The Author's Argument
Python conquered the world by being fast to write, smooth to read, and low-barrier to pick up—all features optimized for "humans writing code," at the cost of mediocre performance, loose typing, and frequent runtime errors. The author's argument is that when the main force writing code becomes AI, the scales tip: generating Rust versus generating Python makes little difference to a model, and strong typing and compile-time checks are precisely powerful tools for constraining the quality of AI output. The human role shifts from writing to reviewing, and when reviewing, code the compiler has already checked over is obviously more reassuring.
The Counterarguments
The pushback in the comments is solid too. The most forceful point: code is written once and read a hundred times, and as long as someone still has to read it, change it, and take the blame, the readability premium won't disappear—plus AI has the most training data on Python, so its generation quality is actually more stable. Others point out that the ecosystem is the real moat—the libraries for scientific computing and machine learning are all on the Python side, and language selection has never been about the language alone. The value of this debate is in posing a real question: should a programming language's design goals now rank "easy for AI to generate, easy for humans to review" near the top?
via: Hacker News