On today's episode of "Loud in tech," we dive into the latest developments in the tech world. We start by exploring a new interpreter technique called "flattening," which promises improved performance in interpreting expressions. Then, we discuss Unison, a language that introduces code-level dependencies as a potential future solution for versioning issues. Moving on, we highlight the repository "a1k0n/a1gpt," a throwaway GPT inference engine that utilizes optimized BLAS operations for AVX and Apple Silicon. In a surprising turn, we take a look at "git-landmine," a repository that allows users to create a local malicious git repo, potentially shedding light on security concerns. Shifting gears, we examine the changing attitudes towards psychedelic drugs and the need for a new classification system. Additionally, we delve into GPU-accelerated SMT constraint solving and the remarkable results achieved by the research team at NYU Tandon. We also touch on the importance of writing as a form of thinking, exploring the influence of large language models like ChatGPT. Finally, we explore the repository "gpt-migrate," a tool designed to assist developers in migrating their codebase. Stay tuned for a thought-provoking discussion on software team performance measurement, CPU cache myths, and more.