What do algorithmic systems and bureaucracy have in common?
I gave on algorithmic systems, strategic interaction, and bureaucracy in the Making Sense of Algorithmic Systems symposium at the Annual Social Psychology Conference in Helsinki on November 18, 2017. The talk lays out early ideas in a domain that is (relatively) new for me. These have been developed in collaboration with Matti Nelimarkka, Jesse Haapoja, Juho Pääkkönen & others – but all mistakes are mine.
To accompany the slides above, here are the key ideas from the talk:
What might post-interaction HCI (Human–Computer Interaction) look like? This is a conceptual shift we are grappling with and trying to make sense of – focusing on direct and observable interaction between one individual and one device feels less and less sufficient (although those interactions, too, remain important). Inspired by Alex Taylor’s thoughts, I like to think of city bike systems as one example that pushes us to think about socio-technical systems and data in new ways.
The more we talk about algorithmic systems, the more we need to ask ourselves to be precise about how exactly they are different from socio-technical systems more broadly. Algorithms, data, artificial intelligence and machine learning are words I’ve heard awfully often this year — and there are problems with how they are used in public (and academic) conversations. There is lots of fear-mongering as well as moments when systems are narrated to hold more power and capabilities than they actually have etc.
One things that seems to be clear is that all things digital and the datafication of everything is attracting a lot of attention in a variety of fields – and critical researchers are already on it, too! There has been a proliferation of critical studies of algorithms and data over the past years. This reading list, collected by Nick Seaver and Tarleton Gillespie is one fantastic place to start from if you’d like to get a glimpse of what is going on. Moreover, we need to keep asking questions about what algorithms are and in what way(s) they are interesting. One important observation underlying the shift to talk about algorithmic systems rather than algorithms on their own is the fact that algorithms don’t exist in isolation. On this account, I recommend Algorithms and their Others, written by Paul Dourish.
Another source of inspiration for me has been this popular piece on the similarities between bureaucracy and algorithmic systems: Rule by Nobody. The analogy does not work 1:1, of course, but there is something to it. And this points to where I think social psychology has an opening to step in and speak up: our field has a lot of expertise on social interactions (also strategic ones) and organizations. These are needed in conversations about algorithmic systems.
For theoretical bases to work on algorithmic systems and strategic interaction, I recommend as a less known book by Erving Goffman, Strategic Interaction. It is a microsociological take of game theory! As I see it, there are (at least) two levels worth thinking about here: First, computer-mediated communication, including questions about how does social interaction play out in the context of algorithmic systems and how do individuals and groups use these systems in strategic ways in interacting with others? Second, human–computer interaction, with questions about how individuals and groups ”game the algorithm” and work around systems that are making it hard for them to accomplish their goals. Here, one might think about Uber drivers strategizing with one another (and against the company and its app) to make more money, but also about the kinds of workarounds that have long been observed as part of the ”normal” repertoire of how people make socio-technical systems work. Goffman’s work gives us tools to consider how individuals can interact with algorithmic systems (and with one another in the presence of these systems) in active, purposeful ways, rather than the dopes fooled by black boxes that popular accounts sometimes make us to be! But we need to be careful in considering what we can take from this work, focused on rich interactional settings (face-to-face).
When it comes to algorithmic systems and bureaucracy, Max Weber’s scholarship is one obvious candidate to work with. I, however, am intrigued to revisit Michel Crozier’s work, especially the book The Bureaucractic Phenomenon, as a resource for thinking about interactions with algorithmic systems. Crozier’s work challenges perspectives that overemphasize the rational organizational structure of bureaucracy, and places emphasis on the strategic efforts of different stakeholders within these organizational systems. Looking at algorithmic systems from this point of view allows for analysing strategic interactions on the system level in a manner that does not do away with the impact of networked systems but also keeps us focused on the possible tensions between the different human actors. Here, too, we need to be careful in applying old tricks to a new show, since as Minna Ruckenstein pointed out in the symposium, the rules in bureaucracies are typically public knowledge whereas this tends not to be the case with proprietary algorithms.
(Finally, while this talk deals with another domain, most of my recent research deals with the so-called sharing economy. If you’d like to hear more, I’d be happy to hear from you. For my academic papers, take a look at my Scholar profile.)
Rajapinta 