Also, not all community members are created equal. I used to be a mod/product lead for a major IT community forum. We had a group of 300 or so hardcore posters

Also, not all community members are created equal. I used to be a mod/product lead for a major IT community forum. We had a group of 300 or so hardcore posters (which the ad team referred to as "inventory builders") who we took behind the curtain and showed our prototypes before we released them. They often had some interesting feedback, but we quickly fell into a strange trap: The changes the hardcore users wanted didn't drive usage -- posting or reading -- when we implemented them. These guys were already spending every damn spare moment posting in our forum, so they had no more to give and all the esoteric bells and whistles (segmented follow lists, advanced karma scoring, multi-tiered private messaging) they wanted were completely irrelevant to the casual user. When we shifted to testing changes that drove casual usage, we saw much better results. The hardcore group wasn't against it, generally, except for those petty few who saw the forum as their private playground and didn't like the riff raff wandering in to post. They were just dumbfounded as to why we were wasting time on these "beginner" features. You don't want to alienate the devoted userbase, but data -- not anecdotal feedback -- is what should guide your decisions.