We took a look at the problem of benchmark engineering, then at the issues of simulator accuracy and how commercial workloads are not well represented in most academic studies. Both of these are weaknesses in methodology that it is important to understand in grasping the nature of architecture research. 

Slides are here