Harnessing advanced customers is something I really believe in. It's also one of the best foundations for startups: You should create a company for which you would be the ideal customer.
In the past, I've been in situations where my suggestions were brushed aside because I was not considered a "typical user", and hence was considered useless for market research. But I always firmly believed that to be the wrong approach, as in any field "typical users" never stretch the envelope. They know of a limited subset of features.
Limited subsets does not create happy user.
As previously discussed all of those seemingly useless (to the "typical user") features are highly likely to be used by somebody, and almost every unusual feature is likely to be the one feature which makes a couple of your users pick you over the competition.
Advanced users are valuable not only because they are likely to be intelligent and innovative, but because they are users that are likely to exploit otherwise unused funcitonality and stretch an application to its limits.
Users like that will inherently in many ways represent a wide subset of an applications user base in that they will know features that are in use by many small, non-overlapping segments of your userbase, and can compare features against each other and can see why these groups of users like or don't like particular features.
Building a startup around what you want for yourself instead of what you think the market may want is the way to go, because the market doesn't have intelligence - the market is just the amalgamation of users that all know about a feature or two they want that most other people don't want.
If you listen to "the market" you will end up with something that meets most needs of most people, but make hardly anyone fully satisfied.
Meanwhile your competition will be busy listening to their advanced users and adding all those seemingly useless bells and whistles.