When I started my research on so-called Knightian uncertainty and the accompanying distinction between measurable risk (or “risk proper") and unmeasurable uncertainty (or “true uncertainty,” Knight 1921, 20; 233), I had assumed that the concept had been promoted exclusively by political economists of the (political) right – e.g., Frank Knight and Friedrich Hayek -- suspicious of central planning and over-confident claims of scientistic economists. In Knight the concept plays a role in explaining the existence of entrepreneurial profit, which is a (lucky) ‘reward’ for action in the face of uncertainty in a (relatively) free enterprise economy. (It is a sad fact that our leading economists ignore the role of luck and embrace a thoroughgoing just desserts theory.)
I had assumed the demise of the concept post-WWII could be explained in terms of a simple narrative, in which technocratic Keynesians wished away the very possibility of Knightian uncertainty in order to promote economics as a privileged and capable policy science. But that story is not quite right: as I have pointed out, (i) free market economists also promoted the demise of Knightian uncertainty and (ii) Keynes also embraced a thoroughgoing version of ‘Knightian uncertainty’ in his (1921) Treatise on Probability—the same publication year as Knight’s Risk, Uncertainty, and Profit. (For the significance of the Treatise on Probability, see Rosser.)
In what follows I revisit Keynes's treatment, which is more radical than I realized (shameless self-promotion: I will be discussing the material at GMU in two weeks).
By "uncertain" knowledge, let me explain, I do not mean merely to distinguish what is known for certain from what is only probable. The game of roulette is not subject, in this sense, to uncertainty... Or, again, the expectation of life is only slightly uncertain. Even the weather is only moderately uncertain. The sense in which I am using the term is that in which the prospect of a European war is uncertain, or the price of copper and the rate of interest twenty years hence, or the obsolescence of a new invention, or the position of private wealth owners in the social system in 1970. About these matters there is no scientific basis on which to form any calculable probability whatever. We simply do not know. Nevertheless, the necessity for action and for decision compels us as practical men to do our best to overlook this awkward fact and to behave exactly as we should if we had behind us a good Benthamite calculation of a series of prospective advantages and disadvantages, each multiplied by its appropriate probability, waiting to be summed. (Keynes, 1937, 213-214)
In the quote Keynes is working with four categories of events, some of which can be known with probability, and others that are not susceptible to probabilistic knowledge.
First, there are events that have unknown outcomes, but do have an ex-ante probability measure (or distribution). These are the source of “probable knowledge.” A paradigmatic example of this is a casino game. In the quote Keynes is silent on the source of the defined probability measure; his position here is compatible with so-called Frequentist (or what Knight calls “statistical”) and objective (or what Knight calls “a priori” 216) interpretations of probability. In the former the probability measure is a product of empirical evidence while in the latter it is based on general principles (see Knight 224).
Second, by contrast, uncertain events involve cases where there is no ‘scientific’ basis at all for a probability measure. It’s unclear how Keynes identifies these cases, but his operation of the category is clear enough—they are events beyond (present) scientific knowledge. The best we can do in such cases is, as Knight also emphasizes, offer ungrounded estimates (Knight 225; Knight suggests that this is always the case when we are dealing with “unique” decision circumstances.).
Third, Keynes also allows that that there are events that fall in between the two extremes; it seems he treats events that do not have a fixed ex-ante probability measure, but that are subject to reasonably informed scientific analysis as partially uncertain in different degrees.
Fourth, there is a further category, deployed for practical purposes, when we treat uncertain events as instances of probable knowledge even though from a theoretical perspective we have no justification for doing so; let's call that "ersatz risk". (See also Knight 226.)
In Keynes’ 1937 treatment the distinction between probable and uncertain events is articulated in epistemic terms; he consistently speaks about knowledge. While Knight is not mentioned in context by Keynes (they seem to have wished each other away), this is, in fact, also Knight’s approach, who resolutely wishes to avoid metaphysical claims and whose emphasis on measurement is entirely epistemic. (In fact, at one point Knight suggests that if we were (Laplacian) infinite intelligences, there would be no uncertainty (207).) It’s possible, of course, that Keynes thinks that there are metaphysical or ontological facts that constrain our knowledge—this is why the game of roulette may be a paradigmatic case of a probable knowledge. But his 1937 position here does not rely on it.
However, as Rosser (self-consciously echoing Shackle perhaps) has argued, in chapter 3 of his 1921 treatment, Keynes had offered a (somewhat different) fourfold distinction that was rooted in ontology (and not merely epistemic in character):
There appear to be four alternatives. Either in some cases there is no probability at all; or probabilities do not all belong to a single set of magnitudes measurable in terms of a common unit; or these measures always exist, but in many cases are, and must remain, unknown; or probabilities do belong to such a set and their measures are capable of being determined by us, although we are not always able so to determine them in practice. (Keynes, 31))
Here, Keynes' first alternative is a metaphysical doctrine. For convenience sake I refer to it as “metaphysical uncertainty”—that is, cases of metaphysical indeterminateness. Unfortunately, Keynes does not explain what might cause such metaphysical uncertainty or how we would recognize it. (The 1921 discussion is a bit unclear on this point. Presumably if our actions make the world and this world-making depends on our beliefs and expectations or Keynesian animal-spirits, then the conditions are definitely ripe for metaphysical uncertainty. But that, in turn, depends on further theses about how beliefs are physically realized.)
Moreover, while the third alternative is again articulated as an epistemic doctrine, there are for Keynes clearly facts of the matter that ensure that really existing probability measures are unavailable to us forever. Again, Keynes does not explain what he has in mind, although the subsequent discussion seems to suggest this is due to our cognitive limitations (see the treatment of what “our minds are capable” of and “human powers” at 32-33). Let’s call those third cases “soft-metaphysical uncertainty.”