Typically the flow goes like this:

      There is a seminal paper, which sets the tone for some topic.

      That first ray-tracing paper.
      That first inverse kinematics paper.
      That first automated behavior paper.

      Corporate university research is applied to this.
      This becomes a specialty.

      And the Ph.Ds and Master's students start marching through Siggraph and other conferences with papers on this topic.

      Cornell, for instance, has been focusing on radiocity and other types of rendering for a long time. It gets incorporated into off the shelf software.

At a certain point, everybody figures out there's nothing new in line drawing algorithms.