Many universities do this, but usually only after the first pass of applications is removed. Also, even though 5% of the applicants are admitted, a much larger

Many universities do this, but usually only after the first pass of applications is removed. Also, even though 5% of the applicants are admitted, a much larger fraction of the applicants are completely qualified (i.e., if you selected a group of applications 3x the size you can admit, and then randomly selected 1 in 3 of those, you'd hardly be able to tell). So at that stage, multiple reviewers would help but not fix the randomness problem.