1) Transactions cost money. There is a hypothetical incentive for a hypothetical manipulator to "pad" the growth (e.g. a miner intending to sell at inflated prices), but the cost of faking those transactions is substantial, and would probably require a concerted effort that no evidence exists for, to my knowledge. Hence, the conclusion based on the least assumptions is that the data is not fake, until evidence to the contrary is presented. (and, "the data doesn't look like commercial adoption" is no such evidence).

2) The no. of transaction graph roughly matches total USD volume transferred roughly matches the no of addresses graph roughly matches the hash rate graph. If the data suggests (even though it is no guarantee for) exponential growth along several dimensions, the most likely conclusion, in the absence of contrary evidence, is that some underlying variable does undergo such growth in fact.


Myself, I'm skeptical towards extrapolation from such data (a point that you made yourself often enough, citing Worldcom stock price as an example, IIRC), but that's a different point: the question whether exponential growth is going to continue or not. Looking at the blockchain data however and questioning whether there was exponential growth until now is a stretch.


EDIT: Added the most pertinent graph. From 100 transactions a day to almost 100k transactions a day, over the course of 5 years.