It feels circular to me: "Most users are better served by a gatekeeper" -> "Because if users had a choice to go outside the system they would" -> "Therefore the gatekeeper must be mandatory" -> "Because most users are better served by a gatekeeper".
At what point do we have to step back and say, "maybe users just don't want gatekeepers?" Especially bearing in mind that, again, anyone who actually did want a gatekeeper could still use one and be fully protected.
We're assuming that every casual user who sideloads an app on Android is doing it because they're uneducated. What if they're not? What if they just feel like it's an acceptable risk?
What if the reason we can't trust the average user to stay in a moderated system is that moderated systems are not scalable, adaptable, or reliable enough to meet the needs of even an average user?
This is because it is actually pretty reasonable to make the judgement that Amazon is trustworthy enough to sideload their applications despite the dire warnings, so what you really want to find out is how many people are making unreasonable judgements on what is trustworthy enough to sideload.
I agree with you that it's a tricky line to walk, but the UX issue can't be dismissed as a triviality.
You can go through the process now with Amazon Underground, Prime Video, and several other Amazon apps for Android. The user is prompted to sideload Amazon's apps that contain functionality disallowed in Google's app store. Amazon shows step-by-step dialogs teaching users how to do it. It would be really interesting to know how many "non-power-user" users have actually stepped through that process at Amazon's behest.