Attempted to complete the test about 25 times, all without success like almost everyone else on this thread, and I also probably tried all the suggestions proposed here and elsewhere.
Then, as some other did in this forum, I cheated the test by clicking at what I judged to be the correct timing, no matter whether I heard the tones or not. Something interesting happened: the test succeeded for both ears, AND completed in about A THIRD of the time it normally takes when clicking without cheating!
Although the results were obviously completely invalid, stating that I had no loss of hearing either side, which I know for a fact to be wrong, it brings to the front possible bugs with the test software itself: If the test uses the results in real time to make decisions how to proceed, there might be a point where the test gets confused in the sequence/branching and consequently reaches the wrong conclusion, but this is just my 2 cents...
In addition, in makes you wonder if only people with no hearing loss can complete the test successfully most of the time, which kind of defeats the purpose. But again, these are speculations.