Martin Fowler's periodic digest covers Thoughtworks' 34th Technology Radar, which is heavily AI-focused and revisits foundational practices like pair programming, mutation testing, and clean code as counterweights to AI-generated complexity. A key theme is securing 'permission hungry' AI agents that require broad access but lack adequate safeguards. Also discussed: Mike Mason's experience with Claude Code producing large, unwieldy codebases and a framework for when human code review is essential. Additional links touch on LLM ghost writing philosophy and the dismantling of the IRS DirectFile program, with reflections on government technology initiatives and public service values.
Sort: