This whole Sherlockian concept of perception and induction (Sherlock Holmes didn’t deduce but induce. Going from specific to general is induction) is highly subjective and that’s its major drawback. It’s limited insofar that it operates on small variables – details – that should enable us to induce the “larger picture”. Our perception of details is relatively objective: we can say that most people perceive the same thing in a pretty similar manner, but interpret it differently.
Absence of evidence is not evidence of absence
If a person does not have any animal hairs on their pants, it would be a very long shot to conclude that the person in question doesn’t have an animal. If, however, a person has animal hairs on their pants, we conclude that the person either has or has been in a contact with an animal. By analogy, if a person wears a wedding ring, we conclude the person is married, and if there is no ring, we shouldn’t conclude that the person is not married. Or?
Need for statistical knowledge
What if the number of people not wearing wedding rings corresponds highly to the number of people not married? From a purely subjective perspective, I think that most people that don’t wear wedding rings really aren’t married. But that’s just my interpretation and that’s the drawback. It would be best to have a statistical research giving us the percentages so that we know for sure. There are many other specific areas that should be investigated (scientifically) in a similar manner. For example, left-handed and right-handed people wearing bags – what would be the statistical percentage of left-handed people wearing their bags on their left side? Of those wearing them on their right side? Of right-handed people wearing them on the left and on the right?
These are some of the questions that beg a specific statistical answer so that one may get a much more precise theory when inducing. I have recently read about the Bayes theorem. It seems to me that its application in induction has a great potential.