I think the key thing to remember is that capabilities, ease of use, and simplicity are not the same thing. Heck, we could even separate simplicity of interfaces and simplicity of implementation. For instance, the old phone have a very simple interface. But implementing the circuit switching behind it is no picnic.
From then it is easier to know what you want: capabilities and ease of use are good, at least in the short term. But simplicity (both of interface and implementation) is likely to lead to even more capabilities and ease of use in the long run: a simple system is easier to modify or expand, and a simple interface is potentially easier to use once you get past the initial learning curve.
A final advantage of simplicity is that we can give it a more formal definition than ease: it is the part that doesn't depend on us messy humans.
From then it is easier to know what you want: capabilities and ease of use are good, at least in the short term. But simplicity (both of interface and implementation) is likely to lead to even more capabilities and ease of use in the long run: a simple system is easier to modify or expand, and a simple interface is potentially easier to use once you get past the initial learning curve.
A final advantage of simplicity is that we can give it a more formal definition than ease: it is the part that doesn't depend on us messy humans.