On suggestion of one of the senior members of MO, I have decided to start a discussion on how to post a question and possible revisions so that it gets accepted.

Recently I posted a question(http://mathoverflow.net/questions/89556/dangers-of-loose-hypothesischanged-the-subject-title) which I felt would be a interesting topic but was not accepted on the grounds of 'not a real question' and 'subjective and argumentative'. My gut feeling is more than the topic of the question it is the manner in which I had asked which must have put off most of the mathematicians. I strongly feel the question is interesting but it needs a revision, may be a major one.

Some possible modifications could be removing subjective elements like small and big and highlighting the pedagogical aspect of the question.

I would really appreciate if some one in the community suggests how to go about this process so that the question is reopened. Thank you.

EDIT: A small change in hypothesis like, changing the space from Zygmund class of functions to Log-lipschitz class or looking for an integer solution in a equation where the solution is easy to find in real or complex case. For instance, Fermat's last theorem is easy to solve on real-line or complex-plane but it engaged mathematicians for three centuries in the integer case. (I have just turned around the example in comment by B. Bischof given below and it is suitable here!)

I thought, and still think, that changing from "asking for real solutions" to "asking for integer solutions" is neither a small change nor a loose hypothesis, but a drastic and dramatic change of the problem.

My misgivings are that the question seemed to invite somewhat superficial answers, and would attract more in the future, getting bumped up to the top each time someone had an idea. (Changing from $L^p$ to $L^1$ is not a small change, it is a big change, and not so much a case of a loose hypothesis as a cavalier approach to precision.)

I don't see how the question could be made interesting. Take pretty much any universally recognized theorem that's considered "hard" in mathematics. Almost always there's a textually (perhaps not conceptually) small modification to the statement that would turn it into an easy true or false statement. My problem with the thread is it pretty much is asking for an endless list of recognized "hard" statements together with various easily-produced modifications of them.