Anti-pattern covered in Rodzik's talk. Tests that contain conditionals (`if p1_score > p2_score then expect ping else expect miri`), loops iterating over 100 products, or calls to production helpers like `cents_to_dollars` are harder to read and often fail to actually test anything: the test's logic mirrors the production logic, so a bug in one passes the other. Symptom: when a loop-driven test fails on 'product 101', the failure message doesn't tell you which product. Rule of thumb: tests should be immediately understandable and should not duplicate production logic.