Don't you mean extinct?
Points and comments are a snapshot, not live.
Phil Tippett's 'extinct' moment mirrors programmer anxiety about LLMs, urging adaptation.
Fabien Sanglard parallels Phil Tippett's 1993 fear of obsolescence when CGI replaced stop-motion to current programmer fears about LLMs. Tippett adapted by co-developing the Dinosaur Input Device and won an Oscar for Jurassic Park. Sanglard advises programmers to evolve by learning LLM internals via Andrej Karpathy's videos and Sebastian Raschka's book, and using LLMs for code generation while maintaining quality through style configs (`GEMINI.md`/`CLAUDE.md`) and thorough code review (commit messages, test coverage). Smaller teams can now revive abandoned projects; LLMs help read research papers and codebases.
Sanglard notes that while LLMs produce 1000x output, code quality still matters: avoid magic numbers, reduce indentation, respect layering. He cautions about mental burnout from context switching across agents. The post cites John Carmack: 'Coding was never the source of value. Problem solving is the core skill.'
What commenters are saying
Commenters split sharply. The dominant camp criticized the article's claim that 'those who refuse to use an LLM will fall behind because they won't be able to produce as much,' calling it needlessly threatening. Several argued volume was never the metric; quality, maintainability, and domain understanding matter more.
A counter-camp shared concrete gains: one used LLMs to expand test suites and tackle reproducible builds on MSVC; another built a single-use interactive web app to evaluate coworker data in days instead of weeks; a third debugged a legacy 8051 codebase with Fable, fixing years-old bugs. One commenter noted managers at some companies have doubled velocity expectations, mandating Claude use-but added legacy codebases often resist LLM assistance.