Almost exactly a century ago, two books were published (Keynes, 1921, Knight, 1921) which sought to persuade economists to distinguish risk, which can be described with probabilistic mathematics, from uncertainty, which cannot. It was an argument that they lost as far as mainstream economic thinking is concerned. In the 1940s Friedman and Savage (1948) and Savage (1954), following von Neumann and Morgenstern (1972), provided an axiomatic characterization of choice under uncertainty which gave rise to an approach commonly labeled subjective expected utility (SEU). Within 20 years, Friedman, who had succeeded Knight as doyen of the Chicago school of economics, would go on to write “I've not referred to this distinction (between risk and uncertainty because I do not believe it is valid. We may treat people as if they assigned probabilities to every conceivable event” (Friedman, 1962). Subjective expected utility implies that individuals—and by extension organizations—make decisions as if they attached numerical values both to possible outcomes and the probabilities associated with these outcomes. Importantly, SEU does not claim that this account describes the thought processes that give rise to these decisions; rather it asserts that adherence to its axioms is definitive of “rationality.” That claim is supported by the demonstration of Ramsey (1926) that individuals who do not observe these requirements can be “Dutch booked,” that is, induced to accept a series of gambles whose net effect is sure to leave them worse off. SEU underpinned the increasingly mathematical financial economics that found practical application on Wall Street and in the City of London and was employed in the prudential regulation of banks and other financial institutions. Thus, the global financial crisis of 2008 invited a critical reappraisal of this approach to the analysis of uncertainty. The most widely recognized critique came from “behavioral economics,” an application of experimental psychology pioneered by Kahneman and Tversky (1996). They showed that subjects repeatedly breached the axioms when confronted with simple choices between gambles. But while the behavioral economics of Allais (1953) and Ellsberg (1961) began as a critique of SEU as a model of how agents made choices under uncertainty, in the version popularized by Kahneman, it became a critique of human behavior. Individuals were poor decision makers, subject to a wide range of “biases.” Even if SEU failed descriptively, it retained its normative force. However, a more fundamental critique disputes the claim that SEU is mandated by rationality. It reasserts the claims of Keynes and Knight that most uncertainty cannot usefully be described in probabilistic language (Kay & King, 2020). When information is imperfect and unequally distributed, it is irrational even to hold, far less act on, subjective probabilities of events and outcomes, which are often incompletely specified and ambiguously determined. Individuals have neither learnt nor evolved to be good at probabilistic reasoning because probabilistic reasoning is not helpful in most real-world situations. Paradoxically, at first sight, the field in which such reasoning is most readily applicable and for which it was first developed—the gambling arena—is one in which most rational individuals decline to participate. Kay and King re-emphasize but redefine the distinction that Keynes and Knight expressed as that between risk and uncertainty, preferring to distinguish resolvable uncertainty—which can be reduced or eliminated by obtaining more information or specifying a well-determined probability distribution—from radical uncertainty, which cannot. Noting that the term risk has become synonymous with volatility in the language of financial economics, they prefer to restore the meaning of the word in ordinary language. In everyday speech, risk is the downside; risk jeopardizes the success of the favored narrative. Thus, conviction narrative theory relates more directly to how people report their decision-making processes, and thus enables them to make better ones. The interview studies of traders in financial markets reported by Tuckett (2011) confirm the experience of most people—that we cope with uncertainty by telling stories. Thus conviction narrative theory offers an alternative account in both the descriptive and normative realm. But, as Fenton-O'Creevy and Tuckett (2021) emphasize, conviction carries dangers as well as opportunities. Indeed one of the few recent works to recognize the importance of narratives to finance (Shiller, 2019) treats attention to narratives mainly as an opportunity for irrational exuberance—as it has been in speculative bubbles throughout history. Financial market participants seize some event—often the development of a new technology—and, exaggerating its short-term impact, bid up the prices of related assets. The profits thereby earned by early adopters of the narrative appear to validate it. This process continues until almost no one is left to be convinced, and at that point the bubble bursts. Fenton-O'Creevy and Tuckett correctly identify the dot com boom of 1999 as an example of this process. But similar observations can be made of the events which led to the 2008 global financial crisis. Bankers and policymakers persuaded each other that the proliferation of complex securitizations of debt instruments, in the words of Federal Reserve Board Chair Alan Greenspan, “enhance(d) the ability to differentiate risk and allocate it to those investors most able and willing to take it” (Greenspan, 1999). When Raguthan Rajan, then chief economist of the International Monetary Fund, warned of the risks accumulating in the system, he was shouted down and eventually left his post. Larry Summers, former US Treasury Secretary and future adviser to President Obama denounced him as a Luddite (Wheatley, 2013). Ultimately many of these securities proved to be worthless and the scale of the resulting bailouts of major banks and financial institutions was unprecedented. In 2022, we are experiencing what may be the strangest of all financial bubbles. Distributed ledger technology and blockchain have been described as “a solution in search of a problem,” but skeptical voices have been drowned by a crescendo of noise on social media. Bitcoin is the first and still the best-known cryptocurrency, but there are now many hundreds of similar instruments. (A nonfungible token (NFT) related to a jpeg created by a “digital artist” known as Beeple sold in May 2021 for $69 million (Kastrenakes, 2021), giving tangible, or at least electronic, reality to Tuckett's description of “phantastic objects”). Special Purpose Acquisition Companies (SPACs), “undertakings of great advantage but no one to know what they are” were launched, often associated with the names of financial wizards including Jay-Z, Shaquille O'Neal, and Martha Stewart; these attracted funds from small investors (Lipschultz, 2021). And in a near-perfect repetition of the 1999 process Fenton-O'Creevy and Tuckett describe, in which companies added to their stock price by attaching “.com” to their names, the NASDAQ listed Long Island Iced Tea Corporation achieved a near fourfold appreciation in value in 2017 by changing its designation to Long Island Blockchain Corporation (In 2021 executives of the company would be charged with insider trading, having passed this supposedly material price sensitive information of their intention to institute the trade to associates) (Egan, 2021). Conviction narratives are necessary if decision-makers are to find the courage to act in the face of radical uncertainty. But conviction is not enough and excessive or unjustified conviction can lead to costly mistakes unless the narratives are effectively tested in debate and discussion. As Fenton-O'Creevy and Tuckett describe, this process of scrutiny must take place within the organization as well as in the public fora associated with financial markets. The concept of the mediating hierarchy, described in the context of corporate governance by Blair and Stout (2001) is useful here. The hierarchy of caricature, in which an executive like Henry Ford demands control of every aspect of production—you can still find Fordlandia in Brazil, where Ford established rubber plantations to ensure that even the tyres were from his own trees—is over. As happened at Ford, the resulting inability to challenge the convictions of the leader is ultimately crippling. In the mediating hierarchy, the role of the boss is not to tell everyone what to do but to assemble views and clarify that a decision has been taken and what that decision is at the end of a process of debate and negotiation. This gives the opportunity for the organization to achieve what Fenton-O'Creevy and Tuckett describe as an integrated state, which supports a central narrative while acknowledging the risks and uncertainties associated with it. The lesson for economics and finance is that the imposition of a model of behavior deduced from a priori axioms not only fails to provide a persuasive description of real decision making but has led to policy choices that have imposed serious economic consequences. Future research on decision-making under uncertainty should be more humbly inductive, rooted in observation of the processes that yield good—and bad—outcomes.