Comment by chemeng

The charitable reading is that this was done as a blinded study, so the testers/researchers would not know and would prevent bias in the testing and data analysis.

There should be a way to unblind in most experiment designs though