Simplicity

Naturally, we prefer simple things over complex ones. If, for example, I hand you an umbrella and I tell you it has five different opening mechanisms for closing and five others for opening, you would say cool, but I'd rather have a normal one. It's an object with a single function protecting you from rain; the extra features do not bring any value. Moreover, if all of a sudden a storm comes and you're well dressed in a costume, you're not finding the damn button to open the umbrella, or the complex mechanism breaks, then you'll throw it straight into the garbage and never buy another one like that. In this case, the extra complexity takes from the value given by the main function. 

I design interfaces for products. My role is to make it as simple as possible for the user to get value out of the product. This means that I must know the various use-cases of that product and optimize UX flows accordingly.
Sometimes I have to accommodate features that I think are irrelevant within the larger context of the app and take from the product's value. In this case, I'll just ask questions about this feature, see if anyone else requested it; if yes, I'll ask more questions about those users, see if maybe they can achieve their goal using the already existing features. If not, I'll make an interactive prototype and test that within a group of customers; if they like it, it gets done. Otherwise, it's up to the CEO to decide.

I think many digital products suffer from too much complexity. Ideally, we want the thing just to work; I call this ironically the "computer make website" CLI command.

The issue with 2D interfaces.
The more you increase complexity in a 2D interface, the harder it gets for a user to create a mental model of it and efficiently project his intent over the 2D space. The learning curve is steep and long, and that will cause him to be searching for features (functionalities) constantly. 

What about 3D interfaces?
The human brain is used to navigating the 3D environment so; a 3D interface could potentially work better, however for some reason, they haven't caught up (probably implementation)

What about AI interfaces?
Well, that's the future.