Just recognize who ran the test in question. A company sponsored DBT never produces a winner that isn't their own product. With that logic, if you would have given the same speakers to ML to run the test, I suspect the results would be much different ;)
Recommended reading...