Whether evidence is direct or indirect is not the measure of our confidence in what we believe true of reality. Eyewitness testimony is notoriously unreliable, while some of the most well established theories of science have only indirect evidence, like quantum theory.
It would be extremely convenient if we could assign a measure of confidence in any given knowledge by a simple enumeration of the relative proportions of direct and indirect evidence, but that isn't the way the real world works. A tight and unambiguous chain of indirect evidence can be of far better quality than ambiguous direct evidence. Evidence must be judged in context with other evidence, and it is this complex confluence and interaction of evidence that we assess.
--Percy
™ Version 4.2