I agree. I'd go further: even if we were running some kind of crazy ML on a per-merchant basis, I still don't think we could cater to every single use case. I don't think that any UI we build -- no matter how many knobs and whistles it has -- will handle everything. The Checkout page points out that you can still build every pixel of the UI yourself, and Stripe will always support that. Checkout is another option alongside the integration we've always supported, not a replacement.
No, PayPal is fucking incompetent. That's why they have like 4 APIs that do the same thing, except they don't have feature parity and you need a feature from the Pay API and another feature from Express Checkout and you can't do either one of those because PayPal sucks.
Even if you ran a controlled study and determined that this checkbox increases conversion by 100x for 99% of your merchants, it is still quite possible that for someone out there with some business in some selected market it harms them (potentially enough to be devastating). There is a reason why PayPal has tons of tiny little options for manipulating parts of the display given to the user, and it isn't because they are somehow incompetent. Unless you are running, per merchant, some kind of machine learning algorithm that optimizes the display for them, thinking you can optimize the flow for everyone and then should purposely not provide options because the merchant might harm themselves if they don't listen to your advice is nothing but hubris.