 |
| A "Rube Goldberg" Machine |
I discussed the "keyhole dilemma" which asks how do I know if I have enough information to correctly diagnose or solve a given problem. The analogy I used was to look at some process (say a Rube Goldberg machine like the one pictured with this post) through a keyhole - which limits what and/or how much you can see of the entire machine. Given you cannot see all the parts of the machine how can you know if what you think is wrong based on your view through the keyhole is enough to correctly diagnose and fix the problem.
(For example, if I could only see the left half of the picture I might conclude that there is a very simple mechanism engaged between "D" (the ball rolling down the ramp) and "P" (the rock squeezing the toothpaste onto the brush) - which would be incorrect.)
In the worlds of physical science, computer science and mathematics humans use techniques to address this dilemma - at least to a degree. For example, determining how a cell divides through biological investigation, using a proof to demonstrate something like the Pythagorean Theorem, or debugging a program.
The same is true to a large degree with the US Justice system and juries.
We see an initial situation, in the case of Casey Anthony her with her child alive, and a final situation, her child is dead. But we are forced to look "through the keyhole" at the process of how things got that way (so we might see "A" in the picture above - throwing the switch, and the toothpaste on the tooth brush - but none of the steps in between). Therefore we do not know how the result follows from the initial circumstances.
Someone like a prosecutor creates a "theory" about how the missing events happened which is provided to the jury. (Now, if the prosecutor knows (has direct knowledge) how it happened, i.e., with some sort of concrete evidence, say a video or a confession, that's a different case. Here we are interested in when only a "theory" is offered the jury.)
So the prosecutor describes a series of events to fill in what happens between "A" and the final result. Of course, in real life there may be several "keyholes" in which case the theory must account for what is visible through each.
What is interesting to me is while their may be any number (perhaps an infinite number) of "theories" about how we get from "A" to the final result how do we find the correct one? Is it even possible?
In many cases it is in fact impossible to determine what is "not seen" through the keyhole - there simply does not exist a record of the missing events or process. To address this type of case prosecutors use what's call "circumstantial evidence" to attempt to fill in the gaps.
"Circumstantial evidence" works like this. If I see Joe shot Ann with a gun - that is direct evidence. However, if I see Joe and Ann go into another room after which I hear, but do not see, a gun shot and I later enter the room to find Ann dead - that is circumstantial evidence that Joe killed Ann.
This process of inference about what happens that I do not have direct evidence for is often used in criminal trials to convict people of crimes.
The problem with inference, particularly in a criminal trial like that of Casey Anthony, is that the process used is neither scientific nor objective. Scientific inference, e.g., I find Joe's finger prints on the gun, may lead me to conclude that Joe in fact held the gun at a given point. But the same is not true for concluding that Joe shot Ann with the gun.
In the later case I am inferring something for which there is no possibility of actual proof. Two people went into a room, one is dead, the other one must have killed them. But this is simply not a fact, nor a provably correct inference.
Now let's imagine a slightly different scenario. We have two keyholes - through one we see Joe fire the gun in the direction of Ann, who we can see through the second keyhole. Now, when we see the gun fire and and drop dead, we can also make a similar inference.
Most people would agree that the first inference (where we only hear a sound) and the second where we see two different but apparently related events are not quite the same. There are various other scenarios we can imagine as well - for example seeing a hand with a gun and the gun killing Ann - but not the holder of the gun, and so on.
I argue that the difference in value as evidence between two events where inferences is used to "fill in" what happens between them is based on probability. For example, seeing Joe and Ann enter a room and hearing a shot leaving Joe as the killer has a 50% probability that Joe killed Ann versus seeing Joe fire the gun through one keyhole and Ann drop dead through another has a 75% probability.
Note that this is an illustration and the point is that different people might assign probabilities differently when weighing two scenarios. Objectively it would be easy to argue about which scenario I provide is more or less likely to make Joe a killer and even though this is essentially a trivially simple case its easy to see how their might be strong disagreement.
Now, as in the case of Casey Anthony, we mix in heated emotion: Ann is my wife or Joe is my husband. Ann is a cute child and Joe is an ugly brute.
Would the assessments still be the same?
This is the first problem with "circumstantial evidence" - not being scientific or objective it is easy to sway those involved such as a jury with emotion. Such emotion also sways the observers of such a trial.
So this is my first problem with "circumstantial evidence."
Is it fair to allow emotion to drive or infiltrate inference? Clearly it does and this trial is a perfect example.
Finger print evidence is "scientific inference" - but inference not swayed by emotion, i.e., they are still my fingerprints whether or not I or you are angry. This is a fact because anyone can inspect my fingers and the fingerprints and decide if they are mine (though if they are, for example, not clear then this may fall into the realm below).
The injection of emotion into the process of inference colors the inference. It makes determining the probabilities of which scenario is more likely harder and unfair because it becomes driven by external factors instead of what ever direct evidence there actually is.
A jury must decide the probabilities of both the defenses or prosecutions scenarios. Since the actual facts are unknowable the result comes down to non-scientific inference.
Which airplane would you want to fly on: the one where the designers "felt really good" about the design or the one designed on "scientific evidence".
Or how about the one where designers where told "they were the greatest designers on earth" or "the biggest idiots on the planet" as they were working through the final details?
No, circumstantial evidence (the result of inference) are only valid if the inference process is objective, i.e., my thinking of how A implies B implies C is done objectively.
Since this in virtually never the case where emotion is concerned or involved I must conclude the circumstantial evidence is unsuitable for logical conclusions.
I think this was the case by the jurors in the Casey Anthony trial.
This is troubling particularly when I see prosecutors in an emotional frenzy: are they really after true justice or are they really just trying to trick jurors into making objectively invalid inferences based on their emotions and feelings - which is not a fair trial.
No comments:
Post a Comment