Dice Are Furniture
I often see people going on about d100 vs d20 vs 3d6, or the forever wars of THAC0 vs AAC (both are wrong, to-hit matrices are superior to either). They are all boring arguments. Here is something worth fighting about instead: your dice system only has the right to be different if it does something interesting with the fiction.
Wait, it's all just statistics?
That's all it ever was.
In the end most dice systems are just oracular random number generators, and they are all pretty damn interreplaceable. A d100, a d20, and 3d6 are all just tools for generating a statistical spread. "But the curve," you say. "3d6 is a bell, the d20 is flat, a +2 is worth different amounts." All true, all beside the point. If you're willing to do a small bit of high school math you could run D&D with a d100, or with 3d6, and it would all feel the same at the table. The specifics of the oracle don't matter. Fight me on it.
Here's an example of when the system does something interesting.
Vampire: the Masquerade 5th edition, for all its warts and bad political takes, does vampires right. You roll pools of d10s, based on attribute and skill. Now, stop yawning, you people in the back. Where VtM does something fun and interesting with it is by saying "oh yeah, and when you get hungry you replace some of them with red ones. You won't like it when the red dice give the wrong results." It weaves the fiction into the dice system in a way that makes the pain immediately obvious to the player. Even if they don't get the exact odds of how bad, the player can grok it in the hand: I'm rolling two red dice because I'm starving is a sentence a first-session player understands physically, before anyone explains a probability to them.
WHFB did an equally good job with its magic system. Roll 2d6, that's the Winds of Magic: how strong the flow of magic is this turn. That's how many dice you can spend on spells. Then, when you cast a spell, you decide to throw a number of dice. More dice equals more chance to succeed, but also more chance you miscast. It makes it legible for the player and gives him a conundrum: "if you want a good chance to succeed at casting your Purple Sun, don't get mad when a chaos demon eats your wizard."
Which brings us to Daggerfheart. The core mechanic rolls two d12s, a Hope die and a Fear die, sums them, adds modifiers, compares against a Difficulty. Whichever die shows higher determines whether you rolled "with Hope" (you gain a Hope token) or "with Fear" (the GM gains a Fear token). Doubles are an automatic critical. Advantage and disadvantage add or subtract a d6. Hope and Fear are spent as metacurrencies: players on abilities and aid, the GM on complications.
I got a little cross-eyed just reading that. There's a huge amount of bookkeeping there. And none of it is all that interesting in the fiction. It's a system feeding into two bland metacurrencies and a generic "and if you roll with Fear something bad might happen!" It's a system that tries to do so much and achieves nothing for the fiction, nor for what it wants the players to feel at the table. The dice may be labelled "Hope" and "Fear," but those names do most of the thematic work; mechanically, they are generic currencies waiting to be translated into fiction later. That is not diegetic in any meaningful sense.
In fact, it detracts from it. There's too much to check and track. Daggerheart purports to be a game about the players and the — sigh — "focus on narrative and character-driven stories," yet you're going to spend a huge amount of time looking at a bunch of dice and managing tokens.
Another one would be Cthulhutech, whose designers thought "now, what if we tried to elicit actual player insanity with our dice system." Start with the generic d10 dice pool, roll all of them, and compare the highest to the target number. Except if you roll matching numbers, which allows you to add them up. And a straight also allows you to add those up! Brilliant — we get players to play Yahtzee while playing our game! This one fails differently from Daggerheart: not by dressing empty currencies in thematic names, but by piling on manipulation for its own sake. Two flavours of the same sin. Doing a lot and saying nothing, versus doing a lot for the sheer Yahtzee of it.
It's fine if your dice system doesn't do anything interesting. If it doesn't add anything (my least favourite word) diegetically, get it to go out of the way so we can get to the interesting part. Sometimes a door is stuck, and it opens on a 2-in-6. Easy, simple. The open door check is not primitive or 'vanilla'; it is finished. There is nothing left to remove.
Either the dice say something interesting about the fiction, or they get out of the way.
Some really interesting systems do the first. Some fail to do either. Many others choose the second path, and that's just fine.
It's okay when your dice are just the table you use to support the game you play.