Thought I would have a quick play, run an unrealistic scenario of an allow wheel, unrealistic in the fact that you cannot have a load over the width of a particular section, so the load is distributed over the entire circumference. The results would be similar to a point, that is to say that the load is only applied to the wheel where contact from the tyre and the tarmac occurs but the deflection would be the same, just not all over the diameter of the wheel.
The rim furthest from the spoke does not necessarily see the most stress because the wall thickness is larger there, as it has to retain the tyre wall. The highest concentrations of stress occur just in from the rim and you can envisage that the crack propagates outwards from there.

The rim furthest from the spoke does not necessarily see the most stress because the wall thickness is larger there, as it has to retain the tyre wall. The highest concentrations of stress occur just in from the rim and you can envisage that the crack propagates outwards from there.