“Code Was Never the Hard Part” Is an Insult to All Programmers

In a recent blog post, developer Senko Rašić takes direct aim at a common trope in the AI era: “Code was never the hard part.” The argument, often used by AI advocates, suggests that LLMs handle the coding easily, so the real challenge is deciding what to build. Rašić calls this “a gross insult to all programmers.”
He opens with a series of pointed questions: If coding is so easy, why were programmers in high demand and well paid before ZIRP? Why was there so much stress and burnout? Why did companies hunt for 10x ninja rockstars and subject candidates to leetcode hazing?
He also challenges the flip side—that “figuring out what to build” is the true hard part. If that were true, why are product managers often clueless, why aren’t they paid more than developers, and why don’t they face rigorous 10-step interviews? Why are business analysts looked down on as pencil pushers?
Rašić argues that “there’s no median programmer” when it comes to wanting to talk to stakeholders. Most developers he’s met have no desire to talk to customers—they want clear priorities and stable requirements. He points out that even developers who claim to be “customer problem solvers” often pivot to monads and memory safety, with only a shallow understanding of actual customers.
The post emphasizes that both code craftsmanship and understanding the customer are critical. “A deep understanding of the system we’re building, together with a deep understanding of why we’re building it,” he writes. He rejects the extreme positions on both sides: “Loudly proclaiming that ‘code is easy’ or, at the opposite end, ‘code is art, a creative human expression that cannot be automated,’ is just burying our heads in the sand.”
He closes with the Spanish phrase “¿Por qué no los dos?”—why not both? The goal is to pursue both the craft and the purpose. “You don’t want cope, you want to thrive,” he writes. But he stops short of endorsing an LLM bandwagon: “By this, I don’t mean ‘jump on the LLM bandwagon.’”
📖 Read the full source: HN AI Agents
👀 See Also

NVIDIA DGX Spark Community Launches Spark Arena for Reproducible LLM Benchmarks
The NVIDIA DGX Spark community has launched Spark Arena, a reproducible leaderboard for open-weights LLM performance using standardized tools and methodology, with current top performers including gpt-oss-120b and Qwen3-Coder-Next.

Anthropic Separates Programmatic Usage from Claude Subscriptions: New Credit Pool Arrives June 15
Starting June 15, Claude subscriptions get a dedicated monthly credit for programmatic use (Agent SDK, claude -p, Claude Code GitHub Actions). Interactive credits no longer subsidize programmatic calls; after the pool runs out, users pay full API rates.

Claude Opus 4.7 adds high-resolution image support, task budgets, and removes extended thinking
Claude Opus 4.7 introduces high-resolution image support up to 2576px/3.75MP, a new task budget feature for controlling token usage in agentic loops, and removes extended thinking budgets in favor of adaptive thinking.

Claude Code source code reportedly leaked, revealing agent architecture details
The source code for Claude Code, Anthropic's AI coding agent, appears to have been leaked, containing the full repository with system prompts, agent loop implementation, and tool calling infrastructure.