I think software visualisation is the other direction for future research. It is atleast as important as other directions of research because the present day software systems are so huge that no single person can ever understand the whole system.
The pace at which human cognitive skills evolve is neglible compared to the speed with which we build new software systems. Given the lack of high-level paradigms that might make programming more easier, the time is ripe for researching mechanisms that will extend and aid human-cognition to understand the a software system. These mechanisms should do to human-cognition what telescope did to the human-vision.
Given that we are building systems almost the same way for the past 50 years, it would be a deliberate mistake and hurting in the long run to ignore this direction of research.
Note: The post is deliberately authoritarian (which i aspire to be 
). I am trying to see if i can write that way.
