Abstract
In his 2006 EGOS colloquium keynote address, James March (2007) identified three waves of organizational research since 1945. The first wave followed the growing demand for teachers in North American universities after the Second World War and the high hopes for social science models, rational analysis, social engineering, and systematic quantitative research to help solve social problems. The second wave reflected the more pessimistic era of the 1960s’ and 1970s’ protests and counterculture movements. Languages of social construction and power supplanted scientific social science amidst the dismay at the social problems of the time. The third wave followed the triumph of capitalism as the Soviet Empire collapsed and social scientists sought social reform through competitive markets, resources increasingly came from businesses, and research focused on individuals and incentives.
The garbage can model, with its focus on organizational anarchies characterized by problematic preferences, unclear technology, and fluid participation, clearly reflects the ambiguities of the era of protest. Yet in choosing simulation over social construction, Cohen, March, and Olsen (1972) favoured rigorous social science models and produced a remarkably optimistic and playful take on those pessimistic times. In the ensuing years, the simulation was largely put to the side and the garbage can metaphor became a trope for disorder and chaos in organizational choice. In a strident critique, Bendor, Moe, and Shotts (2001: 169) complained that thinking around the model ‘lack[ed] the rigor, discipline, and analytic power needed for genuine progress’. Their rhetoric may have been excessive, but they had a point. Forty years later, this volume provides Cohen, March, and Olsen (Chapter 2, p. 28) an opportunity to clarify, distancing themselves from the received version of the model: ‘A central notion of the original paper was not that the world was inexplicably chaotic but that the appearance of chaos came from the application of an erroneous model to an orderly temporal reality.’
How do we reconcile our images of chance and disorder in garbage can research with these claims of order? In fairness to the subsequent followers as well as critics, the order in the original ASQ paper is not immediately evident. Metaphors like ‘garbage can’, ‘organized anarchy’, and ‘streams’ of problems, solutions, participants, and choice opportunities all suggest disorder. The simulation, however, describes what Cohen, March, and Olsen (Chapter 2, p. 26) now call ‘temporarily unfolding processes of participation and attention’ (p. 26). The temporal order is something like customers arriving at a coffee shop, where service depends on the timing of the arrivals and the matching of customers to baristas. At certain times in the day, more people will want coffee, so arrivals will increase. Without more baristas at those times, customers will have to wait.
The garbage can model depicts a more complex process that merges multiple arrivals: problems, solutions, participants, and choice situations. To appreciate that temporal order we might imagine a peculiar cinema complex with ten different movie theatres (choice situations). When moviegoers (participants) arrive they go to the movie theatre that they think is most likely to show a movie and discover that movie genres (problems) have been randomly assigned to a movie theatre. It is as if at the moment a news story appeared in the New York Times online, an appropriate movie genre is assigned to a theatre: a terrorist attack creates an action movie genre, a human interest story creates a family drama genre, a sex scandal creates an adult movie genre, an international story creates a foreign movie, and so on. Specific movies (solutions) would be randomly sent to a theatre based on when the studio released the movie.1 Movies are played only if moviegoers can fit the movie to all the genres assigned to a specific theatre. Moviegoers may be more or less motivated to make a movie fit. If a movie is shown, the theatre is shut down for cleaning and unavailable until an entire cycle of showings ends. If no movie is made to fit then no movie is shown and the genres (problems) and participants go to the open theatre (available choice situation) closet to playing a movie.
This is clearly an absurd way to run a business. A rational owner would select movies to draw the largest audience and moviegoers would choose movies that they wanted to see. But Cohen, March, and Olsen were not describing a business; they were modelling the social behaviour of the time. In this volume they describe (Chapter 2, p. 21) the behaviour that informed their model: The setting was a vigorous social science community super-imposed on an academic world stirred by sentiments of revolt expressed in demonstrations, drugs, clothes, sexuality, hair styles, and exhilarated frustration. That world was somewhat, but not thoroughly, Marxist. Somewhat, but not thoroughly, ‘organic.’ Somewhat, but not thoroughly, Maoist. Somewhat, but not thoroughly, Buddhist. And rather American: populist, splintered, and subordinated on the weekend to football, or the beach.
This is closer to how we experience life. Each day a newspaper’s front (home) page presents a set of problems, the opinion page lists solutions, and people must attend to both. The garbage can models that process temporally, not causally.
The scholars contributing to this volume reveal how radical that approach remains 40 years on. Two patterns are prominent in the contributions. First, the rigors of modelling are fully embraced. If this volume is any indication, simulations are now the order of the day. Fully half the chapters use a simulation, and those chapters reflect a flourishing online exchange of models. Bendor and Shotts (Chapter 5) may set the standard, delivering on their call for rigor, discipline, and analytic power in models offering testable predictions (Bendor, Moe, & Shotts, 2001). Other chapters also respond admirably to that call, incorporating such ideas as choice ordering (Denrell, Chapter 4), spatial ordering (Fioretti, Chapter 6), dispatchers to rationalize arrivals (Seshardi and Shapira, Chapter 7), participant differences (Knudsen et al., Chapter 8), team formation (Trotizsch, Chapter 9), authority and socialization (Morgan and Carley, Chapter 10), and heterogeneity in skill and ambition in problem-solving (Knudsen et al., Chapter 12).The easy exchange of models may be something of a mixed blessing, though: the models, not social problems, often drive the analysis, and frequently the chapters want for empirical work to test the conclusions of the models.
Second, many of the contributions to this volume struggle with placing agency, intention, and consequence in the context of the garbage can model. Most chapters seek some form of agency and control in the decision process, either through models of individual preferences (Denrell, Chapter 4; Bendor & Shotts, Chapter 5; Levinthal, Chapter 13; and Lomi, Conaldi, & Tonellato, Chapter 14), of individual difference (Knudsen, Wargelien, & Yi, Chapter 8; Trotiszch, Chapter 9; Knudsen, Stiglietz, & Yi, Chapter 12; and Augier & Guo, Chapter 16) or normative order (Seshardi & Shapira, Chapter 7; Morgan & Carley, Chapter 10). The contributions by Denrell (Chapter 4), Bendor and Shotts (Chapter 5), and Fioretti (Chapter 6) set this agency amidst chaos and chance and turn to random walks and stochastic models. These contributions show how the garbage can model conflicts with our hopes for individual agency. The original model had little to say about individual agency, and up until this volume, there were few alternatives for thinking about individual choice in garbage can settings.
Some of the most interesting contributions in this volume do offer alternative views of agency. Gibson’s contribution (Chapter 3), with its remarkably detailed conversation analysis of the Cuban missile crisis, suggests one avenue. Choice, in Gibson’s story, is not about individuals, incentives, options, or optimizing. It is about a narrative developed in dialogue with a broader set of actors. Drawing from audio recordings President John F. Kennedy made of his meetings, Gibson provides a fine-grained analysis of the deliberative process, right down to the length of pauses, voice inflections, and turn taking or overlapping conversation. Kennedy and his advisors only faced a single problem – the Soviet Union had started to install nuclear missiles in Cuba – but had no similar experience to draw on, meaning they were learning from a sample of one or fewer (March, Sproull, & Tamuz, 1991). They did have multiple solutions, all of which Kennedy and his advisors realized had a high chance of leading to nuclear war. Gibson’s analysis shows how they told and repeatedly retold narratives about the possible futures. Eventually the advisors began to suppress the worst consequences of one option, giving the President grounds to craft a narrative making it the obvious best choice. Even that solution did not define the outcome of their first choice. Rather than act, Kennedy and his advisors focused on crafting a narrative to present to the public and to Khrushchev.
Kreiner’s contribution (Chapter 15), an account of an architectural competition for the design of a school and library outside Copenhagen, describes a lower-stakes choice in which the narrative is also essential. He describes how shortly after the presentations of the various models – but before having selected the criteria for a choice – the architectural experts on the panel had already arrived at their choice. Faced with the unanimity of the experts, the lay members of the panel subsequently suppressed their opinions and deferred to the experts. The entire panel then worked on a narrative that would justify that choice. This process, like the Cuban missile crisis, suggests a form of trimming to create a narrative, first, by suppressing voice in the face of expertise and, second, in shaping a narrative to justify a winning design.
If these dynamics of choice diverge from the bounded rationality of the Carnegie School, the findings of Knudsen, Warglien, and Yi (Chapter 8) suggest why: we may overestimate the merits of individual agency. These authors conduct an experiment and rather than constrain where individuals can go in a choice situation, they let them go where they wish. Nearly two-thirds of the time subjects do what the garbage can model predicts: they go to the choice closest to a solution. A third of the time, however, subjects go to the problem needing the most effort. The experiment shows that when subjects can make their own choices, their performance is no different than the original model. Yet when subjects are assigned randomly to choice settings – giving them no choice in what problems and solutions they encounter – performance goes up. This is why a garbage can world is so hard to embrace: when the flow of problems and solutions is unpredictable, individual choice adds to the variance, suggesting that our standard models of human intelligence are less effective than we think.
So where do we find order in that world? In a garbage can model our experience of the world depends on who we encounter, what problems are contemporary, and what solutions present themselves. Agency and order are possible in the contingent experience of the garbage can model, but they do not exist in any specific choice. Instead, order is found in the dynamics of the streams: what is produced and how each stream is produced. Ocasio’s chapter on attention in the volume is an especially effective guide to that order. For 20 years, Ocasio has encouraged us to attend to attention. Here he pulls attention as a thread running from Simon (1945), through Cyert and March (1963), and into Cohen, March, and Olsen (1972). Like many of the chapters in this volume, Ocasio normalizes the garbage can model, but his approach is distinctive: he treats the independence of problems, solutions, participants, and choice situations as an extreme form of loose coupling. That gets us out of the images of chaos and chance without returning to consequential logic in the moment of choice.
Instead, it opens up a way to reconsider temporal order in the intersection of the four streams. We cannot always influence specific decisions, but we can heighten the probability of a solution arriving, diminish the perception of a problem, or change how streams come together. Structures can program how and when people make choices and what problems and solutions will be attached to those choices. For instance, academic institutions as organized anarchies may have problematic preferences, but they also use ritual events like convocation and commencement to shape preferences and to draw attention to problems and solutions. In other settings, people intentionally isolate themselves from a flow of problems and solutions – as in sabbatical leave, health spas, or monastic settings. Religious traditions have long used rituals to shape attention over the course of the day, as in Muslim salat, or year, as in Jewish Yom Kippur or Catholic liturgical traditions. These rituals shape both the flow of available solutions and perceptions of problems, giving myth and ceremony meaning beyond apparent intent.
If these do not appear to us as typical garbage can settings, perhaps it is because the model retains some vestige of the pessimism of the era of protest. When we think of garbage can settings as chaotic and disordered, we see the contingency of temporal order working against individual agency, but we miss the ways that individuals can use that contingency to work for change. It is an old notion that the garbage can model renews. The great Irish poet Paul David Hewson (better known as Bono) reminded a University of Pennsylvania graduating class of the issue at their 2004 commencement exercises. He told them that ‘America is not just a country, it’s an idea’ (Bono, 2005). He was calling them to a commitment to ideas, and he was speaking in Philadelphia, near the historic heart of the American Revolution, so he reminded the graduating class that the idea of America came with what he called ‘some liner notes’: the Declaration of Independence and the Constitution of the United States. Bono was speaking of a broader issue we take for granted – structural efforts to grant access to problems and solutions – as he reminded the class that the future was not fixed.
Our imagination of how we shape the future may be limited because of the era we live in. When we test organizations against economic rationality and limit the important experience to the moment of choice, we may miss how we can shape the temporal order of participation and attention. One profound example took place amidst the events leading up to the triumph of capitalism. In 1988 Bruce Springsteen electrified the youth of East Germany when he appeared in East Berlin before 300,000 people and, in the middle of the concert announced (in German) ‘I’m not here for or against any government. I’ve come to play rock ‘n’ roll for you in the hope that one day all the barriers will be torn down’, then launched an emotional rendition of the Bob Dylan song ‘The Chimes of Freedom’. (The clip is available on the internet.) After decades in which the East German government tried to control the flow of ideas, Springsteen marked a propitious moment with an image of freedom that is a recurring theme in his music (see Kirschbaum, 2013, for the drama of that concert). How much he shaped the future is a subject for debate, but he surely shaped attention.
In this era, after the triumph of capitalism, these examples may appear non-traditional to organizational scholars, but perhaps we have it backwards: business is non-traditional and other settings are truly traditional. For a long time we have treated organized anarchies as if they are unique and poor models for organizations, forgetting the merits of choice settings outside business. As academics this means that we diminish the very setting that we have embraced, the university, by treating it as a model of chaos and disorder. For the romantic ideals of rationality that may be true, but there is more to the world than rationality. At their best, universities do something essential: they shape perceptions of problems, contribute to the flow of available solutions, and prepare participants to engage in a complex world. Classrooms, conferences, journals and academic calendars all shape attention.
Indeed, the conference that produced this volume is one such occasion for shaping order and attention. The editors assembled a group of scholars committed to a particular problem, but from diverse perspectives. The volume that resulted will reward those interested in what simulation models can bring to organizational research. It offers rich images to those seeking a language for the micro- and macro-foundations for garbage can settings. And it reminds us that garbage can processes are about more than disorder and chaos; they are a product of temporal order. The volume also celebrates the process that produced the model itself – the confluence of three scholars, who in their commitment to social science some 40 years ago in Irvine, California, shaped a stream of problems, solutions, and participants. Those who were so fortunate as to know one (or more) of these three scholars are richer for the devotion to scholarship that they modelled. It lives in the book, in the ideas that will be passed on, and in the traditions that are handed on to future generations of scholars.