Many of us are enthusiastic worldbuilders. And many of us worldbuilders also try to make our RPG settings "original".
Yet there is a danger to making a setting that has too many unique elements. After all, the setting isn't just for the game master, but also for the players. They need to understand the setting enough to play convincing characters that exist in that world, and act accordingly. And the more setting information they have to absorb to do so, the more difficult it will be to get buy-in from your gaming group.
Personally, I think settings work best when there are a few strong, central unique elements that support strong themes on what the setting - and the campaign - are supposed to be all about, scaffolded by numerous familiar, common tropes from the intended genre (such as science fiction or fantasy). This should help convey the information the players need most effectively.
But I am curious how you see this issue. Do you regard this as a problem at all, and if so how do you approach it?
There is only a problem when your world is not consistent within itself. That means, if the players learn a fact about your world, they must be able to use this fact to predict outcomes of their actions. Otherwise, they might as well be monkeys with a typewriter, i.e. their actions could just as well be random.
But other than that, you are also completely correct that if you have too many things that are different from any established experiences the players could draw from, play also grinds to a halt. Like, let's say you build a setting that doesn't run on any physics we know of, there are no 3 directional dimensions, there is no time, etc etc - I don't know how that would look, but you would basically be asking your players to learn a completely new way to live and make decisions in this space, which is possible, but the simplest thing might require lots of thinking and problem solving.
With that in mind, it's important to keep most of the inner workings the same as either our current established reality, or some well-known fantasy one.
So long story short, I agree with you, it is important just to have a few things that work differently than "expected".
However, "expected" can be very different depending on which players you find. So, as always, I think the most important thing is just to have a proper "session zero" or player selection process, where it is discussed how the setting looks and how far away from established norms it is. If everyone agrees to take it very far away from those norms, then you can be more "original", if not, then you either can't or need to find different players (and if you can't find any players, then you probably went too far :p).