There’s a lot of writing about game design in both theory and practice, but what most of it boils down to is that the opportunity to make meaningful decisions on a regular basis is fun.
– Alexander Williams
That puts a lot in a nutshell. Why is too much randomness a problem? Your decisions have no meaning if they do not affect the outcome. When the outcome is known at the start due to radically uneven opponents, again your decisions have no effect on the outcome. If the outcome is mutable, how meaningful was that decision when it is wiped away with the next tide? And of course grinding is when you have stopped making decisions and are just carrying out a known algorithm 1000 times until you level, get that rare drop, etc.
The extreme case is “no decisions,” like Progress Quest or Candyland, but the less that it matters what decision you make, the closer you get. If randomness or the starting state determines the outcome far more than your decision does, you could just as easily make the opposite decision and get the same outcome. If the outcome is mutable, and 30 seconds after you’re done the result is wiped away, you could just as easily make no decision or just not show up and it makes no difference. The game plays itself, with the player just cranking the wheel to make it go through the motions.
In some sense, part of the point of games is to have low impact, mutable decisions. You get to fight dragons, blow things up, and conquer the world without any risk to yourself or others. No matter how important your decisions are within the game, once the game is over, you declare a winner and are done. But your decisions need to matter within the game or else your participation in the game does not matter even within the game, at which point it is recursively pointless.