I think that we're all missing an important point with regard to the battery pack degradation. Degradation is due to two things: calendar life and use (and environmental conditions, but I'll just assume those are constant and ignore them). The calendar life effect depends just on the battery chemistry while use depends on the number of cycles of a given cell.
It's the last point that's important here: if you've got a 300 mile battery and you drive it 100,000 miles, then you've used 333 cycles of the battery. Do the same thing with a 160 mile pack and you've used 625 cycles. Therefore, you'd expect the use part of the degradation to be nearly twice as large on the 160 mile battery than on the 300.
Put another way, I think that it's wrong to say that at 100,000 miles you'll be at the same percentage of the battery capacity regardless of the pack size. You'd expect the percentage loss to be smaller (by less than a factor of 300/160) for the 300 mile pack than for the 160. How much less depends on the relative contribution of calendar age to the capacity loss.
Taking a complete wild guess at it, if we say that half of the loss is calendar loss and half is due to use, and say that the 70% after 100K miles for the Roadster battery also hold for the model S if it had a 244 mile battery, then we see that 410 cycles (= 100,000 /244) is 15% loss (half of the 30% overall loss), so the use loss rate is about 0.0366% per cycle. The 300 mile battery after 333 cycles then would be at ~73% (219 mi), while the 160 mile would be at 62% (99 mi).
This makes the argument for the 300 mile battery a little stronger. Of course, how much stronger depends on the relative contribution of age and use which I just completely made up in the previous paragraph. It also may vary based on the different cells used in the 160 and 300 packs, though IIRC the 160 and 230 packs use the same cells, just more of them in the 230 so it should really be linear there.