I would like to consider the benchmark domains used at the recent AIPS-98 and 2000 planning competitions and discuss the properties of these problems. I will consider three in detail: blocks world, briefcase and logistics. I will outline some of the algorithms used to attack these problems such as FF and Blackbox. I will then argue that the competition misses two important points about real world planning: the domains are much more knowledge-rich than these and we are usually not interested in "any plan" but want the best one. I will present Pacifica, as used in O-Plan and other research, as a potential benchmarking domain.