Matti Nelimarkka, can digital systems have an effect on society?
– Public administration has, in fact, been using information technology as an important part of its operations for a very long time. Technology pervades everything we do in society. If you apply for unemployment benefit, you will encounter the Kela web system. Or if you participate in political discussion, you will increasingly use a computer, Facebook, Twitter or another platform as an intermediary.
All systems created by humans can be inherently political. Their design always involves value judgements, some of which are unconscious and some conscious. Consequently, systems can have an effect on society and how it functions.
How much power can coders have?
– The shocking truth is that we don’t know. Looking at the research conducted in the social sciences over the past 20 years, you see that algorithmic systems do have power. At the same time, the structure of that power remains, unfortunately, rather obscure.
For instance, when I entered the word ‘idiot’ in Google’s image search, it displayed images of Donald Trump. However, we don’t know whether this is the result of a coder’s practical joke, or whether a large number of people have linked the word ‘idiot’ to images of Trump, making the system draw a connection between the two.
Anecdotally you can say that coders indeed have power, but the reality remains unclear. In my research, I wish to gain a more in-depth understanding of how much political power coders actually have.
How can the power of coders be made visible?
– As a rule, prior research has observed retrospectively how information systems work, and then attempted to come up with insights on, among other things, the power of coders. I am interested in exploring the process by which systems are created – what kinds of power struggles are being waged and what kinds of interests are being emphasized.
The fact that it is challenging to keep track of causalities complicates any effort to make power transparent. For example, if Kela were to introduce a new hard-to-use online service, we on the outside would not be able to see why this happened. Of course, a malicious developer may have consciously made the online service difficult to use, but it may equally well be that time was running out and corners were cut – without this involving any grand scheme to wield power. To the user, the alternatives produce an identical effect: the online service is difficult to use.
How are coders monitored to ensure that future software complies with laws, regulations and other guidelines?
– We do have a range of rules, such as the Web Accessibility Directive of the EU, as well as a growing number of requirements associated with open data and criteria for data storage. However, no guidelines take a stand on how software should be engineered, with the exception of accessibility and ease of use.
In other words, someone could design the Kela system to veer toward the left or right, or closer to traditional family values or modern values, while making sure that it complies with the relevant accessibility criteria.
Software purchases are made by someone drawing up the requirements specification.
However, that is preceded by the legislature first enacting a law pertaining to, for example, unemployment security. After that, someone involved in administration interprets what this law means for the system and attempts to communicate the matter to others. Then, designers and coders try to implement, on the basis of this interpretation by the administrator, an information system that puts into practice this person’s understanding of the law. This means that several kinds of translation take place in the process. The final result may well follow the original spirit of the law, or not.
You have offered the idea of alternative operating systems. What are they?
– Political parties provide alternative budgets, that is, their own versions of the budget drawn up by the government, to enable us citizens to understand what different parties would like to do. I have wondered whether we should communicate on key information systems in the same way. For example, if we had a Kela operating system implemented by the green left, the right could present an alternative system, illustrating how they think it should work.
Each party would have their own design team shaping what digital solutions should look like. This could be important for democracy, since our citizenship is relayed through digital systems.
Your group has also studied views on digital solutions and environments across the Finnish political spectrum. What did you discover?
– We carried out a study (link in Finnish) in which we presented different user interfaces to politicians; different options for the same thing. The goal was to reduce political polarization and the quick-fire exchange of words taking place on social media. We received responses quickly, with the study subjects describing their ideas very naturally and openly. However, their responses reflected their party-specific worldview.
For Kokoomus (the National Coalition Party) and Perussuomalaiset (the Finns Party), freedom of expression was very important. Then again, Vihreät (the Greens) and Kokoomus endorsed the idea that peer-reviewed scholarly articles should be included in political discussion. However, reading such articles requires a high level of education, and their inclusion in the system would exclude a large share of voters. The responses by members of Keskusta (the Centre Party) highlighted community, including a type of village community.
Such research on the political elite can produce relevant knowledge on the political nature of user interfaces. Politics has a strong presence in any encounters between people and computers.
Why are user interfaces often difficult to use?
– For many reasons, but there are indeed a number of interfaces that require time and effort. The University’s system for travel invoicing is my personal experience of a poor user interface. Every time I pay for my expenses I end up wondering whether it is worth the hassle trying to invoice the University or whether the sum is small enough for me to just pay it out of my own pocket.
Poor user interfaces can also stem from, for example, not understanding what users really want to do. Then you end up creating a system that people just have to live with.
Designing well-functioning user interfaces is expensive. That users have different priorities and demands poses another big problem. They often want to do things exactly the way they have always done them.
Is ease of use the primary value of user interfaces?
– It depends a lot on what you want to do with the system and who its anticipated users are.
For example, if you think about launching nuclear weapons, the system may be designed to be complicated to force the user to pause to think several times during the process. Therefore, the ease or difficulty with which a system can be used may also depend on what people are expected to do with the system.
– There is a great deal of concern about surveillance capitalism, pursuing profit through the utilization of personal data, and what the right balance is between the common good and the economic good. But when people wish to put more emphasis on the former, they forget that the common good is not an unambiguous concept. Different parties and ideologies have very different ideas on what the right kind of common good is and the right way of making things happen.
Facebook has been calling for regulatory assistance from governments, but instead of solving the problem, that would turn it into an ideological struggle. Appealing to the government does not solve the fundamental challenge of having systems with a spectrum of values due to the pluralistic nature of society. You may come across values that make you think a mistake has been made, while these same values may be just fine from someone else’s perspective.
However, it is important to understand that all human-made systems can be political in themselves, which is why studying digital power is essential. Regrettably, powerful institutions often do not wish to be studied, but, then again, the gathering of data is not supposed to be easy and straightforward.