MindManager sucks

or for that matter all mind-mapping applications that I know.

Although it offers a really nice, effective and lightweight way of _structuring_ concepts, my problem right now is that I can not use it because I can not force the concepts in a simplistic tree.
In my head the concepts pop up in different places depending on the context  or point-of-view one is currently looking at. What I’d like to have is something where I can mesh concepts (and actual text) in 1..n qualified graphs – depending on the point of view / context / focus, without re-creating relations (simple aggregations or level of details) in various documents.
All this with an UI that is not getting in the way (aka I am looking neither for  a Wiki not a  Microsoft product)

Help!