One point raised by previous noders has been that the assertion "the burden of proof always lies on the person making an assertion" is often assumed, but rarely proved, and as such any person making the assertion is to an extent violating their own principles of argument, by not fulfilling their burden of proof. While to an extent this is a valid concern, it is flawed in that it is possible for one to make a claim or state a fact without evidence. For example, I can simply say "the sky is blue" without any evidence, and it doesn't mean that I am wrong, I'm perfectly right. However, it simply means that I have not proven my case to anyone who has any doubts that the sky is blue.

As for an actual proof of why the burden of proof must lie with the person making an assertion, it is quite easy to show that if the burden of proof were reversed, you could effectively create an infinite loop, whereby one person asserts something with no evidence (and no evidence to prove him wrong either), thereby "winning" the argument. Then, his opponent could simply assert the contrary position, again with no evidence either way, and by our standards have "won" the argument.

For example, if two people are debating a point:
Person 1 asserts with no evidence that position B is correct.
Person 2 denies this, without evidence, and because in this example the burden of proof is on the person denying a claim, he loses the argument.
Thus, the "winning" position must become B.
Person 2 immediately turns around and asserts position A is correct.
Person 1 has no evidence that person 2 is wrong, and so loses, meaning that position A is now the "winning" position.
Then, person 1 turns around and asserts position B and so on.

Thus, if we were to place the burden of proof on a respondent, then unresolvable arugments such as above could occur, and such a thing is clearly unacceptable. Therefore, the burden must fall upon the person making an assertion.

One other criticism which has been raised is that this principle might be contradictory, on the basis that if one person asserts B, and another asserts not(B), then it is no fairer to place the burden of proof on B than on not(B). In addition, it has been argued that to deny a claim simply because there is no proof is Argumentum ad ignorantiam.

However, these criticisms come more from a misunderstanding of the burden of proof than of any shortcomings of this principle. The fact that one fails to fulful the burden in proving B does not mean that not(B) has been proven, it merely means that B has not been proven.

In reality, the effect of the burden of proof is not to prove that an assertion is false, but rather to maintain the status quo against an unproven assertion, without making any claim of if one is necessarily true or false.

For example, consider a situation where there are only two mutually exclusive possible cases, A or B, and there is no initial position, as the given question has never been discussed before. If person 1 asserts position A without evidence, and person 2 denied their claim on the basis they were shifting the burden of proof, then all that person 2 has done is shown that A has not been proven, they have not proven not(A), which would be in this case exactly the same as proving B.

Instead, in cases where there is no initial position, the burden of proof falls on both sides to establish their position, and until either side can do so, the question remains unresolved. The difference between this and Argumentum ad ignorantiam, is that in the latter argues that because A has not been proven, that not(A) must be true, which is indeed a fallacy, but not the same as the principle of the burden of proof.