The recently released CS2023 Curriculum Guidelines propose a "CS Core" of topics that any student with a bachelor's degree in Computer Science must know. While the process of generating CS2023 included multiple rounds of feedback from the CS community, the end result is necessarily the result of compromise and may not meet the needs of all institutions. In particular, the size of the CS Core, as measured in instructional time or courses, may not be consistent with all programs. Small programs with limited faculty, liberal arts programs wishing to ensure space for academic exploration, or departments wanting to explore less traditional curricular structures may find a large CS Core limiting. The 2024 report "Piecing Together the Next 15 Years of Computing Education" edited by Decker and Weiss discusses the challenges of our increasingly "overstuffed curricula" and the barriers this can erect against broadening participation in computing.

ACM/IEEE curriculum guidelines for computer science, such as CS2013 or the forthcoming CS2023, provide well-researched and detailed guidance about the content and skills that make up an undergraduate computer science (CS) program. Liberal arts CS programs often struggle to apply these guidelines within their institutional context and goals. Historically, this has been addressed through the development of model CS curricula tailored for the liberal arts context. We take a different position: that no single model curriculum can apply across the wide range of liberal arts institutions. Instead, we argue that liberal arts CS educators need best practices for using guidelines such as CS2023 to inform curriculum design. These practices must acknowledge the opportunities and priorities of a liberal arts philosophy as well as a program's mission and identity. This paper reviews the context and motivation behind computing in the liberal arts. We also review the history of liberal arts CS educators and ACM/IEEE curriculum guidelines. We present data and trends about liberal arts computing programs, discussing how this informs curriculum design. Finally, we propose a process that guides programs to work with curriculum guidelines through the lens of institutional and program missions and identities, goals, and situational factors.

A liberal arts context offers unique opportunities for curricular innovation that can inform the implementation of computing curricula more broadly. The SIGCSE Committee on Computing Education in Liberal Arts Colleges has collected 18 model curricula during affiliated events at SIGCSE symposia over the past two years. Here we conduct a distillation of the curricula and discuss themes across the curricula including: flexible pathways through majors, interdisciplinary initiatives, and preparing students for a range of careers and their first job. Our discussion focuses on how liberal arts colleges are empowered to think creatively about computing curricula and how research intensive universities, community colleges, and K-12 can leverage these approaches for their context.

We present the algorithm visualization capabilities of the METAL project. Using METAL's graph data which represents highway systems, a selection of interactive algorithm visualizations are performed. Progress of the algorithm is shown by changing the colors of the graph's vertices and/or edges overlaid on Google Maps and in color-coded tabular form, including contents of important data structures. Advantages include the real-world data set and the variety of data sizes available, enhancing student engagement. While many visualizations and visualization tools exist for graph and related algorithms, most focus on small, synthetic graphs. We describe our algorithm visualization capabilities, which include implementations of sequential search, graph traversals, Dijkstra's algorithm, and convex hulls. These can be executed on graphs ranging in size from a few vertices and edges to hundreds. We also present results of a survey of students who have used METAL's algorithm visualizations.

Undergraduate computer science programs at many small colleges often include only one course focused on hardware. Many important concepts are covered in such a course, including the basics of computer architecture. By the end of such a course, students should have a good understanding of how a binary machine instruction is executed in hardware. Unfortunately, even a simplified diagram of a datapath is often difficult for students to master. We present two approaches that use lab exercises to help to address this problem. In one, students build a working model of the datapath out of ordinary materials; in the other, a software simulator is designed and implemented. These approaches are described and their merits discussed.

It is often a challenge to find interesting and appropriate data sets to use as examples to demonstrate graph data structures and algorithms. Goals for the data are often conflicting. The data should include examples small enough to work through in a class example by hand, but some large enough to demonstrate important behaviors of a structure or algorithm. Data should be freely available in a convenient format and should have some real-world relevance. Visualization of the data and results computed from it is helpful.