You launched early and your users' feedback is all useless or wrong. How to get useful feedback?

fix your bugs and apologize to your users.

How could all your user feedback be wrong?

No. I didn't say that they don't like it and there have been no noticeable bugs.

It was very well received by all of those who I heard from. The problem is that the whole "release as early as possible and get user feedback quickly" strategy has not been effective for me at all. My question is how I can get better feedback from users, not how I can suck less.

I think the problem there is that you aren't understanding what they're asking for. I'm sure they are being perfectly clear in telling you what they want but something weird happens when users describe features and wish lists. They often don't know what they want. Finding that out is YOUR job.

IMHO, what you need to do is ask them what would make it better and then dig in to find out what they really mean. Why do they need that feature? There's no script or formula to do this, but that's one question I've found to be useful.

For example, User A says: put a link on my profile, which when clicked shows my age, sex and location.

You should ask, "why do you need that?"

The User will often respond in some way which will give you a clue as to what they're really looking for.

So, the response to that question might be, "oh, well, people always email me to ask me what my a/s/l is and I'm tired of telling them."

So, using this (very lame!) example, they're not looking for a link - they're looking for a way to show their a/s/l on their profile. You can give them what they really want, not what they're asking for.

I hope this makes sense.

You shouldn't blindly implement features just because your users are telling you to do so. While it is important to listen and react to your user's concerns, it's also your responsibility to maintain the integrity/consistency of the application. Also, if you don't believe that the feature you're building is going to have any effect, then it's already heading towards failure because you're not going to be committed to its success.

I agree with jkush -- if you don't agree with the user, you should engage the user and see if you can get the feature to a common ground that you can agree with.

Also, if your feedback loop is just an email, maybe you should add a forum to make the loop more interactive.