Having endured several such calibration sessions at FB/Meta I can also assure you that we are not even talking about 'large' numbers before there is a strong expectation that the ranking totals within an org will meet the predicted curve when it comes to the percentage at exceeds, meets, and meets most. I distinctly remember wasting more than 15 minutes of my life arguing with a skip level manager that someone who had course corrected and delivered solid work while digging themselves out of a PIP was at least a meets before eventually realizing this manager had a quota of meets most to fill and the person we were discussing was destined for that bucket just to make the numbers look as expected.