To start a new GCVS edition (presumably an electronic one), combining the stars of the GCVS IV and Name Lists (and maybe eventually merged with automatic-survey variables), we need to revise the GCVS classification system. We get conflicting signals from the community concerning the directions of this revision: • Experts in particular variable-star classes ask to introduce many new types reflecting new knowledge; • people interested in stellar variability in general tell us we have too many types, different to understand and keep in memory. Our approach: the classification scheme must be simplified, more detailed classification should be given in specialized catalogs for particular types, but the GCVS classification should take into account new astrophysics.