Ullman's argumentation is very wiggly when it comes to impact on performance. On one hand, he claims there's very little proof, on the other hand he claims there is no point producing proof because they wouldn't be believed. Furthermore he says their hands are tied as they'd need their client's approval to publish such numbers, conveniently ignoring the possibility of anonymizing and grouping data. Remember they state to being included in 70-80 games yearly.

Finally, and this is an old song, but anyways: preventing loss in sales due to piracy is an exercise in wishful thinking. The amount of pirate players who would have paid in an alternate universe with no cracked version can't be reliably estimated.