The basic tests that run using TEST aren't the same as when using ATTEMPT (there are two extra tests before the random tests). If that last test (which is tripping so many people up) was included in the basic tests that a user gets when clicking TEST, then this would be a much better kata.
Reasoning: If a user fails on one of the basic tests after passing all the basic tests with the same code, it's understandable to assume that the tests are broken (even though they aren't).