Haven't read the whole paper yet, but it doesn't seem that deep. If you asked any theorist if a TM can do something interactive, they would have said no, and I

Haven't read the whole paper yet, but it doesn't seem that deep. If you asked any theorist if a TM can do something interactive, they would have said no, and I would imagine that most people who know about TMs would understand this. "A TM can do anything" is shorthand for "a TM can compute any function."