The implication of this "problem" is an interesting human biased: there appears to be cases where we judge the more detailed explanation as more probably, while statistically speaking, it isn't. If a person is described as "being concerned with discrimination and social justice", we tend to judge P(X = activist and Y) as more probable than P( X = Y). This is detailed in "Thinking, fast and slow" by Kahneman - a wonderful book if anyone haven't read it.
I did take notice that a sub-set is likely to be smaller than the set it came from: the set of all accountants vs. the sub-set of all accountants who are also plumbers. So therefore P(accountant) > P(accountant and plumber). Unless P(accountant) be zero, or P(plumber|accountant) be one, in which case neither is greater.
Reflecting on my own human biases, I think I ended up reading it as if this were a multiple-choice question, and I had to pick the "best" choice. In other words, it isn't enough that this person is an accountant, but that this person is also a plumber. Or put another way, if I had to wager on whether anyone would pay just any old accountant to go mess with their water heater, I would find it more surprising that the particular accountant in fact hired had no demonstrable plumbing skills. I'd imagine Kahneman could teach me more on this.
My favorite word problem joke goes something like this. If there are three cans on a fence, and you hit one with a thrown rock, how many cans remain on the fence? Answer: Two. Now, if there are three birds on a fence, and you hit one with a thrown rock, how many birds remain on the fence? Answer: None - they'd all start flying off when there's a rock hurtling towards them.
So the question is really "Is this person someone who accepts money, or is this person someone who accepts money and is a licensed plumber?"
If you interpret it literally, it uses the law of the excluded middle inappropriately and would be just as true if you replaced "accountant" with "President" or "extraterrestrial angel."
I'm more inclined to believe this is a (poor) illustration of the idea that a sub-set is likely to be smaller than the set it came from: the set of all accountants vs. the sub-set of all accountants who are also plumbers. In which case - duh! - P(accountant) > P(accountant and plumber). As long as one can indulge in the naivety of this pedagogical tale, that the narrator (the "I" who said "I went to a person") wasn't too particular on whom he hired for the job, then sure, why not?, 'twas likely the accountant that fixed it.
In the real world, which has teachers and mechanics in addition to accountants and plumbers and other occupations, and where people may tend to be pickier on who they'll let near their water heater - leave alone pay someone to mess with it - would this conclusion of "accountant" still prevail necessarily?
Event A is "accountant and plumber". Event B is "accountant and not a plumber". Hence A unioned with B means "accountant"; or, if that's too general, "is an accountant and may or may not be a plumber too". In any case, both events A and B qualify the person to be an accountant.
So what's the controversy? When it comes to water heaters, we're all accountants?