Radical scepticism has argued at length that there can be no certain knowledge, nor even probable knowledge; that nothing
can be known to be more probably true than false, and hence, of any two propositions neither one is more probably than the
other.
A first retreat from this radical scepticism is to admit degrees of confidence in our opinions, while admitting that we lack
certain knowledge of such degrees.
Once such degrees of confidence are admitted, even before anything further is said about what they might be, it is easy to
find examples of comparisons of degrees in which strong confidence seems reasonable.
A fundamental example (if not entirely uncontroversial) is the relationship between a claim and the evidence supporting the
claim.
Unless the claim is logically entailed by the evidence we have for it, it may (logically) prove false even though the supporting
evidence is sound.
This suggests that our confidence in the conclusion perhaps should be less strong than our confidence in the evidence.
Even in the case that the inference is logically sound a comparison is justified, the conclusion will be no more certain than
the premises (assuming that the argument is our only reason to believe the conclusion).
In the face of doubt about the conclusion, we may therefore retreat from asserting the conclusion to describing the evidence
which leads us to suspect that the conclusion might be true.
This example is central to the moderation of scepticism which became known as positivism.
Taking David Hume to be perhaps the earliest positivist (despite his having preceded Comte who introduced the term) we find
a clear distinction made between analytic truth (which he described as expressing relations between ideas) and empirical truths
(matters of fact), the former alone possessing any degree of certainty.
Positivists affirm that "positive science" should go no further in its prouncements than is strictly warraned by the evidence,
and that the scientits conclusions should therefore go no further than to summarise the experimental evidence.
A general positivistic epistemic retreat might therefore be effected by taking all scientific laws as empirical hypotheses.
The derivation of results from such hypotheses, following a nomologico-deductive method, would deduce particular conclusions
from some collection of laws and initial conditions, and would assert no more than the logical (analytic) truth that the conclusions
obtained are logical consequences of the given premises.
The evidence for the hypothetical laws might then consist in empirical observations checking that what happens under the relevant
circumstances complies with the predictions of the theory.
Instead of asserting the theory, the scientists provides a summary of the observational data.
David Hume did not accept "relations between ideas" even when supported by mathematical proof, as beyond doubt.