Generally I like Keith Fenner's videos, but I must admit that I was negatively surprised by that segment: why would you ever want to reduce your accuracy by more than five folds. Ples, Like Bimikkalson said, sweeping the indicators in that way, you risk to have errors in reading larger than what you want to measure, caused by the weight-deflection of the indicator.
I cannot find any single reason why that method could be preferable in any case. If your bed is so badly worn that you wouldn't get accurate enough readings, it would behave even worse when you're cutting anything.
Actually, in order to align the tailstock, the bar doesn't even need to be perfectly cylindircal: you measure near the headstock, you flip the bra end-to-end and you measure near the tailstock.
If the bar is long enough (like the one Keith had) and the tailstock is not too far from being aligned, you're probably close enough with a single iteration (i.e. you adjust the tailstock exactly to the value you measured at the headstock).
Paolo