Instead of posting it as a poll, I would recommend giving open ended questions. While some poll data is valuable, as it focuses on what you want, ideally, your beta testers are there to help point out issues or potential that you didn't think of. Limiting it to a poll means you'll only get back the same ideas your thinking about already.
When asking a poll, make sure your asking questions that you can do something with the answers.
How do the controls feel?
I would expect answers like "good", "bad" and "confusing". If I received answers like "bad", that would tell me nothing about how to change it. You really need to ask why as well. Phrasing the questions differently can get people thinking more open endedly. Such as "Tell us about what you liked or didn't like about the controls."
Are the controls easy/intuitive?
This is another question where recieving the answers of "1" (not easy) only tells me that it needs to change, but not how. Essentially polls should be questions that if you recieve an answer, should tell you what to do next.
Here is an example of a great poll question. "Between Version [current] and version [previous], do you feel the controls have improved, gotten worse, or are about the same?" If I receive Improved, that tells me that the changes I've made are good, and I should keep them in. If I recieve Gotten Worse, then that tells me I'm headed in the wrong direction.
4.How difficult was Level (1, 2, 3)?(1- 5 scale)
5.How fun was Level (1, 2, 3)?(1 - 5 scale)
These are great questions. but let the individuals just answer for the levels they want to. I.e. if no one is saying anything good or bad about level 2, no reason to focus on it, when people are telling you that level 4 is too complicated. But diffuculty and fun are important measurements, that tell your designers they need to rethink this. I would recommend also asking what part of level X was particularly good/bad, and why?
The second thing I would say is don't manage it yourself as a forum. Let the testers tell you exactly how they feel, and let them vote on the importance of other peoples issues. UserVoice.com is a free service for this. Users sign up, (you can also limit registrations to just approved beta testers)
But people will post a bug, a feature request or idea, and use a point to vote on it. people only get a few points. Then all you need to do is look at the items people vote as the largest issues. I've used this in the past with great success.