Lauren Tan says adversarial code reviews waste tokens compared with empirical testing

Lauren Tan (@poteto) says adversarial code reviews are mostly a waste of tokens and that those tokens are better spent on empirical testing combined with formal verification.

In a post dated November 2, 2026, she says she has reduced use of her pstack interrogate skill and may remove it entirely.

Tan says abstract code review leads agents to get lost in nitpicks and YAGNI concerns, which she reports gives little value. Her claims are based on her own experience and are not supported by measured results in the source.