Fetching from the wire…
Top 5 · 2026-04-13 · source-backed
Oxide Computer CTO Bryan Cantrill published "The Peril of Laziness Lost" on April 12. His thesis: the best software exists because humans are lazy, and LLMs aren't.
The argument goes like this. Larry Wall identified laziness as a programmer virtue decades ago. Not sloth. The kind of laziness that makes you write a script instead of doing repetitive work manually. The kind that makes you create a clean abstraction because you don't want to maintain the same logic in five places. Human laziness is a constraint that forces crisp design.
LLMs have no such constraint. Work costs them nothing. They'll happily generate 500 lines where 50 would do. They'll duplicate logic across files because deduplication requires caring about maintenance burden. Cantrill calls the result "a layercake of garbage," systems that grow larger but not better.
Simon Willison amplified the piece the same day, and it's been bouncing around my feeds since. The timing is perfect. Apple is seeing an 84% quarterly surge in App Store submissions driven by vibe coding tools, pushing the 2025 total to 557,000 new apps. The Harvard Gazette reports 92% of US developers have adopted vibe coding practices and 60% of new code in 2026 is AI-generated.
I use Claude Code every day. I'm deeply in the camp of "AI makes me dramatically more productive." But Cantrill is identifying a real failure mode that I've seen in my own work. When I let the agent run without tight feedback loops, systems get bigger without getting simpler. The harness engineering story above connects directly: guides and sensors are how you inject human laziness back into the process. The CLAUDE.md that says "prefer editing existing files to creating new ones" is literally encoding the laziness virtue.
The practical takeaway for builders: your job isn't to generate code anymore. It's to maintain the design pressure that keeps systems simple. Review every agent-generated PR for abstraction quality, not just correctness. Does this code need to exist? Could it be simpler? Would a lazy human have done it differently? That's the question.
Each link below shares sources, entities, or timing with this story.
Alibaba released Qwen3.6-27B on April 22. Dense architecture. Open weights. 77.2% on SWE-bench Verified, within 3.7 points of Claude Opus 4.6. On SkillsBench, it scores 48.2% versus its own 397B MoE predecessor's 30.0%. That's a 77% improvement with 14.8x fewer parameters. Let...
Roo Code announced it will archive its VS Code extension repo on May 15 and merge back into Cline, the project it originally forked from. CEO Matt Rubens said the team needs to "constantly destroy and recreate to keep up with what's newly possible." Translation: the extension...
Spotify's Portal team published Xirp on August 10: a vendor-neutral agentic development environment that manages concurrent sessions across Claude Code, Gemini CLI, and Codex, each session isolated in its own git worktree so dozens of agents can work the same codebase without...
BitTorrent creator Bram Cohen published "The Cult Of Vibe Coding Is Insane" and it hit 569 points with 463 comments on Hacker News. The most-discussed vibe coding essay to date. Cohen's trigger was the Claude Code source leak. When the 512,000 lines of TypeScript leaked via a...
Starlette 1.0.0 shipped March 22, reaching stability after 8 years at 325M downloads/month as FastAPI's foundation. Simon Willison immediately identified the problem: LLMs are trained on pre-1.0 code and will generate incompatible patterns. He built a custom Claude skill from...
Uber's COO Andrew Macdonald told Business Insider what a lot of engineering leaders are thinking but won't say publicly: "Getting harder to justify money spent on tokenmaxxing." The backstory: Uber's CTO revealed the company burned through its entire 2026 Claude Code budget by...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.