets of the elections in various formats, including CSV, JSON, and PrefLib.
  • New information visualization techniques, providing people with easily interpretable information and, in the end, helping them making informed collective decisions. In particular, we used the Edge-Compressed Majority Graph technique for the visualization of the majority graph of a preference profile.
  • Using the Platform for Experiments in Laboratories

In 2017, for the fourth time, several experiments were conducted by a team of researchers on the occasion of the first round of the French presidential election. CoCoRICo-CoDec played an important role in these experiments. Three of the seven organizers of the “Voter Autrement” operation are part of the ANR project: Sylvain Bouveret, Jean-François Laslier, and Isabelle Lebon. This operation aimed at testing other voting rules than the one used for the actual election. It was composed of two series of experiments.

A first series of experiments, in situ, was held on April 23, the day of the first round, in sixteen polling stations in five French towns: voters were invited to test different voting rules after exiting the official polling station.

A second series of experiments were conducted on the Internet during April 2017: 37 739 volunteers agreed to participate before the first round and between the first and the second round, to test three voting rules (randomly selected) and to answer questionnaires helping to analyse the results better. For this, a specific platform was developed: https://vote.imag.fr/. This online experiment, open to everyone, allowed for testing various voting systems: each visitor of the website was invited to say how they would have voted if the President of the Republic were elected according to such or such system. Each participant was asked to test the following systems: (1) voting by note or by approbation; (2) voting by ranking (variants of Borda and IRV); (3) a second stage of voting by note or approbation, using a different scale than at stage (1). After that, each participant was asked to express their opinion on the different candidates on a continuous scale with the help of a cursor, and to express their preferences on random pairs of candidates. Finally, each participant was asked to fill a questionnaire about their political views and sociological characteristics.

A major conclusion drawn from the two series of experiments, and especially from the online experiment, shows that it is possible to test various voting systems on a large scale, since it gathered 37 739 participants. Two interns funded by the projet helped preparing the raw files of data obtained from the in situ experiments and analyzing them.

The raw data of the online experiments, together with a companion paper explaining the experiment, have already been made accessible to the public on the Zenodo platform. See https://doi.org/10.5281/zenodo.1199545. The same procedure is on its way for the in situ data, which will be also be made available soon on Zenodo.

  • Testing and Using the Platform for Pedagogical Uses and Dissemination
    • April 2016: Participation to the dissemination contest Relais d’sciences, Le Dôme, Caen (http://www.ledome.info/). During this contest, which lasted one day, we had to discuss about Whale and social choice with a group of laypersons which were invited to participate. The aim of the discussion was to build the project of a workshop or exhibition to be shown during the science fair (Fête de la Science) in October. Three other teams of scientists were involved in the contest. Even if we did not win, this day was very beneficial to us because it was the occasion of having fruitful discussions about how social choice can be beneficial to real people in everyday life situations. Also, this day helped us build a mini scenario we could use to introduce Whale to an audience, and then ask them to participate online to collective decision making experiment. The scenario was designed to be a very simple, comprehensive, and fun for all the audiences: we ask a group of participants to decide together the features of a movie they’d like to see together : first they had the choose the kind of movie (Biopic, aventure, comedy, etc.), secondly the main actor/actress, next the place of location, and at last an object that would be present in the intrigue. Thanks to Whale, at each step, we introduce to the public new ways to report the preferences (yes-no ballots, grades, rankings) and several ways to aggregate these information. At the end of the experiment, we present to the public the poster of the fake movie they have decided to see together. Hence, with this fun scenario, we are able to discuss with the public the importance of the different ways to make decision, and make them manipulate Whale. For more details, see: Movies.
    • Participation to Fête de la Science, Le Dôme, Caen, October 2016. We were involved in two different actions. The first one took place on Thursday, October 13. The aim was to present Whale 4 to a group of laypersons, discuss about its features, and try to invent new use cases for it. The aim was to raise new ideas of real applications for this kind of platforms. The second action took place on Saturday, October the 15 and was regular science exhibition where people were able to experiment and manipulate social choice concepts with Whale 4, using the scenario on the choice of a movie developed earlier that year. With this fun scenario, we are able to discuss with the public the importance of the different ways to make decision, and it also proved to be a nice opportunity of seeing Whale 4 used in a real situation.
    • The scenario we developed during the science contest in 2016, and presented afterwards in la Fête de la Science in October 2016, was used several time in 2017 and 2018 to present Whale: again during the Fête de la Science at Le Dôme on October 2017, at a workshop in Current Issues in Economic Design, CNAM, Paris on November 14th 2017, at seminar in Caen on the 15th of December 2017, during an encounter between high school students and researcher in Rennes on January 29th 2018, and before the municipal council and civil servants of the town of Saint-Etienne-du Rouvray (76800) on April 13th 2018.

All these presentations and experiments in front of diverse attendances proved that the Whale platform was easy to use with various technologies (computers, tablets, cell phones).

Media

Finally, our experiments in situ and on the internet when the presidential election held attracted significant attention from the media: newspapers (among others, Le Monde, Libération, Mediapart, le Dauphine Libéré, La Croix, Huffington Post), radio (France Inter, France Culture, France Bleu Isère, RCF Radio, Media journal/Radio Tendance Ouest), and television (Télé Grenoble, C News, France 3 Auvergne-Rhône-Alpes, France 3 Région). For more details, and links, see https://vote.imag.fr/about#media.