- 1. I am proposing here to reconstruct Keynes's view as recommending that rational agents endorse a rule for adopting states of probability judgement relative to diverse potential states of full belief or certainty, the weakest con- firmational commitment in the sense of Levi (1980) allowed by probability logic. Ramsey seemed more inclined to assume that the degrees of belief or subjective probability would or should be always cardinally measurable, whereas Keynes did not. Like many of his frequentist opponents, Keynes thought that, in the absence of logically compelling determinations, degrees of belief or subjective probability ought to be indeterminate. Ramsey refused to countenance this and deployed some unimpressive and question-begging arguments against Keynes (see Ramsey, 1990a, pp. 56-7.) Ramsey's insight that probability logic could not be both a logic of truth and a logic of consistency registers a more profound objection to Keynes's approach. But that does not touch the reconstruction I am offering.
- 2. Venn (1888, ch. 9) did make some remarks pertinent to the selection of reference classes and direct inference. But Keynes was essentially correct. Venn thought that the reference class used in direct inference is a practical matter. Keynes complained that Venn was too narrow in restricting probability only to statistical probability. In his review of the 1866 edition of Venn's book, Peirce (1867) complained for the same reason that Venn is too much of a conceptualist because belief probabilities derived from information about frequencies according to Venn are based on appeals to reference classes without objective grounding. Peirce complained especially about Chapter 17 of Venn's book on extraordinary stories.
- 3. Ramsey did not explicitly mention Keynes's discussion of weight of argument; but there seems little doubt that the note was written in response to Keynes's discussion.
- 4. B. Skyrms (1990) discusses the Ramsey note (Ramsey, 1990b). Skyrms is anxious to suggest that Ramsey is a forerunner of "Jeffrey updating" or "probability kinematics" and cites this note as an intimation of it along with some notes taken on a paper of Donkin's from the 1850s. I myself can detect no such intimation and suspect that Ramsey may not have even entertained the idea.