Beyond bugs and features: A case for indeterminacy

Spandrels of San Marco. [CC License from Tango7174]
In 1979, Harvard professors Stephen Jay Gould and Richard Lewontin identified what they saw as a shortcoming in American and English evolutionary biology. It was, they argued, dominated by an adaptationist program.[1] By this, they meant that it embraced a misguided atomization of an organism’s traits, which then “are explained as structures optimally designed by natural selection for their function.”[2] For example, an exaggerated version of the adaptationist program might look at a contemporary human face, see a nose, and argue that it was adapted and selected for its ability to hold glasses. Such a theory of the nose not only ignores the plural functions the nose serves, but the complex history of its evolution, its shifting usefulness for different kinds of activities, its mutational detours, the different kinds of noses, and the nose’s evolution as part of the larger systems of faces, bodies, and environments.  So how should we talk about noses? Or, more importantly, how do we talk about any single feature of a complex system?


In their critique, Gould and Lewontin drew an analogy by between non-adaptive traits and the spandrels of the Basilica San Marco, in Venice. Spandrels are the triangular sections that form at the corners of arches. If you are building a dome held up by four arches, you will end up with a spandrel on each corner.[3][architecture pedants: see footnote] Spandrels, in other words, are not a “selected-for” trait; they are the necessary byproducts, or “free riders,” of the intended architecture of the dome. Yet, over time, spandrels were co-opted for artistic purposes, developed their own aesthetic functions, and established new generic conventions. Spandrellic ornamentation, in other words, emerged as an art form out of the latent area created by the negative space of the archway.


A spandrel. From: me.

The spandrel serves as a useful analogy to describe some inevitable dynamics of complex systems: what appear as primary features may in fact be the byproducts of some other systemic element; these byproducts of design features are nevertheless potential resources for future co-optation; and the often arbitrary division between primary cause and secondary byproduct obfuscates how systems (biological, architectural, technological) develop, over time, in ways that cannot be so neatly atomized.


The view from the corner

I have neither the expertise nor the desire to weigh in on the implications of spandrels for evolutionary theory. However, Gould and Lewontin’s radical analogizing is an opportunity to test our own thinking about how technical systems emerge and develop through time. That is one caveat, here is another: media systems are not organisms and they do not “evolve” in any way that would (or ought to) make sense to a biologist—despite what some have argued. The kinds of evolutions we might be interested in are marked by the politics of difference, access, and aesthetics. They are shaped and warped by power and control. Yet, many media scholars can agree with Gould and Lewontin’s approach to evolution by treating the history of media systems as integrated materializations of the complex pressures of constraint and exploitation.

Gould and Lewontin were not sending young biologists out to find and categorize the equivalent of arches and spandrels in biological organisms. (In addition to being beside the point, such a determination is only potentially possible where a sizable enough fossil record exists). Likewise, advocating for a strictly “cause” and “byproduct” approach to technological history would be misguided. We already have language for this kind of determination: “It’s not a bug; it’s a feature.” This well-worn maxim has intentionality embedded in the spandrel analogy. What appears to one person as a problem is, in fact, a deliberate choice; or, what appears to one person as a logical constraint, is a potential exploit. But even these categorizations begin with an adaptationist’s atomization, ignoring the fact that bugs and features emerge over time, that bugs become features, and that features become bugs, and that all of this happens in unpredictable arrangements of densely connected bugs, features, and unruly human and non-human actors. What happens, I want to ask, when we turn our attention from a pre-given architecture to a spandrel theory of constraints, byproducts, excesses, and unforeseen consequences that mark media technologies and infrastructures?

Accidents, mushrooms, and leap seconds

In her history of SMS messaging, Collette Snowden takes exception to the received claim that SMS was an “accidental” innovation. One version of the SMS story contends that while it was designed to be included in the GSM network, no one imagined its full capabilities and ultimate triumph as a central mode of communication. This would be an adaptationist view of media technology, that does not see SMS as the result of decisions made by engineers, technicians, and policymakers and, later, appropriations by a diverse group of users. SMS was perhaps unexpected but it was far from accidental, Snowden argues, and “framing SMS as an accidental success conforms to a recurrent narrative theme in the history of technology, which positions a specific technological development as the result of fortuitous or unplanned events, seemingly beyond human control or intervention.”[4] We might call the accidental view of technological development the “bad spandrel” view, or the penicillin view. In such a view, spandrels are flukes of circumstance; by treating them as such we ignore the full entanglement of human decisions, embedded physical constraints, and the economics of materials, labor, and space.

Tsing Mushroom.gifYet, the accidental view has some appeal for those who want to trouble the notion that large firms, single actors, and innovators are largely responsible for determining the development and use of new technologies. In her study of the matsutake mushroom, Anna Tsing advocates for embracing indeterminacy. The matsutake mushroom became common in Japan around Nara and Kyoto, following the first wave of deforestation. Here, a delicacy appeared in the ruin of industrialization. Then in the 1980s, as Chernobyl contaminated European mushroom supplies, and forests in the Pacific Northwest were clear-cut, governments came under pressure from environmentalists to halt logging. Again, the matsutake appeared in the aftermath. In these moments, a mushroom can, Tsing, says, “illuminate the cracks in the global political economy.”[5]

In my own work on the history of technology, I often light on the unexpected excesses of human choices. In studying the history of time standardization, I ended up at the latest in a long history of spandrellic challenges: the leap second. With the advent of atomic timekeeping and the creation of new, atomic-based time standards, a conflict emerged. While previous time standards were based on the (always fluctuating) movement of the earth and other celestial bodies, atomic time was based on the unceasing and unimpeachable vibration of cesium atoms. As more technologies were linked into International Atomic Time, a divergence emerged with our current civil time standard, Coordinated Universal Time. A choice had to be made: would we de-link timekeeping from the earth’s movements, therefore slowly separating time from daylight? Or, should we insert “leap seconds” into the time stream to keep civil and scientific time aligned? The choice, for the past few decades, has been to insert leap seconds, manually, into a time-synchronization system that is already highly precarious. The question, then, is a leap second a “free rider”? On the one hand, they could easily be read as byproducts, since they are the result of a contemporary conflict, a gap, between measurement systems. Just as spandrels first appeared as filler between arches and walls, leap seconds paper over differences between atomic and terrestrial timekeeping. On the other hand, leap seconds are a manifestation of a historic truism: that the time of day is determined by a human relationship to daylight. Leap seconds, in this reading, ratify a socially maintained view of timekeeping’s relationship to light. Far from being able to determine the free rider status of leap seconds, we can actually understand this debate through the lens of primary and secondary features—as the attempt by some to argue that the primary feature of a timekeeping system is its long-lasting accuracy and, by others, that the primary feature is a relationship between the earth and the sun. Such a view embraces indeterminacy while using the indexing function of these features to surface ongoing struggles to name what is truly important.

When I try to think spandrellically (which, yes, I know), it shifts my focus away from singling out the shiniest object and classifying technologies into existing hierarchies  (primary vs secondary; intentional vs accidental; bug vs feature). Instead, like Tsing following the matsutake mushroom, spandrels are the markers of change, and they can force us to see the patterns in which those markers appear, and the relationships that they illuminate. A spandrel perspective is an indeterminate, relational heuristic for describing system evolution that resists ascribing power and responsibility to any single actor. Far from seeing some systemic aspects as determining, intentional, or accidental, we can use this perspective to study the conditions in which primariness is asserted, in which simple adaptation is credited when the explanation is far more diffuse, and we could focus on the conditions in which “useful accidents” tend to appear.





[1] S. J. Gould and R. C. Lewontin, “The Spandrels of San Marco and the Panglossian Paradigm: A Critique of the Adaptationist Program,” Proceedings of the Royal Society of London. Series B. Biological Sciences 205, no. 1161 (1979): 581; Stephen Jay Gould, “The Exaptive Excellence of Spandrels as a Term and Prototype,” Proceedings of the National Academy of Sciences of the United States of America. 94, no. 20 (1997).

[2] Gould and Lewontin, “The Spandrels of San Marco and the Panglossian Paradigm: A Critique of the Adaptationist Program,” 585.

[3] “Spandrel” is, more precisely, usually the name given to the triangular space above an arch on a flat wall, while “pendentive” is the term used to describe a three-dimensional spandrel. Gould later recognized this fault in their terminology but notes that “spandrel” is sometimes used to describe pendentives and that the changing to the more precise architectural term did not change the analogy.

[4] Snowden, Collette. “Csting A pwr4l spLL: D evOLshn f SMS (Casting a Powerful Spell: The Evolution of SMS).” In The Cell Phone Reader, edited by A. Kavoori and N. Arceneaux. New York: Peter Lang, 109.

[5] Tsing, A. L. (2015). Mushroom at the End of the World: On the Possibility of Life in Capitalist Ruins: Princeton University Press.