The voting method described in [Tangian 2017b] has been experimentally approbated during the 2016 election to the Karlsruhe Institute of Technology (KIT) Student Parliament [Tangian 2017c]. Under this election method, the voters cast no votes but are asked about their preferences on the policy issues which are declared in the party manifestos (like in voting advice applications, e.g. German Wahl-O-Mat). Then the degree to which the parties match with the electorate's policy profile is expressed by the parties' indices of popularity (the average percentage of the voters represented on all the issues) and universality (frequency in representing a majority), and the parliament seats are distributed among the parties in proportion to their indices. This way it is hoped to bridge direct and representative democracies and to make the latter 'more representative' and, respectively, 'more democratic'. The voters are no longer swayed by politicians' charisma and communication skills but are directed to subject matters behind personal images, ideological symbols and populist declarations. It is supposed that a method that focuses on properties of decisions proposed (e.g., political and economic implications of Brexit) can make vote more profound and responsible. Indeed, the 2016 experiment proved that the method can increase the parliament's representativeness. At the same time, it revealed that the critical point is the selection of questions by the electoral committee: they can be favorable for one party and unfavorable for another, or they can poorly discriminate between the parties, finally causing an equalization of sizes of the party factions in the parliament (regarded by some as the method's malfunction). In the given paper, we describe a similar experiment during the election to the KIT Student Parliament in July 2017, where the problems mentioned are tackled. The parties are asked to formulate the questions themselves and to answer all of them, including the questions by other parties. The collected 94 questions are then reduced to 25 using an optimization model aimed at contrasting as much as possible between the parties by maximizing the total distance between the vectors which characterize their policy profiles. The 2017 experiment confirms that the alternative election method significantly increases the parliament representativeness while avoiding the accusation of partiality in the question selection. However, the equalization of parliament factions is still persisting. Analyzing this effect, we find that the student parties' positions are insufficiently diverse to reflect voters' policy preferences, meaning that the student parties fail to consistently represent electorate groups, representing different sets of voters on each issue. In mathematical terms, the question selection based on maximizing the distance between the party vectors reduces their space location almost to a two-dimensional plane, which is inadequate to represent the multi-dimensional space of voters' policy profiles. To surmount this effect it is suggested to replace the actual question selection criterion by the one which enhances the multi-dimensional space location of the party policy profiles.
