The first thing to be discovered was that pyramidal 
     organizational structures, with many layers of 
     supervision, and with functional division by specialty, 
     simply did not work. The communication chain between 
     top scientific or engineering leadership and work centres 
     was too long for either the scientific or managerial 
     message to be communicated. But in these research 
     organizations where work actually got done, when one 
     studied them he found that whatever the organization 
     chart prescribed, groups of researchers with different 
     competences as required by the problem in hand were 
     working together, cutting across organizational lines; 
     that they were establishing most of their own design 
     criteria for the work as well as their intended patterns of 
     association; that the patterns of their group association 
     at work followed the organization of their competences