I too followed the "Midnight Dinner Train" up the hill that night with the intent of getting some high-end digital audio of what was a common event 100 years ago (that is a 2-8-0 helping a 4-6-0 with the daily passenger train).
168 has had a shuffle to it since it went back into service, but that night it had clearly gotten worse. At first I was inwardly cursing a engineer Jeremy for trying to run 168 hooked up too high (come'on, Jeremy, you know better than that!!") But it turns out he had 168 dropped down 4 notches lower than he usually ran her. 168 has pretty widely spaced notches on the Johnson Bar quadrant, four notches is a big difference. With the eccentric strap getting loose, extra lost motion was introduced into the valve events making 168 go lame. As it was the "back" eccentric, the effect was most noticable the higher the Johnson Bar was hooked up. With Stephenson gear, when in full forward gear, most all the valve motion is created by the "forward" eccentric, as one hooks the bar toward center, the "backward" eccentric imparts motion into the valve events, as well as the forward eccentric.
Having to run 4 notches deeper would make any engine tear at her fire and use a lot more water than normal. It is no wonder they had to stop twice to blow 168 hot and fill the boiler.
Still it was a magical night to hear such a mis-matched doubleheader - a mismatched sound that was a daily event 100 years ago. 425 always seemed to be pedaling harder to keep up, when indeed she was providing over 2/3 of the power to get up the hill.