A few weeks ago I had the pleasure of attending the excellent PhD defence of Holly Robbins at TU Delft, and afterward participating on a panel on the related theme of “Designing for Legibility of Computational Black Boxes” along with Thomas Binder (The Royal Danish Academy of Fine Arts), Marcel Schouwenaar (The Incredible Machine), Irina Shklovski (IT University of Copenhagen), and Ron Wakkary (Simon Fraser University).
Here are my opening remarks/provocation related to the theme:
While digital technologies have brought many kinds of possibilities, changes, and even ‘revolutions’ of many kinds, the emerging form that Johan Redström and I call fluid assemblages is particularly striking. Many of the things we now use in everyday life are fluid assemblages that do not have only physical forms, whether a music-playing app that does what a record player used to do plus much more or other kinds of things that have digital capabilities and connectivity added (such as connected speakers or that longstanding dream of ubiquitous computing, the refrigerator that orders milk). Whereas purely physical ‘older’ things were relatively stable and predictable, fluid assemblages are dynamic, contextually configured, and continually updated. They also have multiple users and kinds of use, and participate in larger assemblages of which they—and their users—are component parts.
It is possible to highlight some of their main features and challenges for design in relation to the idea of opening up the black box of computation. Specifically, a key problem is that the box is not a box.
Boxes can be opened (binary: open/closed), and this is what enables access to what is inside.
Boxes have clear boundaries that contain and exclude.
Boxes can be clearly identified as boxes.
The bigger the box, the greater the complexity it can conceal (more or less).
Boxes do not change their contents while closed.
Fluid assemblages can appear to fit in boxes, but this is an illusion.
The box of computation is not at all like a box because it is a fluid assemblage.
Fluid assemblages cannot be opened—they must be rendered legible in some way that will always be limited. Something always withdraws (hat tip to object-oriented ontology).
Fluid assemblages have no clear boundaries. They are connective and impure, always capable of assembling and being assembled in new ways. And this assembling does not have to be visible.
Fluid assemblages cannot be clearly identified or pinned down to a specific location or time. They manifest in a variety of ways, but none definitive.
Size and appearance of things that are fluid assemblages in terms of how they manifest as things for use has very little relation to their level of complexity. Think of the latest smartphones: they are basically like sheets of glass, yet the things they do are highly complex, and can have very real consequences in the world.
Fluid assemblages are constantly changing. These changes can be rendered legible (for example, through a new version announcement) but can also be hidden (for example, as A/B testing done as a matter of course without explaining to users which version is being tested on them and which metrics are being assessed).
Fluid assemblages may appear to fit in shiny black (or white) boxes, but they are in no way contained by them in any way that we have been used to until now.
In conclusion: To open up the black box of computation, I suggest that we need to think outside the box and instead enter the assemblage.