Matt McCormick has published a piece titled "Yes, Code Review Is Still Necessary" arguing that despite the rise of AI-powered coding tools, human code review remains a critical practice for software teams. The article appeared on Hacker News on August 15, where it gathered modest attention with a score of just 2 points and one comment at time of writing.

The Case Against Skipping Reviews

McCormick's central argument centers on the unique value human reviewers bring to the development process. While AI tools excel at catching syntax errors and following coding conventions, they often struggle with context-specific issues like business logic flaws, edge cases that could cause production incidents, and architectural decisions that might create technical debt down the line.

Knowledge Transfer in Distributed Teams

Beyond bug detection, code review serves as a mechanism for knowledge sharing across engineering organizations. When senior developers leave comments explaining why certain approaches were chosen, junior engineers absorb institutional knowledge that would otherwise be lost. This mentorship function is something no AI tool currently replicates effectively, according to McCormick.

Balancing Speed and Quality

The article acknowledges the tension between review thoroughness and shipping velocity. Teams under pressure to deliver quickly may view code review as a bottleneck rather than a safeguard. McCormick suggests pragmatic approaches like focusing reviews on high-risk changes while allowing low-stakes patches to ship with lighter touch.

AI as Reviewer, Not Replacement

McCormick doesn't dismiss AI entirelyβ€”rather, he positions it as a complement to human review. Automated tools can handle style checking and basic pattern matching, freeing humans to focus on higher-order concerns like maintainability and alignment with product goals.

Key Takeaways

  • AI coding assistants don't eliminate the need for human oversight of complex systems
  • Code reviews serve educational purposes beyond quality assurance
  • Teams should calibrate review depth based on change risk rather than applying uniform rigor
  • The best workflow combines automated checks with targeted human attention

The Bottom Line

The low engagement on McCormick's piece is tellingβ€”maybe the dev community has already settled this debate in practice. If your team ships without code review, you're either moving faster than wisdom allows or you've found a risk profile that genuinely doesn't need it. Most shops haven't.