I was made aware of Aronson's sequence by Greg Ross' Futility Closet article on it three weeks ago. A couple of things caught my eye. The first was his use of "nine billion one million second" to example the "few T-less ordinals" that "don’t arrange themselves to mop up all the incoming Ts". It would have been a little more compelling if 9001000002 was actually in the sequence — which it is not. The closest t-free ordinal that is is 9001000702.
The second thing was Greg's "We had supposed that the sentence would end with … letter in this sentence. But an infinite sentence has no end..." English number names have been well-defined only up to 10^66-1 — although I fully expect (once Mathematica debugs its IntegerName function) that that will go up to 10^306-1. There exists a longer realization but it may take some time for Mathematica to decide to incorporate it and it isn't obvious to me if the machine-generated naming scheme is potentially infinite. All strictly increasing, current English number-name sequences are necessarily finite, whether or not it is so recognized.