So far a decent chunk of the questions I've asked on this site are open in the research sense. E.g., a question like "does problem X have complexity Y" will result in answers giving the best known algorithms with complexities > Y, while also confirming that no one knows whether Y is possible.
How should such questions be handled? Should I mark as accepted an answer that I'm confident describes the best known complexity, or leave them open to indicate that no one yet knows? There are arguments either way: in many cases it's unlikely that the question will be completely closed any time soon, but leaving it open does reflect its actual status.