Even when witness descriptions are similar, to prove two people are the same man, part of the proof process is to prove that they are not different. Meaning proving he (x2) are not other random men with similar looking attributes. We must prove these two men are same man, but also prove that they are not different other random men.
In our legal system (as I understand it the basics of it), part of the system is proving that the offender did the crime, but the other part is proving that the offender didn't do the crime by virtue of presenting reasonable doubt. "
Just because a person has been declared "not guilty", it does not mean that he is innocent. "
A similar process occurs in science and the use of the null hypothesis and alternative hypothesis in the experimental method.
In science we consider two sides of the same question of whether there is a relationship between two variables.
a) null hypothesis = NO statistical significant differences exist between two variables or parameters.
b) alternative hypothesis = statistically significant differences exists between two variables.
Using our current 'taxi man' and 'apartment man' example.
If we assume these two men are the same man, not only should we look for evidence to prove that they are the same man, (eg same or similar height, hair colour and style, body shape, two legs, two arms and a head). But we should also look for evidence that disproves the assumptions that they the same man, meaning we set out to disprove the null hypothesis.
We need evidence to prove they are the same man AND to find NO evidence that disproves (or rejects) the assumption that they are same man. The latter part meaning we must look for evidence that shows they are different men. If we don't find evidence showing they are different men, (we can then reject the null hypothesis) and then we can assume they are the same man.
....and yes it took me ages to write this post and get my head around it!
(happy to have some science nerd proof read and critique this post for me!)
https://www.thoughtco.com/null-hypothesis-vs-alternative-hypothesis-3126413