User-Agent information is inherently unreliable. I don't see why people are upset when a misguided practice that already provides false results with an unknown margin of error gets even worse. If they insist on using such a metric then it's their responsibility to assume the risks that it incurs.