Senior dev with 30 years argues bootcamps teach syntax but not the durable skills: system decomposition, data flow, and knowing which decisions are load-bearing.
Key Takeaways
The real bottleneck is never typing speed or syntax recall – it is knowing what code should be written, in what order, with which contracts between parts.
Durable programming skills include: mental models for decomposition, data-flow intuition, spotting cheap-vs-expensive decisions, and debugging as a discipline.
AI coding tools (Claude Code cited directly) amplify seniors who already know where the seams go; they accelerate plausible-but-wrong output for juniors who lack that judgment.
Recommended path: go deep in one language by shipping and maintaining a real thing for a year, then pick a structurally different second language and find the overlap – that overlap is programming.
Pair-programming with someone 15 years ahead is called the highest-bandwidth learning available, yet almost no one is putting it on YouTube because there is no financial incentive to do so.
Hacker News Comment Review
Commenters broadly agree with the thesis but push back on curriculum as the fix – the consensus is that these skills are acquired on the job through mistakes, not taught in structured settings.
A practical objection surfaced: most hiring pipelines screen for a specific language anyway, so learners rationally optimize for syntax over systems thinking regardless of what is ideal.
One thread debated whether language must come first – the counterpoint being that experienced developers reason about unfamiliar code across the C/Java/Go family without deep knowledge of each.
Notable Comments
@hackthemack: most companies hire for language X and do not care if you could pick it up in weeks – market incentives undercut the article’s advice directly.
@somewhatgoated: “Who still does boot camps in $currentyear? You can get all these fundamentals for free…from an LLM” – ironic given the article’s AI-amplifier warning.
