Often new members that aren't part of another SE site ask "bad" questions that reflect that they haven't even taken the trouble to Google the topic they are concerned about. An example might be this question. I think those questions shouldn't be downvoted too soon since this might discourage people who have no experience with Q&A sites from asking further (and hopefully better) questions. In my opinion they should be told what's wrong with their question first, and given some advice and time to improve it. What do you think about the matter?
We are about posts, not users. Posts should be upvoted solely on the basis of their merit and not on the basis of the user who wrote them--there even is a large school of thought that rep should be hidden away on posts.
That being said, I am a bit lax on downvoting newbies--in the sense that I comment first, downvote later in borderline cases. But that's all.
(Shameless self plug): You may want to see the voting faq
Personally I think that down vote is not necessary for the moment, downvote is useful when we have a lot of questions and we need to focus the attention in the most useful. But we don't have a lot of questions!
It is normal for users coming from other Q&A sites (e.g. Yahoo answer) to post these sort of questions so be downvoted is very annoying for new user that most of the times don't understand why. Comment is much more constructive.
I strongly discourage downvote on newbie questions! Newbie are potentially good future users. When students ask a silly questions to the professor, I don't think is right that the professor punish them, these is the right method to discourage good questions too. I will use downvote only for spam questions and in some cases for wrong answers.
I think we should copy TeX.SX's policy of never downvoting a question under -1. That shows the user that they've screwed up, without unfairly punishing them. It is also then easier for them to have the question upvoted if they improve it (Which I have seen happen once feedback is given)