forecasting is not an exact science, far from it, but the way you did it is incorrect. A better way is to increase the difficulty based lets say on the average of the last 5 periods + 10% or  -10% depending on the direction. Your exponential increase eventually runs into requiring more and more grains of sand to the point that there's none available. The difficulty will run into a wall where GPU and ASIC farms will start closing. Many smart (cough cough) people is mining to be rich. Wrong move, they will at best break even and will be served divorce papers instead :-) Also count on BTC going down to previous levels. The current BTC price requires new money coming into the system at a pace of 100 million dollars ++++ year. That will not happen