SPECULATION AND EVIDENCE
is that gases are composed of spherical molecules, or that some hypothesis correctly explains both why gases are composed of spherical molecules and why mechanical hypotheses have been successful.[30]) Analogous claims are justified in the case of “only-game-in-town” and “authoritative” evidence.
What follows from this about speculation? If the A- definition of potential evidence is employed, and if you theorize using h and know that there are only “meta-inductive facts” in favor of h, then h is a speculation for you. Matters get trickier in the case of “authoritative” and “only-game-intown” facts in favor of h. Let's start with the former. Suppose you know that the following is the case:
e: The recognized authorities announce that they have discovered (potential) evidence that h is true, and that for this reason they believe that h is true, even though they do not announce what this evidence is.
Suppose that the evidence they have discovered really is potential evidence that h. Finally, suppose that these authorities are so good that you can rightly claim to know that such potential evidence for h exists, even though you don't know what it is. If you use h in theorizing, then you are not speculating, since you know (from authority) that there is potential evidence that h. This is the case even though e is not potential evidence that h (it fails to satisfy the explanatory connection condition). So it is not that e is “authoritative” evidence for h. It is that, in this case at least, if you know that the authoritative fact e is true, then you know there is (non- authoritative) potential evidence for h.
On the other hand, suppose that that the evidence the authorities have in mind is not in fact potential evidence that h, though they believe it is, and perhaps, given their ep- istemic situation, they are even justified in doing so.
Then, despite the fact that the authorities believe they have discovered potential evidence that h is true, they haven't. And they, as well as you, are speculating (in a sense of (spec) employing potential evidence) when they and you use h in theorizing. So, in general, knowing the truth of an “authoritative” fact such as e is not sufficient to prevent h from being a speculation. Nor, for analogous reasons, is knowing that h is the “only game in town.”What happens if we use an objective Bayesian B-concept of evidence (one that is upgraded to require that e is true and that e increase h’s objective probability so that the latter is greater than %, but not to require satisfaction of the explanatory connection condition)? Such a Bayesian needs to argue first that meta-inductive, “only-game-in-town,” and authoritative facts do, or at least can, increase the probability of a hypothesis, so that they really do or can constitute (objective Bayesian) evidence.[31] Next, he has to argue that these kinds of facts can increase the probability of a hypothesis sufficiently to make the probability greater than %, so that the hypothesis is not a speculation (in an objective sense). Elsewhere I have given arguments to show that at least “only-game-in-town” facts fail to raise the probability of a hypothesis, so that they don't even satisfy the Bayesian definition of evidence.[32] Furthermore, even if the Bayesian can show that meta-inductive facts can increase the probability that some theory of the type in question is true or viable, it doesn't follow from this that the probability of the particular theory in question has increased. Maxwell argued from the success of mechanical theories in astronomy and electricity to the probable success (or the increase in probability) of some type of mechanical theory of gases. He did not claim that his particular theory of spherical molecules acting only by impact was probable, or that its probability was increased.[33] “Authoritative” facts—ones stating just that the authorities believe that h is true—are subject to Laudan's “pessimistic” induction: “Authorities” (those deemed to be such) from Ptolemy to Aristotle to Newton to the present time have gotten it wrong so often that the only inference possible just from their success rate is that they will get it wrong again—not that what they believe now is made more probable by the fact that they believe it.[34]
However, for the sake of argument, suppose (contrary to what I have just claimed) that the fact that h is the “only game in town,” or that h is a theory of a type that has been successful, or that a vast majority of the authorities believe h, could raise the probability of h beyond Th This would satisfy the upgraded objective Bayesian concept of evidence.
And since the probability of h on e would be greater than Th a necessary condition for potential evidence would be satisfied as well. However, the major claim I want to make is this: So raising the probability of h, and thus making it reasonable to believe h, is not enough for what scientists want or should want of evidence. Even if h is the “only game in town,” even if h is a hypothesis of a general type that has been successful in other areas, even if the authorities believe h to be true, and even if these facts provide a good reason to believe the hypothesis, or a better reason than without them, when scientists seek evidence for h, they want something more. I turn to this next.10.