Having solid fundamental mechanics design & core logic is the most important thing in games because it's future proof. This is somewhat related to my post on why Risk vs Reward doesn't scale well.
Let's say you're designing an action game, and it's time to work on its hardest difficulty. You wanna have an absolutely batshit mode that will push even the best player to their limit. So you start working on it, trying to make it hard but well balanced & fun.
But there's a bit of a problem - you aren't a great player. Perhaps your reflexes are lacking, perhaps your execution and hand-eye coordination isn't there, perhaps you lack consistency, perhaps you're not good at coming up with outside the box strats. Whatever the reason, you realize that you cannot at once make sure your game's balanced yourself, but also make a difficulty mode that goes far beyond what you can do as a player.
There are some things you can do to alleviate this. Playtesting from great players can help, but they are hard to come by. And they won't necessarily get great at your game. You might try isolating parts of your game to make sure each chunk is beatable. But this ignores how a long enough series of barely doable challenges can become impossible when put together, among other things. You might approach this from a theoretical perspective and do some math to make sure things are beatable, but you can (and will) miss key parts of your equation.
THIS IS WHERE THE FUNDAMENTALS COME IN!
Difficulty doesn't just test the players, it tests the fundamental design of your game. Every hitbox, every input buffer, every instance of RNG. Everything good will shine brighter and everything that sucks will suck much, much more.
Each genre and game has its own set of fundamentals - they are the things that lock into focus as you ramp up the difficulty. Usually this will be anything related to input precision, mechanics and techniques that allow the player to react to RNG/chaos driven variation, the quality and consistency of hitboxes, the clarity of player character states, etc.
It's easy to brush aside fundamenal flaws in an easy game. A forced hit? You have a health bar! An unclear moving hitbox? The pecision requirements aren't strong enough for those 3 pixel movements to matter! RNG that creates wildly different encounters in terms of challenge level on restart? All of them are beatable anyway! Excessive coyote frames and input buffering? The platforming isn't THAT tight!
For you to truly be able to trust yourself and your own game as the difficulty ramps up and these things start breaking, you need to have a very solid, intuitive grasp of the fundamentals. That's the only way to make games that go beyond your personal skill level but don't devolve into unfun frustrating pieces of shit - kusoge.
Or just become so great that you can test it all yourself, like the creator of Crimzon Clover, CLOVER-TAC did.