Redux is just one piece of the puzzle. Let me try to explain the problem it solves. Let's say if you are building a complex SPA with lot of UI components in the view and state changes impacts several parts of the view. You can build this view in several ways:
* Plain javascript with jquery - We can surely achieve it but the code quickly becomes a soup of callbacks and events handlers. After a while the code becomes unmanageable. This is pretty much the approach you're suggesting with getters/setters.
* AngularJS's dirty checking is one solution. It works and the code scales in large teams but perf suffers in complex views due to costly digests. It doesn't provide a way to conditionally exclude a sub-view from the digest cycle. This will become the main bottleneck (for perf) as the application grows.
* React and Angular 2/4 - React (shouldcomponentupdate) and Angular 2/4 (onpush change detection) leverages immutability very well to cut down re-render cycles. This is where Redux shines with its immutable way of changing state. It's uni-directional flow is another great benefit to keep the entire flow predictable and maintainable.
I am not saying all these things are needed for every SPA app out there. But these optimizations are needed for complex and performance sensitive applications. Check this out - https://medium.com/@dan_abramov/you-might-not-need-redux-be4...
Also, I know keeping application state as a single state tree is nothing new. I started this side project (https://github.com/guptag/FinCharts) almost four years ago with React/Immutable.js without any state management libraries (recently moved to electron from nodewebkit). But this code won't scale in large teams (entire ui state is declared in one file, cannot extend the functionality like Redux's middleware support etc).