Daily Digest
Why TDD and AI coding tools exploit the same psychological flaws
Drew DeVault draws a provocative parallel between Test-Driven Development cults and GenAI adoption, arguing both exploit developers' psychological need to feel competent while potentially undermining actual code quality.
- TDD's hidden influence on architecture: While ensuring test coverage, TDD shapes codebases to be "testable" rather than well-designed, and provides no guarantee that passing tests verify the right behavior for actual user needs.
- The dopamine trap of development metrics: Both TDD's green test suites and AI's rapid output create addictive feedback loops through coverage percentages, CI badges, and productivity metrics that make developers feel competent regardless of actual code quality.
- AI coding agents as the new performance theater: GenAI tools let mediocre programmers experience the rush of 10x developer productivity, building "cathedrals" quickly that have beautiful test coverage but rotten foundations underneath.
- The psychological cost of shortcuts: Developers chase these tools despite knowing the externalities (environmental costs, job displacement) because the feeling of finally being "great" at programming overrides long-term concerns.
Latest Articles - Page 18
Memory Safety Is ...
Memory safety is a property of implementation, not program execution; most definitions incorrectly focus on the wrong aspect of the system.
The Second Great Error Model Convergence
Modern languages (Go, Rust, Swift, Zig) have converged on a new error handling approach distinct from traditional exceptions, requiring call-site annotations.
Git's HTTP server side design does not scale
Git's HTTP server uses CGI, which doesn't scale well for modern web applications due to performance limitations.
Shipping at Inference-Speed
Developer shifted from reading code line-by-line to streaming AI-generated code, working at the speed of AI inference rather than human comprehension.
Parsing Advances
Describes building a resilient parser that collects all errors instead of stopping at the first one, with a warning about infinite loop bugs from not consuming tokens.
On Friday Deploys: Sometimes that Puppy Needs Murdering
The post humorously discusses the perennial debate about whether deploy freezes during holidays help or hurt software development teams.
Arcane Cheese with Doomtrain Extreme
A humorous Final Fantasy 14 blog post challenges non-players to spot which gaming terms also appear in the Bible.
Wrapping up Blogvent 2025
Blogvent 2025 has concluded. Check out all the December posts from this year's series!
Unix "find" expressions compiled to bytecode
Author developed a bytecode compiler for Unix find expressions, which surprisingly performs better than the tree-walk interpreters used in real implementations.
Advent of Slop: A Guest Post by Claude
Claude AI autonomously solved Advent of Code 2025 puzzles using a web browser skill, working independently to read prompts and submit solutions.
Newtype Index Pattern In Zig
Using 32-bit indexes instead of pointers saves memory, improves CPU cache efficiency, and boosts performance by reducing memory bandwidth bottlenecks.
Making the "End of Year Developer" nature documentary
A behind-the-scenes look at creating a nature documentary-style film about developers "working" during the holiday season.
2025 was for AI what 2010 was for cloud
AI has become mainstream foundational technology in 2025, similar to how cloud computing transitioned from experimental to essential in 2010.
A Year Of Vibes
The author reflects on 2025 as a transformative year, leaving Sentry to start a company and shifting from hands-on coding to AI-assisted programming.
CSS for markdown blockquote attribution
Markdown-generated HTML blockquotes need proper CSS styling to look good.
Generating data shapes with Hypothesis
Hypothesis can randomly generate schemas (not just data) to test a Hasher class, avoiding false positives from type mismatches like [0] vs [False].
I like when apps are "finished"
Apps don't always need updates—sometimes software is just good enough as is.
Reflections on AI at the end of 2025
AI researchers now accept LLMs aren't just "stochastic parrots." Chain-of-thought reasoning improves outputs through internal search and RL.
I have become more of a writing utensil snob
Writer admits to being a pen/pencil snob with too many writing utensils, but insists they have it under control.