The article argues that ticket-based tracking systems have fundamentally undermined agile software development by replacing collaborative teamwork with individual task assignment. Originally, agile emphasized early delivery and frequent feedback through lightweight user stories on physical cards, but commercial tools rebranded tickets as agile solutions, shifting focus to solo work, detailed specifications, and utilization metrics rather than team collaboration and continuous learning.
A Hacker News user seeks recommendations for trusted resources and authors to learn best practices and methodologies in AI-assisted software development, citing information overload and mentioning Simon Willison as an example of the type of expert source they're looking for.
An article argues that AI accelerates software development but not deployment. While building products with AI tools like Claude and Devin produces fast demos, getting them production-ready requires expensive security, reliability, and privacy work that takes months. Research shows AI increases code volume and security risks, slows actual delivery despite feeling faster, and has delivered tangible value to only 4% of companies adopting it.
Tehas is a software factory powered by Pi coding agents that automates GitHub issue processing and pull request reviews using self-hosted models in a sandbox environment, with a dashboard tracking workflow stages from ticket intake through deployment.
A hiring manager argues that home assignments remain valuable in engineering interviews despite AI coding assistants. Rather than eliminating assignments, interviewers should use them as conversation starters to assess candidates' understanding of design trade-offs, architectural decisions, and ability to work effectively with AI tools.
The article traces how programming languages designed to eliminate programmers (COBOL, SQL, 4GL) instead created new specialized professions. It argues that despite recurring industry promises that natural language will replace coding, more programmers emerge under different titles, and programming languages serve fundamentally different purposes than human language.
A programmer reflects on how AI tools like Claude have fundamentally changed software development, enabling faster project creation with less hand-written code and AI-assisted code review, while raising questions about what programming education should emphasize as the field shifts toward prompt engineering and architectural direction rather than syntax production.
PostgreSQL 19 experienced an unusually high number of feature reverts during beta—approximately 8 to 10 significant ones—due to a combination of factors: insufficient development time, architectural defects discovered through improved LLM-assisted code review, and reduced senior developer availability during the security vulnerability response period. The author suggests most issues were addressable with additional time and focus, though the community prioritized maintaining the release schedule.
As AI agents increasingly automate software development tasks like code reviews and end-to-end product builds, the nature of programming work shifts from creative problem-solving to steering and verification. While this improves productivity and output speed, it removes the engaging aspects of development that many programmers find rewarding, similar to how assembly lines transformed craft-based car manufacturing into specialized industrial labor.
A keynote exploring how software developers understand code through the philosophical concept of verum factum—the idea that we truly know what we make. The speaker argues that software development is a 'symmathesy,' a learning system where developers, code, tools, and users learn from each other, and that this understanding deepens through collaboration, better tooling, and expressive code rather than external optimization.
A researcher gave four coding agents ($100 budgets) to build PDF editors and found significant usability issues in all implementations, attributing this to agents lacking human-like goal pursuit and UI interaction patterns. The experiment reveals that while LLMs can now produce functional code, human value in software development centers on maintaining focus and identifying friction points that agents systematically miss.
A developer argues that solving real problems remains the most rewarding aspect of software development, even when coding agents generate much of the code. While some developers miss the satisfaction of hand-writing code, the author contends that faster implementation enables more time for refinement and understanding user needs through iteration.
A solo developer rebuilt Adobe's Creative Suite in Rust using Claude, creating open-source alternatives like PhotoCraft and LightCraft with AI integration and agent-native capabilities. The project has sparked debate over legal concerns, code safety, and whether AI-generated software can replace Adobe's industry-standard tools.