From a development efficiency perspective, Famous is not as good as Angular or React. From a rendering perspective, it's not as good as three.js. It's not yet a leading tool. Jack of all trades, master of none. Maybe it will make more sense when VR platforms become mainstream.
What was released today was a rewrite of the Famous core engine, so you can't really compare it to Angular or React. Three.js is a far better comparison, though the two have different specialties since Famous isn't trying to be pure WebGL.
The Famous team is working on a framework that will interface with their core engine, making things far cleaner at the application level while facilitating less painful integration. Comparing that to various client-side frameworks will definitely make sense.
To expand on this, Famous's Node class [0] is rather like a DOM node, except far more powerful. You could imagine a React component that renders a tree of virtual Nodes, diffs that with an existing tree, and transitions properties accordingly. But that will be built on Famous, just like react-canvas [1] needed the canvas API to exist. Baby steps.
I think that pulling the core out into a separate engine is definitely the right direction and it will be interesting to see what framework they come up with. It definitely needs a clean slate - The Famous/Angular stuff looked promising but it came across as a bit bulky.
