One of my main research projects analyzes the history of using the business computer primarily in the insurance and manufacturing industries. By business computer, I mean the computers such as the IBM 650 and 700 series and UNIVAC that were primarily used for administrative tasks as opposed to the computers used for computational purposes and on the shop floor. I focus on insurance and manufacturing because there were two of the largest adopters of computing technology and there are significant differences in how they thought about, engaged with, and ultimately used the computer. From a historical point of view, there is a surprisingly limited but growing research stream on how organizations and industries acquired and implemented the computer. The archives for this work is quite substantial and come from the meeting proceedings, journals, and reports of the various occupations that developed different perspectives and uses of the computer as well as consultant and academic studies, conferences, computer companies, and business media. Methodologically, this work involves a combination of comparative history as well as more quantitative approaches.
One main area of my research investigates the socio-cognitive processes involved in how different industries came to understand the computer. There has been growing interest in the role of cognition in the development and evolution of technologies, but to date there has been limited empirical studies of how these cognitive processes evolve. Methodologically, I take the perspective that the meaning of the computer is best revealed by how it fits into the conceptual schemas organizations use to interpret office technology. This treats the emergent computer conceptualization as embedded within a network of other categories and relations. The meaning of the computer, in turn, reflects how the category and its relations with other categories emerged and changed over time.
My empirical strategy to capture the full schema evolution involves building out the conceptual networks from the articles and books in which different market participants discuss what the computer is, how it should and is actually used. Using Carley’s cognitive mapping approach, I build the networks from the various nouns and relations used to describe the computer and its use within these texts. Below is an example of a aggregated conceptual network from my work with Chris Bingham (UNC) in which we map the emergence of the insurance industry’s cognitive schema (1947 – 1975) primarily from texts produced from the various trade and professional societies. This included mapping close to 400 texts. The diagram shows the conceptualization of office technology prior to the computer from 1947 – 1954. The nodes represent aggregated categories and the lines represent ties between these categories, where larger node size and thicker lines mean higher frequency.Larger Picture of frequency
Note how tightly coupled “machine”, “Card”, and to a lesser extent “Clerk” are, forming a cohesive conceptual structure. Chris and I observe that this pre-existing conceptual schema influences the emergent computer schema in the sense that while other competing analogies highlighted new features of the computer, at least initially the initial schema of the computer closely resembled the above schema. Overtime, however, the machine concepts and relations decreased as new more decision-oriented, brain-like concepts and relations emerged. We identify the socio-cognitive mechanisms involved in this process. For a more complete description of the socio-cognitive processes involved in this transition, please see our paper.
With JoAnne Yates (MIT), I am investigating the role of power and authority in the convergence of the conceptualization and use of the computer within the insurance industry. Even before the computer was commercially available, the insurance industry converged on thinking of the computer as a transaction-processing machine and initial uses of the computer centered on these administrative applications. We explain this homogeneity in terms of how managerial elites from the administrative, accounting, and systems occupations established themselves as cognitive authorities of the computer and then controlled the acquisition and implementation processes. They maintained this conservative conceptualization of the computer by filtering broader institutional understandings of the computer such as cybernetics and AI. This work exposes the important role of inter-occupational struggles within organizations as well as the role of the professional associations in this process of developing an understanding and use of the computer.
Building on the occupational theme, another project investigates how different occupations within the manufacturing industry interpreted and used the computer to expand their authority and jurisdictional control. Unlike insurance, there was no centralized control over the interpretation of the computer within manufacturing. Some occupations, like accounting used it to reinforce existing practices, some like purchasing and marketing were not as interested; whereas, others like production planners and systems men leveraged the computer to try to create new task areas and mobilize the profession. In work with Brayden King (Northwestern) and Greg Liegel (Chicago), we compare the production planners’ and systems men’s strategies as well as their structural position within the occupational system. We argue that groups that are more central to the organization and have more task overlap with other competing occupations are more likely to use a differentiation strategy; whereas, groups that are more peripheral to the organization and have less task overlap will be more inclined to use an integration strategy. Moreover, our historical analysis reveals that the integration strategy, although it appears to be more conciliatory on its surface, has significant transformative potential, inasmuch as it enlists the cooperation of other groups in the new technology’s use, thereby expanding the influence of the peripheral group and potentially altering organizational processes.
Prior to academia, I worked as a consultant helping firms build and implement large scale software applications and as an equity analyst on Wall Street following the business application software sector. I build upon these experiences in my research about how the business application software industry emerged. Related to my conceptual work, one project with Christopher Bingham analyzes the history of the classification system for manufacturing software applications (think of the history of enterprise resource planning or ERP firms like SAP and Oracle). This work involves gathering archival records of the various market participants who tried to classify software, dating back to the introduction of the business computer in 1954. This archive includes software firms, occupational groups such as production planners who used the software, industry research firms such as IDC and Garnter Group, consultants, etc … We observe that the logic underlying the classification system shifted several times from an initial focus on technical aspects associated with the computer to a user/occupational system to finally a process-oriented system. These different logics influenced the characteristics that defined and differentiated the various software categories. Finally, we identify a relational structure in this classification system such that well formed categories like marketing applications that were not as well integrated into the relational structure did not develop as well as other categories that were well integrated. We use this analysis to make the argument that system-level characteristics, such as logics and relational structures influence the development of individual categories and that this perspective has been missing from the mostly category-level analysis in conceptual development.
Related to the occupational, work, and professions analysis with the computer, Greg Liegel and I also are investigating why programmers have not been successful at professionalizing. Several explanations have been advanced, but we argue that they do not factor how the technical constraints of the computer and its development toward a modular structure influenced the programmer’s ability to establish its own knowledge domain. From the technology management literature, we treat computers as a system and our historical analysis combines the historical evolution of this system as it was enacted with the nature of work performed by programmers. We note that initially the lack of differentiation between hardware and software meant that the practice of programming was a physical activity that had significant overlap with other occupations such as systems work and machine operators. This impeded initial efforts to define a unique domain of knowledge. Over time, the computer system became more modular. While increased modularity differentiated software from hardware, it also fragmented the practice of programming which made it more difficult to identify the core body of knowledge that defines programming.
Another series of projects in the software industry resonate with my interest in the role of technology use in industry formation and evolution. I have a series of projects with Michael Cusumano (MIT) and Fernando Suarez (Boston University) that considers services as a firm activity that surrounds the usage characteristics of a technology. The technology strategy and evolution literature has either ignored the role of services or assumed that services play a greater role as a technology matures; however, our analysis of publicly traded software firms from 1990 – 2006 shows that the relationship between services and firm performance is more complex. We find a non-linear relationship between a product firm’s fraction of total sales coming from services and its overall operating margins - that additional services start to have a positive effect on the firm’s overall profits when services reach a majority of a product firm’s sales. In a separate paper under development, we explore the role of services at a theoretical level arguing that it represents a different kind of knowledge exchange process and capability that influences product development throughout its lifecycle.
My work in financial markets builds upon my interest in implementation and enactment in the technology sector. Here, I apply this perspective to better understand the relationship between regulations and market order – a significant and growing research area. With Damon Phillips (Columbia), I examine how investment banks applied FAS 157, or the “Fair Market Rule” as a means to better understand how market order get established following the introduction of a new regulation. The Fair Market Rule requires publicly traded firms to classify their assets and liabilities into one of three categories depending upon the methods used to calculate their fair value. This rule received significant notoriety as many blamed it for magnifying the financial crisis. The categorical nature of the rule allows us to plot and compare how different firms actually applied the rule to make accounting decisions.
While neoinstitutionalists typically cast market order in terms of convergence to a common understanding, we surprisingly observe that investment banks did not converge to a common application of the rule. Yet they acted in a coordinated way akin to White’s market schedule and Stark’s coordination without homogenization. We investigate the various social factors found in economic sociology – isomorphism, peer coordination, and audience evaluation – to determine how this structure emerged. Conceptually, we argue that there are qualitatively different market structures and introduce the concept of comparability as a more general way to capture market order.
I also examine innovation and market emergence within creative markets. Similar to the technical markets, I am interested in how music classifications form. This work has identified a different kind of process than typically assumed in the category formation literature and my work in technology. The category dynamics of jazz do not rest on identifying salient characteristics that define is genre or on underlying institutional logics. Work with Young-Kyu Kim and Damon Phillips highlights the sequence of identities inherent in the production and adoption of creative products that forms the basis for defining that genre. More specifically, we explore how the identities of the producers and early adopters influence the long-run success of jazz recordings. We argue that early jazz songs were more likely to be covered if they followed a pattern of having black originators and white early adopters, independent of other quality and resource factors. The collective work in jazz and technology helps expand our understanding of category formation to include more complex and varied processes. It also lays out the groundwork to identify the conditions that favor one kind of formation process over the other.