We had a helpful meeting with NSERC staff Rawni Sharp and Judie Foster on June 26.

Several details about the creation and operation of the new GSC structure were clarified,

which I will try to summarize below.

The most relevant document to understand the big picture is the one called "Conference Model Description".

The large units, called GROUPS, formerly called "Broad Area Panels", are the level at which budgets will be

allocated.  Within each group are Sections, and  NSERC's intention is that old GSC 14 would remain a single section, with a section chair, irrespective of which group it was placed in.  A section is comprised of several units, and each unit is comprised of one or more building blocks. (If there were very few (< 10) applications in a one or more building blocks, these would be grouped together into a unit.) 

(NSERC did acknowledge that communicating the new structure is turning out to be challenging.)

As examples, if GSC 14 became a section in a "Math/Stats Group", then there would be two sections for math, corresponding to old GSCs 336,337, and one section for statistics, old GSC 14.  There would be crossover in the areas of finance, probability, actuarial science, and possibly one or two others.  If GSC 14 became a section in a "CS Group", then there would be three sections for computer science, one for statistics, with crossover in statistical computing, computational statistics, machine learning, etc. 

Applicants would identify 2 or 3 units for their application, and section chairs would review the full set of applications to that section to divide the workload appropriately.  The applications would be rated  within the units, and these ratings would not be changed by the Section Chairs and Group Chair.  The budget would need to be balanced across the entire group.  NSERC is working on mechanisms to ensure consistency of ratings within and between groups and sections. 

One of our most important tasks in the immediate future is to suggest both a coherent set of building blocks/units, and a means to group all our building blocks/units so that we optimize the correspondence between applications to that group and expertise in that group.  We can make that case that the best such grouping is CS, or Math, or something else, but NSERC's dominant concern is to have applications reviewed fairly.