It seemas that Cornell's (1968) approach to PSHA started as an open research basis for further development in many directions. After McGuire's publication of the FORTRAN program to assess PSHA, everything fell into a rigid formalism leaving no ground for extreme value analysis, analytical approximations that facilitates heuristic guidance. Also, it became almost compulsory to integrate over aleatory uncertainties instead of treating them separately.

Can't we break away (especially in research) from the methodological "Cornel-McGuire" procedure? It is a perfect a approach for a perfect world (total probability theorem without any questions), but there are many examples of cases that need exploration along other routes?