In the linked forum post, Jeremy adds a write skew test that not always detects a write skew, resulting in multiple counter increments for the same number (i.e. both threads increment the counter to a given number).