The goal of the work presented here is to better understand the affordances and limitations of using blocks-based environments in formal computer science contexts, and then apply that knowledge to the design a new tool for learning to program that draws on the strengths of both the blocks-based and text-based modalities. To answer these questions we conducted a study comparing three versions of a blocks-based programming tool in high school introductory programming classes. In doing so, we seek to provide evidence to better inform educators who are tasked with making consequential decisions around how learners are introduced to computer science and, more generally, to contribute to our understanding of the relationship between programming tools and the understandings and practices they promote.