Our focus is on the effects that dated ideas about the nature of science (NOS) have on curriculum, instruction and assessments. First we examine historical developments in teaching about NOS, beginning with the seminal ideas of James Conant. Next we provide an overview of recent developments in philosophy and cognitive sciences that have shifted NOS characterizations away from general heuristic principles toward cognitive and social elements. Next, we analyze two alternative views regarding "explicitly teaching" NOS in pre-college programs. Version 1 is grounded in teachers presenting "Consensus-based Heuristic Principles" in science lessons and activities. Version 2 is grounded in learners experience of 'Building and Refining Model-Based Scientific Practices' in critique and communication enactments that occur in longer immersion units and learning progressions. We argue that Version 2 is to be preferred over Version 1 because it develops the critical epistemic cognitive and social practices that scientists and science learners use when (1) developing and evaluating scientific evidence, explanations and knowledge and (2) critiquing and communicating scientific ideas and information; thereby promoting science literacy.