In my experience UKG are generally more consistent across figures/distributors etc, where as AFA are definitely more lenient on trilogos and Uzay/Ledi's etc than they are on Kenner and Palitoy MOCs. But there are always exceptions, like this example. Inconsistency is another reason why I'm not keen on grading, human error and subjectiveness are of course unavoidable when a person is making the decision, so they can't begin to claim grading is fair and consistent.
I'd be much more keen on grading if it was somehow electronic and based 100% on facts rather than opinion, eg as mentioned before a 95% grade if 95% of the card was perfect. But we've all seen mint MOCs with the POP removed score anything from 40 to 60, when in reality the other 95% of the card is flawless.