In reply to Robert Durran:
> I am afraid that I cannot see how knowing the names of the two boys effects the problem. We still do not know how the two children were selected. At random? As many boys as possible? The two eldest children? In the absence of this knowledge, the problem is ambiguous and the only thing we know for certain is that the family includes at least two boys.
We can also ignore the names and do a population sampling approach to get 1/2.
Of all the three child families containing at least two boys, 1/8 will have three boys and 3/8 will have two boys and one girl (the former has to be BBB, the latter can be BBG, BGB or GBB, focusing on birth order).
So, if we see two boys run out of a house that we know contains three children, we know that there was a 3/8 chance that the house had two boys and a girl in it to start with, and a 1/8 chance that it had three boys in it to begin with.
But, the puzzle doesn't ask for that answer. It asks for the odds on the sex of the child left in the house. To get to that we have to make a further comparison, which involves answering these questions:
1. Of houses containing three boys, what is the chance that the first two children out will be boys?
2. Of houses containing two boys and a girl, what is the chance that the first two children out will be boys?
The answer to 1. is that it is certain, the answer to 2. is 1/3.
We now notice that while it is three times more likely that households containing three children, at least two of whom are boys, will contain one girl rather than no girls, it is three times less likely that the first two children to run out of such a house will both be boys. Therefore, if we see two boys run out of a house that contains three children, at least two of whom are boys, there is a 50:50 chance that the remaining child is a boy.