Detailed view of a printed circuit board.

How algorithms can promote participation

Algorithms influence our lives not only on Facebook, but in many other areas of life as well. Algorithmic decision-making (ADM) processes are used, for example, to observe and evaluate how often break-ins occur in certain neighborhoods so the police can deploy their resources accordingly. Insurance companies also use them to monitor how automobiles are driven. We have therefore developed a method which can be used to evaluate ADM processes to see to what degree they can affect social participation.

The potential social impact of ADM processes – more concretely, their impact on opportunities to participate – is a key characteristic for carrying out a differentiated analysis of those processes. To what degree can ADM processes influence whether or not each individual is treated equally and included when a society builds political will? To what extent do algorithms make it possible for each person to be part of social, cultural and economic developments?

A newly created tool can help provide answers to these and similar questions. The tool was developed on our behalf by Kilian Vieth and Ben Wagner, authors of the working paper Teilhabe, ausgerechnet (Calculated Participation). The tool makes it possible to analyze the potential impact of ADM processes that evaluate and judge people. The analysis is based on seven independent variables. The user awards points for the case in question, depending on the degree to which each criterion applies. One criterion, for example, is the role played by the operator of the relevant process. Does the operator enjoy sovereign power or dominate the market? If so, this criterion is said to have a high potential impact and is assigned a value of 2. Once all of the criteria have been evaluated, the values are summed to identify the potential impact on participation.

The process is a first attempt to make the debate about algorithmic processes more objective. The proposed classification method focuses on criteria in three areas: actors, inclusion in the social context, and potential implications for basic civil rights. Other focal points are also conceivable, and alternative ideas and suggestions are welcomed. Since this is an initial approach to the subject, the limited scope offers several advantages: Many of the assessment questions can be answered with relatively little research, and the results are readily comparable. Moreover, the criteria are designed to reduce complexity and increase flexibility. That means the scoring system is also well suited for use in workshops and as a basis for discussions during strategic consulting sessions.

In contrast to intuitive estimates, the tool's questions can be applied to a range of situations. The point of the analysis is not to determine how effective a given process is, but how it might impact participation – positively or negatively. The greater the impact, the more requirements must be put in place and the more effort must be made to ensure positive effects are exploited and negative effects, such as discrimination, are avoided.

The working paper is part of a larger exploration of "Participation in the Age of Algorithms and Big Data." An analysis of international examples of ADM processes and the discernible opportunities, risks and required responses has already been published.