I think the MB structure for classical music is not its limiting factor, and it would be unwise to focus too much on that until it is. In fact, the more people understand the structure, the more I am convinced they will like it and appreciate its intelligence. I think the biggest practical problem is the sheer amount of work required for somebody to enter a new classical album into MB (having entered maybe 40-50 albums myself over the last 6 months). It is a strong disincentive.
Part of the problem is the amount of persnickety detail that the style guidelines require. I'm not saying that they are wrong per se, but it is a major headache to attempt to keep on-side of them all, particularly when your gut reaction to something is that it doesn't look right. I should be clear that I understand where the style guidelines come from, and why they're needed, and even why most of them ended up the way they did.
As a newbie here, and an outsider, the whole MB thing has the feel of a hacker community with its IRC and user-unfriendly web interface. I am pretty sure it must put people off. I've been doing this since January and I still can't get the hang of scripts. I think a major initiative needs to focus on making the whole thing easier and more intuitive to use, particularly since the whole point here is that we don't just want more albums entered, we want richer metadata entered.
Also, as a 'big-picture' kind of a guy, I can't help but think the most productive single initiative would be one that brought some of the bigger hitters on the labels onside. There are two challenges - one would be to convince them that having their albums catalogued in MB would be in their best interests (as opposed to just not in their worst interests). The second would be to provide them with tools that would make it easy for them to move in the MB direction once the first criterion is met. I have some thoughts on how that could be done, but this isn't the moment to expand on them.
Finally, a perspective from the user community, which is where I fit in. Of all the attributes of a rich metadata set, the one that people probably clamor for the most is genres, and MB doesn't support them. I know why, and I fully understand the nature of that particularly ugly beast, but it is something MB needs to face up to, and, probably, find a way to bite the bullet. But geez, that's going to be tough. Although it would be less tough (I think) in classical.