As K–12 computer science expands in the United States, students encounter a growing array of programming tools. Many introductory experiences use block-based environments, where programs are assembled by snapping together visual blocks instead of typing code. While these tools can support learning, high school students often perceive them negatively, even when they support the same underlying logic as text-based coding. Using a constructivist grounded theory approach, we interviewed 17 high school students to trace how early experiences, tool design, peer discourse, and cultural framings shape these views. We find that students develop informal folk theories: that computer science is about accumulating languages, that block-based programming is for young children, and that limitations in programming activities stem from the block modality itself—beliefs that can shift when students encounter counterexamples. Our findings call for more deliberate design and sequencing of tools that are attentive to the meanings students construct as they progress, and that promote more expansive notions of programming beyond modality.

Authentic learning, characterized by engagement with real-world problems and tools, has long been of interest in education due to its impact on student motivation and learning outcomes [2, 7]. In computer science (CS) education, however, students and teachers face the challenge of balancing the desire to teach and learn "real" programming with the need for a gentle and scaffolded introduction to this highly abstract and cognitively demanding discipline [4]. As a tool-dependent discipline, the tension between authentic and scaffolded is particularly evident in the perceived in-authenticity of educational programming tools. While scaffolded blocks-based programming tools are approachable [14] and beneficial for learning [3, 10], they are often perceived as less authentic by high school students [4, 14], which can be demotivating. Conversely, "real" text-based programming, while authentic, can be difficult and intimidating, creating a barrier to learning and engagement [10, 14]. This dichotomy exemplifies a challenge in CS education: how can we provide an authentic learning experience through tools that are both approachable and representative of authentic programming practice?

Modeling complex dynamical systems poses significant challenges, with traditional methods struggling to work on a variety of systems and scale to high-dimensional dynamics. In response, we present DynaDojo, a novel benchmarking platform designed for data-driven dynamical system identification. DynaDojo provides diagnostics on three ways an algorithm’s performance scales: across the number of training samples, the complexity of a dynamical system, and a target error to achieve. Furthermore, DynaDojo enables studying out-of-distribution generalization (by providing unique test conditions for each system) and active learning (by supporting closed-loop control). Through its user-friendly and easily extensible API, DynaDojo accommodates a wide range of user-defined \texttt{Algorithms}, \texttt{Systems}, and \texttt{Challenges} (evaluation metrics). The platform also prioritizes resource-efficient training with parallel processing strategies for running on a cluster. To showcase its utility, in DynaDojo 0.9, we include implementations of 7 baseline algorithms and 20 dynamical systems, along with several demos exhibiting insights researchers can glean using our platform. This work aspires to make DynaDojo a unifying benchmarking platform for system identification, paralleling the role of OpenAI’s Gym in reinforcement learning.