I'll stick my neck out a bit here and say that questions such as this could be good provided that answers could theoretically be supported with citations, history and other facts. Just as Programmers SE asks that people post answers backed up by their experience, answers to this type of question should be able to be validated beyond the strength of one's convictions or opinions.
A fantastic answer to that question would not just summarize the arguments for and against compulsory voting, but also describe how it's worked and changed things in practice in the countries that enforce it.
Answers like:
I personally feel that compulsory voting is horrible because moderates are typically uninformed and just pick a candidate at random
... are far from useful. However, if you have actual data that indicates the above - the answer ceases to be an opinion:
In the People's Republic of Atwoodistan, compulsory voting unearthed strong evidence that moderates simply choose candidates randomly. The effects included ... .... ....
I agree with Casebash that unnecessarily partisan responses should be down voted, and I'd extend that to say that questions that are likely to only receive that type of reply, or pure opinions should be closed as not constructive. However, in this case, useful answers have appeared, and there's a clear potential for even better answers to be posted.
This seems to be more of a question on how 'good subjective / bad subjective' should be interpreted (well, more implemented) here.