So you're using "not now" as a shortcut to identify malicious design. That's fine but personally I'm not. I would go as far as claiming as most people don't feel this way about such a tiny wording detail, but I obviously haven't ran any A/B testing to confirm either interpretation.
> And, for the sake of argument, how should respectful software behave? If I clicked 'not now', are you _ever_ going to ask me again? If so, how do you decide on the time-period?
Honestly this question seems bizarrely focused on making the software give you a definite promise on what it will or will not do. It's a valid design question how to respectfully prompt the user for choices, but if that's your concern then this specific wording choice is an unimportant detail to get stuck on.
> If not, why make 'no' more complicated than 'yes'?
The reason was given my my opening argument: "no/yes" demands a conscious choice whereas "not now" doesn't. "not now" is plausibly less complicated than "yes" for a majority of users, and cognitive overhead generally matters more to people than imaginary contracts about future choice prompts.