It is, without doubt, 1 in 2.
The mathematical model showing 1 in 3 is wrong in that it ignores the fact that 2 of the potential 4 scenarios - GG, BG, GB and BB - are discounted by the fact that a man with a boy can't have 2 girls and he has to have had either the boy first or the boy second -...
Either the boy is the first child, in which case it eliminates GB, or the boy is the second child, in which case it eliminates BG. Either way it leaves you with 2 out of 4 options: 50/50.
You've also eliminated GB by saying that the first child is a boy, as is effectively the case here. Therefore, it's 50/50 and anyone arguing otherwise is wrong.
Chances of the other child being an identical twin are approx. 3/100, otherwise it's 50/50. So approximately 51.5% chance that the other is a boy.
Although I fear I may have missed the trick part of this question.