Make That 1.5 Qualms

Apropos the previous post, I had another thought regarding defeasibility. Switching from a presumption of innocence to a presumption of guilt introduces another complication due to the assymmetry of proving something did happen and proving that it didn't.

In general its easier to demonstrate that something happened then something didn't happen. An event that occurs will often leave some sort of discernable traces in its wake, positive evidence which can rebut the presumption of innocence and thus satisfy the requirement that such a presumption be defeasible. Put more clearly, when we ask "What sort of evidence would we accept to prove that event X happened?" its pretty easy to come up with a long list for a wide variety of common Xs.

Now consider the opposite case of trying to prove an event didn't happen. A thing which does not happen will, by definition, fail to leave traces, so rebuttal must happen indirectly by considering the state of the world and seeing if its logically consistent with what we'd expect to see if some event X didn't occur:

• The dog always barks at strangers in the backyard.
  
• We didn't hear the dog bark.
  
• Therefore the state of the world is consistent with the hypothesis that a stranger did not enter the yard.
  

Which is not to say that no direct evidence will ever exist ("Your honor, here's a videotape of the yard during the time in question, which demonstrates conclusively that no one crossed the yard"), but rather that such evidence will be less common and the challenge of proving an events' non-occurence more difficult.
But there's another hitch as well: Sometimes an event is know to happen but leave discernable traces. In this case the state of the world will be consistent with the hypothesis that the event happened and the hypothesis that it did not:

• The dog usually barks when a stranger enteres the yard.
  
• We didn't hear the dog bark.
  
• Therefore it's likely, but not 100% certain, that a stranger didn't enter the yard.
  

In this circumstance its possible to build a probabilistic case that the event didn't occur, but doing so be more difficult than the opposite.
All of the above points to the challenges inherent in proving that something didn't happen compared to providing that something did. Such considerations inevitably become fact bound, so its diffcult to generalize beyond the following: When evaluating a claim its important to ask what evidence could be produced to prove, at whatever level of confidence is desried, the truth of that claim. If that list is empty then the criteria of theoretical falsifiability is violated. If that list is non-empty, but such evidence as it contains is unavailable for practical reasons, then the criteria of practical falsifiability in violated. In either case there's a failure of due process; it is not fair to bring a charge if there is no evidence which could reasonably be produced to disprove it.