I would say for me it's: Conflicts are resolved in terms of success vs failure

There's a huge tendency for system designers to try to boil things down to the 'opposed roll' or 'roll vs DC', maybe with fancier dice mechanics or things to modify the rolls or whatever. We're so used to 'roll to see if you succeed' that people have a blind spot for coming up with alternatives.

The problem is, boiling things down to random success vs failure is often a really bad way to handle particular types of situations. For example, in situations where retrying the task is logically permitted and has low cost, you get behavior like 'roll until you succeed' or swinginess of attacks in combat. In other situations, projecting a very nuanced situation onto the idea of a single person succeeding or failing destroys that nuance (see pretty much any system with a Diplomacy skill or similar) - if one person 'wins' the Diplomacy situation then that implicitly directs things away from the possibility of some sort of set of compromises where each participant gets some but not all of what they want. Success vs failure also often runs afoul of the issue of the game grinding to a halt or being very swingy: if you fail to pick the lock on the door in the dungeon, do you just go home? So either you succeed, or you fail but keep trying things until one works, or the game grinds to a halt.

The other issue is, success vs failure mechanics have a tendency to lead to the thought process: 'If I want to make this more complex/nuanced, I should make it into a sequence of checks!'. This particular thought suffers from a failure to understand probabilities - the probability of success decays exponentially with the number of checks in sequence that you require (furthermore, its not more complex since there are no additional decision branches between rolls - you could just calculate this joint probability and roll it directly). So you have things like 'if you want to pull off this special move, do a skill roll followed by an attack roll', which is almost always a bad deal (since you're basically rolling twice and taking the worse of the two rolls).

So, what I'd like to see is more focus on different types of outcomes than just 'success' or 'failure': resource pools, bidding systems, wager systems, trade-offs, etc. There's all sorts of ideas in these directions which tend to be mostly explored by fringe RPGs, but which I'd like to see used more widely. I'd also like to see a general philosophy shift from 'what is the chance of success?' to 'what are the consequences of action?'.