So what? If you propose to have someone do some amount of running that they wouldn't otherwise have done, they must pay for that running in one of two ways:
1: Eat more. This has a linear relationship to the amount of additional running.
2: Do less. By doing less, you can redirect calories that would have been burned anyway to the task of running. You cannot, by definition, redirect calories that are budgeted for the basal metabolic rate.
By paying for the extra running "in kind", by not doing activities you would otherwise have done, it is easy to show a nonlinear relationship between "total running done" and "total calories expended". But that's spurious, it has nothing to do with the energy cost of running. There is a linear relationship between "total running done" and "total calories expended on running", and that is what matters when calculating the cost of running.
