Computational notebooks are popular tools for data science and crafting computational narratives. However, their 1D structure introduces and exacerbates user issues, such as messiness, tedious navigation, inefficient use of large displays, performance of non-linear analyses, and presentation of non-linear narratives. In this Ph.D., we address these issues through the design, exploration, and evaluation of computational notebooks which use 2D space to organize cells, or 2D computational notebooks. Specifically, we explore whether users would use 2D space, design and evaluate a 2D computational notebook prototype for individual work, explore how users collaborate in 2D space for data science and education, create and validate a theoretical understanding of how nonlinear processes in data science cause problems when forced into a linear, 1D computational notebook, and build upon the foundation we have made to refine 2D computational notebooks. Our work contributes insights on if and how expanded space usage can improve computational notebooks.