Sure and looking at paper results seem to show that versions without it seem to have been doing better in major events than versions with it on the whole, but that's not analytical at all. Relatively small samples, no underlying data and all that, and you'd have to break down how many of each...