To be honest, this provides a great opportunity for research.
Anecdotally, we believe that we can differentiate the top 5% and the bottom 5% of applicants, but there is a large swath of people in the middle. I personally can attest to this from being an assessor for multiple MMIs and having trouble differentiating applicants. In truth, most of the people McMaster interviews would make great physicians. So maybe we do not need to separate applicants by decimals (how they did on one station vs another, or a mild difference in GPA). Maybe the interview should just be used for cutoffs and if everyone can meet that standard, we use a lottery system?
By taking the top 100 students and randomizing the rest (from 101-550 or so), you have two cohorts that Mac can study to see if there are any actual differences in scores (PPI, CAE, etc.) between applicants.
I suspect there won't be, which could really change the admissions process going forward.