July 11, 2026
AI wrote it. Humans panicked
Why Write Code in 2026
Coders revolt as AI hype meets a simple comeback: sometimes you still need to touch the stove
TLDR: The article argues that even in an AI-heavy future, people still need to write some code themselves to truly understand and improve what they’re building. Commenters split between proud hand-coders, pragmatic AI users, and jokers comparing the whole thing to choosing home cooking over fast food.
The big argument here is deliciously simple: if artificial intelligence can churn out most of your software, why are humans still typing code at all? The writer’s answer is basically, “because you need to feel the mess.” He says software builders aren’t just making apps anymore — they’re building the whole machine that lets AI helpers make changes fast. But if humans only sit back and approve whatever the machine spits out, they stop noticing when the whole thing gets shaky, confusing, or quietly terrible.
And the comments? Oh, they were very awake. One camp went full old-school pride: I still write my own code, it’s better, cleaner, and I actually know what’s going on, declared one commenter in the digital equivalent of slamming car keys on the table. Another took a more balanced, less dramatic line: humans should craft the important parts and let AI handle the boring filler — what one person hilariously called the “little islands” of judgment connected by “obvious goo.” That line alone deserves its own merch.
Then came the meme brigade. The funniest clapback compared the whole debate to asking, “Why cook food in 2026 while McDonald’s exists?” Brutal. Another commenter zeroed in on the article’s core fear: fragility. In plain English, if software feels flimsy and weird when a person touches it, letting a robot keep piling onto it is probably not going to end well. The community mood was clear: AI can help, but nobody’s ready to hand over the kitchen and call it dinner.
Key Points
- •The article says software engineers increasingly build the infrastructure and guardrails that let AI agents make and ship code changes.
- •The infrastructure described includes proactive guidance such as prompts, skills, AGENTS.md files, and knowledge bases, plus reactive checks such as tests, linting, type systems, and evaluations.
- •The author argues that writing code remains useful because it helps humans think directly in the execution environment and maintain attention on system architecture.
- •The article links hands-on coding with detecting fragility, improving documentation and architectural consistency, and strengthening testing strategies.
- •The article argues that coding agents should be viewed more like newly onboarded interns than compilers, and that human involvement is still needed for judgment and ownership.