Background: Tinkering, an ad hoc approach to solving a problem, involves manipulating objects to characterize and build knowledge about a particular system in an exploratory way, often with the goal of getting some product/idea to produce desired behavior. Tinkering contrasts more deliberate activity toward understanding how a phenomenon works. Purpose: Some researchers have argued that tinkering is an unproductive engineering design practice because it does not always lead to progress in design and/or conceptual learning while others have argued for tinkering as a productive design practice. We speak to this disagreement regarding the productivity of tinkering for novice designers by closely analyzing tinkering episodes. Design/Method: We present a microgenetic account of two tinkering episodes to contribute to a more refined understanding of what tinkering is. The research setting is in a summer camp for high school girls using a design-based Arduino component. Results: Our analyses demonstrate that tinkering is neither universally productive nor unproductive in engineering design; we show that tinkering can help participants make progress toward some instructional goals such as supporting the engagement of students in some engineering design practices. We also show that tinkering can hinder progress toward other goals such as developing a robust solution to a design problem. Conclusions: We argue for a more nuanced understanding of productivity (of tinkering and other design practices), which is local and defined with respect to specific goals and actors.