Everyone will have an opinion on it. And it is all about whatever works for the individual. But just for instance sake, say you have a 2,000 foot wide field and are going back and forth. If you run a 30' head and cut 29.5' with it or more, that will take 67.7 passes or corners. If you run an overlap of 1.5-2 foot which seems to be about what I've been seeing in the picture on the net, that would be 71.4 corners. Other than the new computer controlled hydraulic driven fan combines, which aint perfect either, each time you leave the crop you loose grain over the back, and each time you re-enter you loose grain.
Now thats for a 30' head in what is likely a small field for that size head. "here" in my area of the pnw, a 20' head is much more common, and a few 25's have come around with the 2388s and larger. So a 20' head in a 2,000 foot wide field would be even a greater loss by not cutting a full head width. The wider the header, the less difference made with some offset. On a single rotor machine, its not so big a deal. But with a conventional or double rotor (I use double instead of twin for pc reasons) it can make a difference in being difficult to set or heavier loss because one side will be starved. When you figure that, you have to divide the header width in 2 and each half gets its half of the threshing unit. So now a 30' head will send 15' to one side and 13' to the other side. That's what, about a 15% difference in material volume?
In our tall fescue grass seed, when swathing a lodged field, some will be leaning "outward" from the header and some inward. I have enough experience, I can do pretty well judging the true line to cut by, but the other machine has an in-experienced operator in it. He will follow the artificial cut line and have a reduced width area. Using the "operator stress" idea, this worked for the swather operator, but did not eliminate the stress as now the combine operator has to watch the machine load in these reduced width areas and compensate for it.
