In discussing the importance of computer games for conveying serious messages through play, Aleks Krotoski uses “play” and “games” interchangeably (Nature 466, 695; 2010). However, this is incorrect in the context of human development: these terms denote separate constructs, with different ontogenies, proximal causes and functions.

Play is mainly a behaviour of juveniles and is not functional in its immediate environment; its benefits relate to creativity and novelty. Games are developed later in childhood and are governed by rules based on deduction; their benefits tend to be specific to the game's dimensions (such as hand–eye coordination). What Krotoski describes are games, not play: interaction with an end in mind.

This conflation is potentially damaging. Science can progress only if constructs are clearly delineated and used consistently; where there is deviation, definitions should be explicit. Inexact usage may also mislead policy-makers: take Krotoski's claim that “computer-based play can support learning in schools”. To my knowledge, there is very little evidence that play, as defined here, affects school-based learning; where it does, it is limited to preschool children. But there is good evidence that games help school-age children.