For all practical purposes the basic physical equations governing the behaviour of a system at the molecular level can only be solved approximately. The key issue in any reliable and accurate computational study in molecular physics and quantum chemistry is the adoption of a suitable model which contains the essential physics and chemistry, is computationally tractable, and preferably amenable to systematic refinement. The provision of advice on the choice of an appropriate model for a specific problem has so far received scant attention. This issue is becoming acute as standard' software...