It seems to me you have a very valid argument. For what AASHTO is saying with distribution of loads to the substructure of the transverse timberplanks can be taken at 20" whatever structural is kept within such width *** could *** be thought apt to resist the load for it *** would *** be getting there already distributed. That's the positive look.
The negative one. It seems AASHTO is saying doesn't want in place any bending member not able to carry at least an undistributed line of wheels. Hmmm. This sets a minimum for such "bending member" leading to big sizes, then the same for transverse.
Now ... I READ (sometimes do) we are talikng of Shear.
Maximum shear wheel loads near supports have almost NIL lateral distribution. How more when the bending members may be thought unable to pass shear ones to others, such in stringers. So, well, this puts the size of the overall stringer if of constant section in the need to meet such sehar when the loads are at maximum, at ends. After this thought, I would either meet the requirement or reshape the structure.