I was chatting with people at SIGGRAPH and remarking on the fact that there is only one Turing Award winner for graphics (so far). The people there made the interesting argument that one of the reasons theoretical contributions appear to get Turing Awards more frequently (more on this below) is that theoretical papers tends to frame a problem in the context of related work far more meticulously.
Now this is something I have heard in other contexts as well (when seeing reviews of papers, reading papers etc). It is not too hard to see why this might be the case; theoretical problems are usually crisply defined, and it is almost always clear what results/techniques influence a given work (though not completely clear).
As an argument in favor of sound referencing, I like this one. We are taught (implicitly or explicitly) that referencing is important to place work in context, to acknowledge other's work in the "society of research", and ultimately to recognize that one's one work is never done in isolation but builds upon that of others. The argument that proper referencing creates a trail back to a source of inspiration is more than 'citation counts count !': it emphasizes what's important about careful referencing: that it builds a structured edifice of knowledge whose value can be perceived objectively.
Back to Turing Awards:
According to my highly unscientific classification method, theory-like disciplines (Algorithms/Complexity/Logic/Programming Languages) took a large, but not overwhelmingly large share of Turing Awards:
(Contains theory-like substance): 18 (11 Alg/Complexity + 6 PL + 1 Logic)
(We build stuff): 10 (3 DB + 7 Compilers/Arch/OS/)
(We are stuff): 5 (4 AI/1 Robotics)
(We count stuff): 2 Numerical Analysis
What category does Dijkstra fall into ? I left him out of this...