I guess the core idea would be to reduce the amount of dice tricks, make the individual Charms special and evocative and to make them Bogstromantic.
The first thing you should probably need to address then, is how 'Special and evocative' you want your new Charm system to feel. Because no matter what, you're going to run headlong into the issue that Exalted is ultimately operating as a Game System and not a construct of freeform prose, which means certain effects are going to inevitably fall into broad mechanical archetypes like "damage-adder," "counterattack," "persistent buff" and so on to interface with how Storyteller is a ruleset built atop trait numbers and dice percentages. None of those things are very compelling or magical in practice, but Charms Need to interact with that system to some unavoidable degree. Not even Jenna Moran could fully rid herself of them either, so you can see places in her works where she tried to at least work them towards a Goal.
3e tried to create a type of Specialness and "magical evocativeness" by introducing unpredictability and obscuring the actual gameplay effectiveness of Charms themselves through endlessly (and needlessly) complex variations to and vague interactions between base effects like rerolling and resolution structure, so we can both kind of agree that doesn't really work and grinds at-odds with the required transparency that the system needs to be a fulfilling and nonarbitrary experience to run.
So instead, keeping in mind what "you can't escape the numbers" limitations you're working within, you'll need to look at what those broad mechanical archetypes are performing
inside the system and how you in particular feel they should be performing that, in a way You find to be special and rewarding
in the gameplay. There's two ways of going about this, and Exalted mixes them pretty liberally. The first is the hardest one, because it requires an understanding of the core system: determining Foundational Mechanics. This is plotting out those base-level interactions between system and Charm, where Borgstromancy is the second step as you tease out what the mechanical foundations you've imposed say about the narrative baseline.
Like, here is an example: Perhaps you think that Counters are too strong as an independent attack action, and so instead they serve as a form of small amount of "automatic damage" inflicted upon an attacker instead of a fully-rolled attack with all the resolution complexity that implies. That's not a bad idea at all, and now you can reduce Other similar "counterattack" Charms to be "automatic effect upon being announced as a target of an action." That is your new Foundation mechanic for counters, and it becomes a gameplay Archetype you can assume anyone who wants to be a good combatant will approach because well, as any DB with anima-flux can tell you, an auto-damage aura is a pretty useful thing to have.
But, at the same time, there still exist other "do not target me" effects in the Solar Charmset like Radiant Majestic Presence which can be folded under your new definition of "counterattack." This is where you finally get to make Statements about the setting, and just how the "untouchability" of a Solar combatant manifests itself, and what ways both counters
and Presence (through providing 'social-based counters') assist in that theme. So you can already see the process here: Define Archetypes across the system, find other examples of them and sort them by relevance to those Archetypes, and then define how each Ability tree which has (or should have) that Archetype approaches it in a way which upholds the narrative you want.
The second method is working the opposite way via Informed Narrative effects, which are more detached from the core systems and are are more "layered on top of" the Storyteller system. This is things like Perfect effects and other "announce it and it happens" pseudo-stunt powers like conjuring light or making a bonfire from your hands, which you can't normally make actions to cause at a whim like you can with say, Jumping. A lot of people tend to work from here initially because it requires less systems-knowledge and more "what I want to do" knowledge, and that means they tend to butt heads with
other Charms more often than Foundational stuff. Since you've pinned down the gameplay archetypes you have in mind prior, you have to look at what the game is telling you About the splat in question first and see what Archetype(s) you have which serve do that thing.
Another example, the book says "Dragonblooded use and create elemental effects," but it now falls to You to assign that trait to Something, or maybe even several somethings. I already mentioned redefining "counters" so that DB Anima flux is now a "counter" by this method, but the standard way is by replicating an archetype within the Charm itself, just as a variant on the existing rules for it. Much the same way that "bolt" Charms are rolled as a Simple action with a "weapon" made out of bonuses applied by the Charm, instead of Modifying an existing attack action.
You're using the Charm rules to recreate the effect you want, in a way which matches other areas in the system but not Overlapping them, usually with some kind of rider conditional (like how a "bolt attack" is not a proper
weapon, and therefore you can't tag it with Archery or Thrown Charms dependent on ammo or projectiles, beyond Excellencies for the roll it is making). You get the general idea here, but it basically comes back to the same ideas as the first: Find the traits you want to highlight, determine which archetype you think works best for that purpose, and use it to highlight that feature in a way that the splat in question "always" has access to it in some way so long as they buy into its required Ability.
Hopefully that all doesn't read quite so daunting as the length implies! But Generally speaking, these are the things you should be looking at first when sorting out how to revamp any chunk of the Charm system, wholly or in part. The rest of it comes down to matters of personal taste, and internal-balancing by running the numbers (to make sure your bonuses and penalties are actually Bonuses and Penalties in the long run), and playtesting it extensively to get the results you feel are right.
