Did you read the paper? It is quite long, mainly because of Kramnik's manual analysis of some selected games. It appears his role was to explore game playing factors, which are hard to catch with statistics.