Experimenting in the Unknown: Lessons from The Manhattan Project

Abstract

Experimentation is paramount to innovation. In fact, innovation scholars and practitioners have espoused Thomke's (2003) book titled Experimentation matters: Unlocking the potential of new technologies for innovation. Unfortunately, companies still experience considerable technical and managerial difficulties with organizing experiments under high uncertainty. This study confronts Thomke's experimentation principles with a high uncertainty context by examining experiments conducted by The Manhattan Project from 1943-1945, the initiative that created the first implosion-type fission bomb. Findings suggest that lack of theoretical knowledge, a crisis of scientific instruments, and absence of pre-established organizations were critical to such highly uncertain experimentation. We conclude with a discussion of the boundary conditions of Thomke's experimentation principles. Finally, we propose five principles for facilitating the management of experimentation in the unknown.

Proposing a Multilevel Approach for the Management of Uncertainties in Exploratory Projects

“…It relies on general goals, purpose, the reasons why (e.g., why we should perform this exploratory project), and outcomes of an unknown situation (e.g., new technology, new market). By “unknown situation,” we mean situations in which firms face extreme uncertainties (in line with Gillier & Lenffe, 2018; Loch et al, 2008). At the beginning of Project 3, organizations, led by MIT, faced difficulties in defining the purpose, goals, and outcomes of the exploration project.…”

“…While trial-and-error learning consists of planning until the next milestone, having an overall understanding about the project, and staying open to the emergence of new information, inductive learning employs imagination and goes to action without clear hypotheses or the next milestone. Therefore, inductive learning advances the current understanding regarding experimentation (see Gillier & Lenfle, 2018). Charting the unknown goes beyond managing uncertainties present in a given project (as suggested by Rice et al, 2008) since such management often requires a multilevel perspective that goes beyond the project.…”

The Logics of Double Proof in Proof of Concept: A Design Theory-Based Model of Experimentation in the Unknown

“…The literature suggests that there are several possible forms of experimentation (see, for instance, Gillier and Lenfle, 2019). In particular, the design literature has focused on distinguishing between exploration-based experimentation and validation-based experimentation (Ben Mahmoud-Jouini and Midler, 2020;Blomkvist and Holmlid, 2011;Cross, 2008;Nicolaÿ andLenfle, 2019, Jensen at al., 2017).…”

The Tension Between Flexible Goals and Managerial Control in Exploratory Projects

“…In addition, if uncertainty is very high, some of the testing principles proposed by Thomke (2003) may need to be adjusted, as demonstrated in Gillier and Lenfle (2018). For example, the organization might need to build or recruit new capabilities to carry out the experimentation needed to answer a posed question.…”