Tag
A software engineer reflects on designing a 54-card deck (Algodeck) that distills computer science concepts into accessible illustrations, and shares insights on how graphic design and software engineering both solve problems through systems thinking.
Gergely Orosz emphasizes the importance of using UTC timestamps or epoch time along with proper date libraries to handle time zones and daylight saving time correctly.
This article explores how domain-specific languages (DSLs) improve the reliability and design insight when using LLMs for code generation, emphasizing iterative design and constrained syntax.
This paper introduces Agentic Service-Oriented Computing (ASOC) as a new research area, proposing six foundational principles and a five-dimensional research agenda for engineering LLM-powered agents as dependable services in distributed ecosystems.
A tweet expressing fear over a perceived major disruption to software engineering caused by an AI advancement.
The article argues that as AI coding models become superhuman, the engineering bottleneck shifts from writing code to encoding judgment, leading to flatter software organizations where customer insight and product judgment become the primary human advantages.
A veteran software engineer reflects on AI's disruptive potential across professions including bioengineering, law, and accounting, expressing concern about the accelerating changes.
A blog post defending the necessity of keys and normalization in the relational model, arguing that they reflect ontological requirements for coherent discourse about reality, countering criticisms about practical difficulties in defining keys and domains.
Armin Ronacher reflects on how the shared language of a software project — the common understanding of concepts, boundaries, and invariants — lives in documentation, code, and conversations, and how the friction of coordination synchronizes people, a process that AI agents might disrupt.
The essay uses the Tower of Babel story to argue that AI-assisted programming removes friction in software development, which may undermine the shared understanding and coordination necessary for large projects, leading to chaotic codebases.
A commentary on how software engineers should recognize when they are engaging in 'vibe coding' (AI-assisted coding) and that there is jagged expertise, similar to models, in development practices.
This paper investigates how language models under coding agents internally represent the evolving program during iterative editing. The authors find that linear probes can decode program properties (e.g., parsing, test pass rate) from residual streams, and surprisingly, these representations predict future outcomes before the agent materializes edits, revealing a 'latent programming horizon'.
An exploration of the Forward Deployed Engineer (FDE) role, its origins at Palantir, adoption by AI startups like OpenAI and Ramp, and reasons for its growing demand in 2025.
Discusses the debate on whether libraries should log errors themselves or propagate them to the calling application, highlighting differing practices across ecosystems like Go's slog.
A tech commentator notes that success stories about rewrites and migrations published by companies are half-truths, omitting negatives and internal incentives, and warns that copying them without understanding context leads to disappointment.
The article critiques Anthropic's narrative that AI will replace software engineering, discusses the controversy surrounding Bun's rewrite from Zig to Rust, and praises Zig creator Andrew Kelley's blunt response to the migration's justification.
The article explores loop engineering as a potentially missing discipline for developing robust production agents in AI systems.
This paper formalizes the 'patchwork problem' where LLM-generated code is locally correct but structurally incoherent across a codebase, proposes a taxonomy of eight failure categories and a hybrid verification framework, and demonstrates that many failures evade current tools.
Amy J. Ko shares her critical perspective on AI-assisted software development, exploring the promises and pitfalls of using LLMs for coding, and recounts her three-month experiment to evaluate the real impact on productivity and practice.
Gergely Orosz observes that the discussion about AI enabling mainstream formal verification has faded, and questions why AI hasn't impacted that field.