- information visualization point of view
- real-world data sets are large - millions of nodes
- exploration vs explanation
- sacrifice generality to achieve scalability
- restricted class: quasi-hierarchical
- push on scale, leave abstraction for complementary work