Workshop on Belief Revision in Strategy and Organization

The “theory-based” view has been a recently developing and rapidly growing perspective in the field of business strategy. The theory-based view casts the business leader in the role of a scientist. Developing a strategy for an organization then involves creating a theory of how the organization can succeed in creating value and developing a competitive advantage relative to its rivals. Those theories typically involve a set of premises (assumptions) and chains of logic that connect those premises to expected outcomes.

Despite the interest and research in this perspective, many important questions remain open. For example, how can managers determine the validity of their theories and when they might need to reject, revise, or refine them? 

In some cases, like scientists, managers can collect and analyze data or run experiments to test their logic and intuition.But in many cases, they cannot. In many cases, for example, these theories of success involve actions that are irreversible: Perhaps they can only happen once because their implementation creates path-dependence. For example, when Sun Microsystems first embraced RISC chip design, it required them to rewrite their operating system and to redesign nearly every other component in the system. Reversing course not only would have come at tremendous cost but also would have left them behind their competitors who had continued to use Motorola CPUs. Or, a failed experiment might prove catastrophic to the organization. What if Dropbox had lost its customers’ data when it migrated from Amazon’s servers to its own proprietary operating system and data centers? 

Without the ability to experiment or explore variation in outcomes, how can managers assess and refine their theories? How can they learn?

We will refer to this problem as the “belief revision” problem. Anyone who has watched the movie Apollo 11, starring Tom Hanks, has seen an example of why theory-based learning often requires belief revision and not just Bayesian updating. In the movie (and in the actual mission on which it had been based), the ground staff needed to form and revise theories to save the astronauts’ lives. 

The ground staff had a theory that they could bring the astronauts back to Earth alive in the Moon capsule. But how could they establish the conditions that would allow for that? They could not run experiments. Any failed attempt would have killed the astronauts. They also could not rely on the analysis of past data. It had never been tried.

The ground staff developed their theory, tested, and revised it, but they did so not through data analysis or experimentation to evaluate the overall expected outcome. Instead, they formulated and tested sub-premises, assumptions, and intermediate conclusions implied by the theory. They revised the theory by revising its logic. 

The Apollo 11 example illustrates the process that we refer to as belief revision. Belief revision matters not just for emergency theory development, but for all problems in which entrepreneurs or organizational leaders hope to achieve something big that has not been done before. In this sense, scaling Airbnb, making space travel cheap, and stimulating demand for electric cars all involve a challenge similar to that of bringing the astronauts back to Earth in a Moon capsule. They often have just one chance to get it right.

The belief revision problem has already received some attention in philosophy, particularly in the philosophy of science. It is also closely related to non-monotonic logics, which create rules for resolving potentially conflicting logical conclusions. But these issues and how they might be applied practically to managers and organizations have received little attention to date. We therefore propose bringing together a set of scholars with diverse expertise, who usually would not interact with each other, to stimulate research on this question.

To address this issue, we are proposing a small workshop focused on developing a research agenda and on exploring formal foundations for solving these problems. The workshop will bring together a mix of scholars who have done research related to this question but who have not been in deep conversation with each other for a variety of reasons: They come from different disciplinary perspectives and therefore typically attend different workshops. 

We see the workshop as having three main goals: First, it will help to create a social infrastructure for this emerging research question. Many of the proposed participants have never met each other. Extended interaction in such an intimate setting would help to foster new connections within this research community. Second, we expect that it will generate a research agenda for the topic. What are the main sub-problems and what technologies could be brought to bear to solve them? Third, we hope that the workshop would generate the seeds of several collaborations and papers that would begin to address topics in the agenda. 

Although the primary expected contributions of the workshop relate to the development of the theory-based view of strategy and to generating practical advice for entrepreneurs and managers facing these real-world problems, the close connection between the theory-based view and the philosophy of science may mean that the workshop and its fruits will also have implications for best practices in science (another rapidly developing field often referred to as the “science of science”).