> I highly doubt you are getting novel, high quality data.

That's not the point. The point is you reject low quality data, aka noise