it's like anything else, when one single testing method becomes the reference, manufacturers become interested in doing good in it. DXO, while still a silly half done software(who does post processing without giving mask or brush to the user???? they don't get my money anymore), it has become the reference for objective data on lenses and cameras. which will obviously leads to choices like "do we care for perceived image quality or about passing a measurement?"
and I'm also of the opinion that we're dumb and for the most part don't change our behavior when it's a gentle sodomy. Nvdia &co did it so many times, browsers, cars... at the time they sell a lot of products, and in the end they only need to get away with it long enough for the next product to come around with the next marketing trick. plus once people get some false information like that, most of the time rebuttals don't get the same coverage nor the same subjective impact. if something as huge as what happened with Volkswagen can end up with them making profits, it's fair to say that cheating is a good idea.
