Pablopikachu
This is exactly what I've been thinking, good thing someone mentioned it. The 1st page is full of the same method posted over and over again even though it's flawed.
You have quoted a post that claims to quote Ron Purewal, but I'd be very surprised if Ron is actually responsible for that quote, because it misunderstands a basic mathematical point, and I would not expect Ron to make a mistake like that. The solutions on the first page of this thread are essentially correct, even if they sometimes do not mention one detail (the detail that the post you quote claims is a serious problem). The solutions really should say the following (it is the final bullet point below that the earlier solutions often do not describe in detail) :
• say we label the letters A, B, C and D
• then the probability letter A is in the right envelope is 1/4
• assuming that, the probability B goes to the wrong envelope is then 2/3
• now to continue solving this way, you must focus on the right letter next. One of the two remaining letters is guaranteed to end up in the wrong envelope, because we just put letter B in its envelope. You can't focus on that letter third, because then you won't know what to do when you come to the fourth letter (you won't know if its envelope has been used or not). You must focus on the other letter third. That letter has a 1/2 chance of ending up in the wrong envelope, and the final letter, as I just mentioned, automatically goes in the wrong envelope.
So the probability A is in the right envelope and the rest are in wrong envelopes is (1/4)(2/3)(1/2)(1) = 1/12, and multiplying by 4 since any of the four letters could be in the right envelope, the answer is 1/3. The solution is fine -- you just need to work through the letters in a certain order.
That's the rationale behind the solutions on the first page, and it is certainly correct. I did post a slightly faster way to do this earlier in this thread also. I'd add that the post you quote claims this method is wrong, because it fails when you try to apply it to 5 envelopes. But it obviously "fails" in that case, because, if you work out by hand how things work with five envelopes, you quickly realize that you have lots of cases that just don't happen with only 4 envelopes -- it is (perhaps surprisingly) a much more complicated situation. For example, with two letters left, when you have 5 letters and envelopes, it might be that both are guaranteed to end up in the wrong envelope (because both of their envelopes have been used) or it might be that exactly one could end up in the wrong envelope, or it could be that each still has its correct envelope available. To solve with 5 envelopes, you would have to work out the probability you end up in each of these cases (unless you use a derangement formula or some other method -- regardless, you will not need to worry about this on the GMAT, because it is too complicated for the test). But none of these complications appear anywhere in the four envelope situation, which is why the problem in this thread has a simple solution, but the 5-envelope problem does not.