I don’t think I understand what is being said here. There is some story about refactoring but then failure, but I’m not sure what that failure was. Seems like the point of article is you should have tests but I’m not really sure how the author is motivating it in this post.
I think the thrust of the post is that if you find yourself focusing on the refactor before you finish building out tests to confirm the system’s current behavior, you can end up in a bit of a pit of confusion. Nail down your current system with tests, then refactor away. Know what there is and then spend time thinking about what the system should be.