The original was specifically developed for the ASIC transition, would you say that difficulty increasing by a factor of 10 over the year after the first ASICS come online is realistic?
It will probably be more than that. But it will also have a different shape than steady exponential growth.
The time period will probably be more or the growth factor of difficulty will probably be more? It will precede in waves with each new shipment of ASICS, but how can I model that?
The growth factor will be more. It will probably be rapid, fairly linear growth for a few months (up to x10 maybe?) and then for the rest of the year it will grow to about x20 of the original amount. This assumes constant rate, a changing rate will cause the difficulty to adjust proportionally with some delay.
So you expect linear growth where difficulty =10*oriniginal_difficulty for the first few months and then where difficulty=20*original_difficulty after that? That's fast.