The ratio of his previous PB to his next PB was 7.21 / 4.94 = 1.46. Assuming this is an accurate ratio for an extremely lucky solve, someone whose best time is 6.84 could get 4.69 on an extremely lucky scramble. Granted, this assumes a linear correlation with no offset, but the numbers are close enough that I'm comfortable saying the margin of error isn't enormous.I don't quite understand your reasoning. What if someone with 12s average gets a 6.84 single (I also don't understand why exactly 6.84)? Do you also think, that he's capable of getting sub4.75? Because for example I have official 6.87 and I am 100% sure that I won't ever get sub5 solve, even at home. It's much more convenient to say, that everyone with let's say low8 average is able to get WR-like time.
Also, you can never be 100% sure you'll never get sub-5. How long ago did sub-7 seem impossible to you?
If you say that someone with a low 8 average could get a WR time, that's still 60 people, if you define "low 8" to be "8.25 or lower".