I agree that it would be good to leave the question open. I found the various answers interesting but far from definitive, and I can imagine some better answers being given.

A common pattern on MO is that an old, popular soft question with very many answers resurfaces after a long hibernation and is quickly closed. Perhaps the those who voted to close had this pattern in mind. But I don't think the "How many mathematicians?" question fits this pattern since it had relatively few answers and could benefit from a few additional, better answers.

Since so far no one (other) of the people that voted to close commented, I feel some obligation to say something in particular since some details so far got a bit blurred and there is even some factual misinformation around, yet will try to keep it essentially 'facts only' (in the way I perceive them), as I promised (myself) to essentially stay out of such discussion.

I woted to close this question (I was neither the first nor the last to vote, most likely 2nd or 3rd). Here is why:

1. The questio became reactivated by an answer that was at the borderline between massively off-topic and spam (now deleted); it was if memory serves well some sort of request (not sure whether it was serious or ironic, due to the way high-school teachers are exclude from being mathematicians in the question, or purely fictional) of some high-school(?) teacher asking how to improve in maths. (So this question did just attract a very low quality answer, cf. bsteinberg comment).

2. Contrary to what is claimed by OP, the question did not get four answers more than two years ago, but it did get three essentially right away in Nov 2009 (so more than two years ago), yet than an additional fourth one only in July 2011. And, this final one seems to me, based on comments, votes and my own judgement, considerable worse than the earlier ones; it got an estimate for US math PhD totally wrong, as pointed out by comments. (So, I observed a steep decline in the quality of answers, cf again bsteinberg comment.)

3. The question has an accepted answer (unusual for CW!). To me this means OP got a completely satisfactory answer and even might not be interested anymore. The latter seems even more plausible as there is no sign of OP stilll being interested/paying attention to the question; via following up on comments or alike. In particular, a specific request for clarification (comment by Martin Jan 2011) received no answer by OP or anyone else. And, this while a new answer was even added afterwards (also no activity by OP related to it, while otherwise quite active commenting); so it is not just one comment that one might have missed.

For all these reasons, I saw simply no reasons to keep this open. But I have no problem if it stays open if some people still care; I would even have voted to reopen myself if I had seen the request in time.

One personal comment: Not sure if the name-calling was necessary. In particular, 'macho users' IMO is really a poor choice of words.

As a side remark: IMHO, the wisdom of crowds can be well sampled by going into the Bigg Market on a Saturday night, or reading the comments left on the Guardian's Comment is Free blog/site/jungle.

First of all many thanks to the users who reopened the question in the astonishlingly short period of a few minutes.
Actually the closing of this question was essentially an opportunity to make a case for soft questions.
I wanted to argue with a specific example at hand that contrary to a widely held point of view, it is easy to get answers to hard technical questions by asking the right specialists. The only obstacle is the self-restraint that a decent mathematician feels toward imposing oneself on a specific colleague. MathOverflow is an ideal remedy: people will answer if they have the time or are in the mood and we can thus ask away without scruples.
The paradox is that softer questions cannot necessary be answered by these awesome colleagues and that there is no alternative to asking on MathOverflow, hoping that a few among the thousands of users have some information and/or out-of-the box idea as an answer. It was probably a mistake to mix up in this thread the specific question "How many mathematicians are there ?" with these general considerations, on which I would however still like to have our community's opinion .

Dear Georges, it seems that you took my comment as an affront? It was not meant that way: all I meant was that back in the day, it seems, very soft-questions were not only more tolerated, but fairly well received. These days, any question as soft as the one under fire here, will probably succumb to rapid closure, and perhaps pointers to the MO Faq. As an anecdote: when my MO questions were less mature, I once asked a question "which math problem has witnessed the largest number of different proofs" --- this question rapidly evoked a vehement closure response from the community, and I just deleted the question; now-a-days I wouldn't dare asking such a soft question!

Moreover the huge number of users due to the success of our site makes it preposterous that just five people with no special legitimacy other than having passed the low barrier of 3000 "reputation" points can prevent all other users to interact with a question.

I wouldn't be opposed if this limit were raised to 10K or 100K, so that only the "highly legitimate" people may vote. But then I guess it is equally preposterous to accept "up votes" from other "un-informed" users with just a paltry 15 reputations points? But then, aren't these the agreed-upon rules of the site? They could be changed if the community decides to. Weighted majority, while not perfect, might perhaps offer a more satisfactory solution to give "weighted legitimacy"?

I wonder what the SE2 rules are on closure? The same as in MO1 now?

@Gerry: Thanks.