What is it that makes witness testimony evidence, or is it evidence at all?
It eliminates the inductive leap.
"Circumstantial" evidence is generally dismissed as "flimsy" evidence, but in fact -- so long as the induction is not thin -- it is more reliable than eyewitness testimony.
Postulate, say, a large gothic mansion. A scream is heard from the next room. The listeners run into the room to see a dead man, stabbed, and another man holding a bloody knife and standing over the corpse.
What happened? Well, all you have is circumstantial evidence. You can certainly propose the hypothesis that the man holding the knife murdered the victim with it. But nobody knows if this is what happened. Investigation is the science (and frequently the art) of explaining observations. A successful hypothesis explains all the observations without relying on assuming things that can't be observed (or worse, observed to be different).
A good example over on sci.space.history, where I've just wrapped up a line of questioning with Scott Grissom. In brief, his theory is that a piece of metal was fastened to the control panel in a particular way. Unfortunately the photographs of the control panel don't show any marks, abrasions, or other signs that such a metal piece was fastened there. That alone won't sink the ship. In terms of parsimony it's a point of worry: you expect to see it, but you don't. But that's an argument from silence. Or is it? In fact, in the photographs you can see marks from some other component -- the flange of the switch that was supposed to be there. So not only does the evidence fail to support his hypothesis, it actually supports a different hypothesis: that the switch was fastened there in the normal manner.
The gothic murder would seem to be a clear-cut case. But what if the "murder" entered the room from a different door, saw the body, screamed himself, and withdrew the knife preparing to offer aid, and was thus discovered? In each of these scenarios there is an inductive leap. We may conclude that the most likely explanation for the observations is murder by stabbing, but absent a way to falsify it or any other hypothesis we are left with the inductive skepticism.
An eyewitness to the process suggested by the hypothesis is essential. Some way to falsify one or the other hypothesis reduces the breadth of the inductive leap. To be sure, that leap can be narrowed by additional circumstantial evidence. But until there is evidence that the knife penetrated the victim's flesh propelled by the accused, there will be inductive skepticism.
Obviously a case made solely on the basis of witnesses is hampered by the possibility that the witness is lying or mistaken. Many independent witnesses will mitigate this. But the best case is still to be made by a combination of eyewitness testimony and circumstantial evidence. Neither alone is conclusive, but both together can be.
Conspiracists take advantage this variance and often like to "mine" the witness testimony for discrepancies as direct evidence of a coverup, etc.
This is simply an abrogation of the historical method. Inconsistency in and of itself does not prove impropriety. The opposite, in fact, is true. But neither conspiracists nor their readers understand this. "Common sense" tells them that if something is true, everything said about it must be consistent and complete.
Now in legal circles there is some disagreement about this. One impeaches a witness by showing his testimony is inconsistent, or that he has lied in the past (thus making him untrustworthy). The rules of evidence that apply in law are not necessarily the best rules to apply in an investigation that may legitimately fail to draw a conclusion.
We have a fairly good grasp of how the image points on a photograph came to be, from the action of the light source on the objects, through lens, to the film, and through the development and reproduction process. But what is the corresponding mechanism with witnesses?
Exactly that. Something is philosophically true or not. It either happened a certain way, or it didn't. The fact that a photograph reproduces without passion the image of an event does not make it that event. The lack of passion in the recollection does not ensure that every detail of that recollection will be clear, unmistakable, and correctly interpreted. Thus an eyewitness' testimony is considered true if it is generally self-consistent, generally consistent with toerh recollections, generally rooted in known good facts and principles, and generally free from interpretational bias or extrapolation.
|