That is all true only if Q* stays within the range of your assumptions.
That seems like a strange way of wording what I already said: "As long as the limit is far above the transaction demand (much greater than Q* in the figure), then the supply is constrained economically rather than algorithmically."You've really lost me here. If the growth in demand for transactions is greater than what is permitted by BIP101, then we'll bump into the higher limit, and fees will increase more than what would be implied by my paper. In other words "as long as the limit is far above the transaction demand (much greater than Q* in the figure), then the supply is constrained economically rather than algorithmically."
I don't understand your concern here. What would be so bad about hitting the higher (and constantly increasing) limit earlier than historical extrapolation would predict? [not that I think that's likely, but still]
To then go back to our previous argument, let's pretend that your 8 MBs blocks get filled a year before your scheduled doubling then what happens?
Then fees would be more than what's suggested in my paper. What would be wrong with that [in the unlikely event it happens]? In fact, it would just be something like the realization of the "high" growth curve shown here:
I guess my point is that modeling transactional demand based on historical growth is a surefire way to shoot yourself in the foot.
There are several "transactional demands" modelled in the above graph. They would all work fine with BIP101. It's just that in the case of the "high growth" curve, the fee market might begin to be affected by the block size limit, rather than strictly due to economic supply and demand.
How do you propose this is "shooting yourself in the foot?"