I'm over-simplifying of course, but these things sound simple until you ask all the "but-then" and "what-if". If someone actually makes them simple in a way that still lets me cover those cases I'll be the first to stop writing authN/authZ code.
AWS very much decoupled authentication from authorization with Cognito and it works very well.
However, we're definitely not aiming it at non-programmers. I don't think completely general low/no-code solutions will ever take off. Everyone wants things far too custom, and at some point you won’t be able to beat the programming language of the time. Companies like retool are doing a great job in specific verticals — and I think it will always stay that way.
I think the next generation of dev-tools are going to be pretty impressive, especially with how easy and modular React components are — I’m bullish that React will become “low-code”
Alternatively, tell me your complicated auth flows that will not have users running away screaming and are, therefor, bad ideas and should be replaced by cookie cutter stuff. Because that is where all of these things end up.
Drop in the plugin you want, configure the API keys, and then wire it into your User model. Then you can customize the urls, views, and emails.
You still have to read a fair amount of documentation and a lot of developers don't like the extra up-front work and reinvent the wheel because it's "simpler".
The simple use cases should just be plug and play. If you want a different solution: here are the tools, here is the documentation, have fun.
If I need, say, authentication, why do I need to install a library and then read a bunch of documentation on how to use it? Why can't I simply plug it in (just like low-code!) and have it automatically work?
I think there is a lot of room to improve developer tooling and that the effort spent there would produce much greater rewards than developing a higher level abstraction, such as a pluggable authentication widget aimed at non-programmers.