Let me be a contrarian and state that the vast majority of the downvotes on poorly written first questions by new users have been deserved. There are several things in a question that will attract downvotes like flies:
- Demanding that people give them code
- Using textspeak
- Demonstrating no research at all (see Matt Gemmell's definitive article about this)
- Asking us to debug their application, usually with no stack trace or console output
- Being completely incoherent
- Writing a rant that attacks one or more popular pieces of technology
Any developer with any common sense will know not to ask something containing one or more of these elements. This is not something that we should have to train new members on, because this kind of behavior is not acceptable in any professional forum or mailing list. Quite frankly, in almost every case that I've commented and explained why they got downvoted for doing this, they kept right on asking bad questions in the same way.
As Jeff has stated, "the world is awash in questions" and we can afford to be selective of what's allowed here in order to maintain a high signal to noise ratio as the site grows. If someone wants to have their question be treated well, they can put in a little effort when asking.
However, this doesn't mean we should be jerks about how we treat newcomers. We should be polite, professional, and provide constructive criticism to those who are open to improving. If there's a good core question that is roughly worded, you're right in that we should edit it to expose the quality content. I've edited nearly 1500 posts, and one of the most rewarding things on the site is to watch an edited question reverse from being downvoted to massively upvoted.
Downvotes, close votes, and the quality filters exist for a reason. I feel that removing the penalty for downvoting questions, the addition of the automatic question ban for repeat offenders, and the expansion of the close votes and tools has led to a noticeable improvement in the quality of the site over the last year.