Knightian uncertainty and the limitations of the Savage Heuristic

• Published date: 01 December 2023
• Author: Nicolai J. Foss
• Date: 01 December 2023
• DOI: http://doi.org/10.1111/emre.12623

EDITORIAL

Knightian uncertainty and the limitations of the Savage Heuristic

INTRODUCTION

The premise behind this special issue of the European

Management Review is that management theory needs to

grapple better with uncertainty. There are two key rea-

sons for this. First, entire fields of management theory—

notably entrepreneurship, strategy, and important parts

of organization theory—are predicated on the existence

of uncertainty. As Townsend et al. (2018, p. 659) argue,

“one of the enduring and fundamental assumptions

underlying theories of entrepreneurial action is that entre-

preneurs operate in uncertain environments.”And in

strategy, the core problem is seen as putting together

bundles of (unpriced) complementary resources under

uncertainty to maximize value creation (e.g., Lippman &

Rumelt, 2003). The upside of uncertainty is that it gives

individuals and organizations the chance to “create some-

thing from nothing”(Baker & Nelson, 2005). The down-

side of uncertainty is that it may be associated with

potentially disruptive events that may negatively affect

the fates of organizations, and their members, as well as

the communities in which these are embedded,

as famously captured by the notion of a Black Swan

(Taleb, 2007). This means that organizations need to be

prepared for uncertainty, build resilience, and adapt

when uncertain events manifest. If organizations can

devise ways to anticipate uncertainty and deal with both

its beneficial and adverse effects, the need for costly

adaptation can potentially be dramatically reduced. Simi-

larly, if organizations can devise ways to embrace uncer-

tainty, much unnecessary experimentation can be

eliminated. The second assumption underlying this spe-

cial issue is that management research in fact has the

potential to offer decision-support in situations character-

ized by uncertainty—but that so far it has not been suffi-

ciently successful in this regard.

In this brief essay, I point to some of the challenges

that face attempts to grapple with uncertainty in manage-

ment theory. First, the field still faces considerable con-

struct clarity challenges with respect to what we mean by

Knightian uncertainty, as almost everything from subjec-

tive probability (in Bayesian as well as non-Bayesian

forms) over ambiguity to sheer ignorance has been called

“Knightian uncertainty.”This matters for a number of

reasons, one being the persistent and powerful Bayesian

objection that “uncertainty”is reducible to subjective

probability, so that per implication, Bayesian normative

principles also apply under such uncertainty, and there is

no need for a distinct “uncertainty”terminology (for a

discussion of these issues in the context of entrepreneur-

ship theory, see Foss & Grandori, 2020). Second, I will

argue that Knight (1921) was right in emphasizing the

uniqueness of decision situations as a source of uncer-

tainty, but that his analysis was incomplete. The problem

is that some uncertainty caused by uniqueness is entirely

“remediable”(Packard & Clark, 2019), notably by run-

ning experiments (see Zellweger & Zenger, 2023). For

example, the decision whether to introduce a new feature

to some product may be shrouded in initial uncertainty.

However, experimentation may quickly resolve such

uncertainty. What I shall call the “Savage Heuristic”—

that we can treat the real, “large”world as if it is a “small

world”—works. However, other uniqueness-induced

uncertainty is not resolvable in this way, such as the

uncertainty that currently surrounds the Power-to-X eco-

systems that are emerging in many parts of the develop-

ing world, that is, efforts to turn renewable energy into

“X”(typically hydrogen). Thus, it is not uniqueness per

se that matters, but the combination of uniqueness and

the “size”of the decision problem. Since “big”decision

challenges cannot be resolved by means of experiments

and making use of past data points, there is a need for a

“logic of imagination.”

UNCERTAINTY: MEANING AND THE

BAYESIAN CHALLENGE

Knight and beyond

The notion of “uncertainty”has often been embedded in

discussion heavy philosophical issues, such as determin-

ism versus indeterminism, free will, and the meaning of

“probability”(Shackle, 1972). Knight (1921) did not shy

away from these issues; indeed, they are crucial to his dis-

cussion. In Knight (1921, Chapter 7), Knight provides a

fundamentally ontological grounding of the argument by

invoking non-determinism and concluding that the future

is “inherently unknowable.”He argues that our cognitive

faculties have evolved as instruments of foresight and

[Correction added on 10 January 2024, after first online publication: Article title has been corrected from ‘Special issue, “Expanding the boundaries of rationality”:

Knightian uncertainty and the limitations of the Savage Heuristic’to ‘Knightian uncertainty and the limitations of the Savage Heuristic’]

DOI: 10.1111/emre.12623

626 © 2023 European Academy of Management. European Management Review. 2023;20:626–631.wileyonlinelibrary.com/journal/emre