Pencils Down
DHH says it's pencils down on writing code by hand. I think he's right about the typing. The real work is understanding systems and the people who use them.
Writing on software engineering, technology strategy, and where the industry is heading. Several of these started as predictions; the pieces check them against what actually happened.
13 entries · newest first
DHH says it's pencils down on writing code by hand. I think he's right about the typing. The real work is understanding systems and the people who use them.
A new infrastructure layer for AI agents is emerging. Sandboxing, micro-environments, tool registries, and orchestration primitives will make the underlying language irrelevant. The stack is being eaten from above.
Software development is splitting in two. A supervisor class will orchestrate AI and define outcomes. Everyone else gets compressed. The dividing line isn't experience — it's judgment.
AI is accelerating release cycles to dozens per week on major platforms. The gap between what software can do and what users have absorbed is becoming a problem of its own.
Quick take on why AI Full-Stack Developers (AIFSD) only work inside digital factories, not as lone freelancers.
Why most software systems have an effective shelf life of about five years, and what that means for how we build and plan.
The languages that win long-term aren't the most powerful — they're the most consistent. When code looks the same regardless of who wrote it, teams move faster.
The case for strategic rewrites over endless refactoring, and when starting fresh is the right call.
Jeff Bezos is back. The founder's return to Amazon's frontlines to drive AI strategy confirms a prediction made in 2023: this company is rebooting for the AI era.
The forces reshaping programming languages aren't about syntax preferences. Hardware shifts, safety mandates, and AI workloads are driving a generational transition in what languages get adopted.
Why multi-model databases are gaining traction and what it means for application architecture.
Full-stack was never about knowing every layer. It's about shipping a feature without handoffs, and AI is about to make that the whole job.
The 2010s SaaS land grab hit escape velocity. Now gravity—and finance—is calling the capsule home.