Originally Posted by
wphamilton
The major (real) objection is that you'll be measuring drive train losses and not frame flex losses, because the latter is much smaller and will be lost in variable "noise".
You are going to need to stay in the exact same gears as well, because the 2% drive train loss on derailleur bikes is best case. it can range up to 4-10% depending on the gear and how well maintained the chain is. If your experiment measures losses of 2% plus or minus an error, the frame flex may be inside that "error", and therefore not measured. It needs some valid way to subtract out the drive train.
And if we keep drive train the same? Of course even identical components won't be the same after installation. We aren't measuring drive train loss anyway, but the change in drive train loss between bikes. That may still be well above any measurable frame loss.