I decided to take a look through these forums after how utterly useless for discussion the BG3 Reddit is with all the fanart and meme spam that happens there, and was pleased to discover that actual constructive feedback happens on these forums.
I think that the main thing that irks me about the way BG3 handles combat currently is that the homebrew rules more or less completely overpower everything else, in terms of the amount of emphasis that is placed on them in the game's balance. People really wouldn't be calling BG3 a 'DOS2 mod' if it didn't have a lot of truth to it. Granted, Larian has been working on reducing such comparisons starting with cantrips no longer generating field effects by themselves and field effects applying their effects on a per round basis instead of based on amount of distance traveled within them, but they still have a lot of work to do. Some of the absolute biggest problems are still there, and it would be harder to rebalance the game around changes to those problematic mechanics, because it's quite evident that many encounters are straight up designed around their existence.
First of all, high ground advantage/low ground disadvantage. Many people on these forums have gone back and forth on how bad or strategic this supposedly is, but it's really, really not a good thing when the majority of battles boil down to controlling the high ground as soon as possible. It also heavily skews the value of AC by sheer virtue of existing - by the way that it skews hit rate probabilities for ranged attacks, AC lower than 16 may as well not even exist if you're on low ground. Not only that, but such a free method of gaining advantage rolls heavily cheapens the value of spells that grant advantage such as Faerie Fire or flat increased modifiers such as Bless (of which the devs have already talked about reaching a rather unfortunate conclusion about the community not using the latter). Most games design high ground/low ground as a means to control enemy or party movement, and BG3 is probably the only game I've played where its importance as a way to improve *hit rate* in such a drastic way drowns out such other tactical considerations.
Granted, it's possible this only exists because backstab advantage is a thing, though BOTH really should be removed.
Jump and Disengage being coupled together and becoming bonus actions, along with shove being a bonus action too. Let's face it, the only reason this is a thing is to enable the devs to go wild with field effects design, and yet Jump being a bonus action makes it so common that it simultaneously cheapens the impact of said field effects at the same time. Not only that, but opportunity attacks are less emphasized with their existence, which makes the balance between melee and ranged characters and enemies rather awkward (and especially when you factor in the earlier high ground/low ground stuff on top of that). The concept of melee pressuring ranged is basically just an enemy-only concept as it currently is too, as players lose way more from having to use their bonus action to disengage than enemies do by means of limited action economy. Oh, and bonus action shove just makes the high ground/low ground concept even more ridiculous.
(Oh yeah, Harpies being glorified winged archers that just spam a modified jump every turn is quite ridiculous design. It's quite the cop out for not having them as proper flying enemies. If they actually flew, spells like Sleep could suddenly gain value by having them drop to the ground to take massive fall damage. Granted, flying Harpies could be even more annoying to melee, if it weren't for the fact that Harpies did not have ranged attacks in tabletop and that their main means of attacking were to swoop down to swipe at you - meaning you could do things like ready action a melee attack of your own.)
Barrelmancy and eating food in combat. The lowest hanging fruit. Abusable stuff like that was a cute and novel idea in DOS, but that was a game series where your suspension of belief was broken so often that nobody should have legitimately taken the writing and narrative any seriously (especially since the writing in those games were already well beyond the 'comically absurd' line to begin with). Importing them into BG3, however, results in a massive gameplay/story whiplash that can't really be ignored. Here we have a story that includes stuff like Illithids that are working towards some secret goal, a bunch of Tieflings and Druids fending off an invasion and trying to survive, and Goblins that are supposed to be treated like an actual serious threat even if they're only being manipulated by some greater power. Yet everyone has random exploding barrels all over the place and can swallow food whole during combat for... Reasons?
The major thing is that all builds have access to using such concepts in combat with little to no limit or drawback to using them. If one designs the game's balance under the assumption that people are going to abuse such powerful tactics that are so readily available, then they're walking a pretty dangerous line between making an RPG and instead creating a glorified puzzle game that just happens to have RPG mechanics. And speaking as someone that had hundreds of hours in DOS2, the latter really isn't something that should be encouraged, because everyone here who actually DID play DOS2 to completion should remember how terrible the balance was by endgame - terrible enough that they had to make a Definitive Edition to rebalance the entire endgame and do a major stat crunch on top of that, and many would argue that it acted as a band-aid at best. Encouraging cheese doesn't mean that you're fostering creativity, it just means you've failed to make the other 90% of options feel any rewarding to consider using in comparison.
Sometimes, giving people more options actually results in less viable strategies. In the case of BG3, the current design heavily prioritizes heavy offense or cheese strategies that may or may not involve repeatedly reloading until you get that first or second round just right. Not only that, but the defensive and proactive options that actually let you do things *during the enemy turn* such as reactions, dodge action, and ready actions don't really exist at the moment. I imagine most of the doom and gloom is also in part due to how no one is even sure exactly how they'll get implemented, if at all, due to some very worrying developer statements about them that are probably emphasized a bit too much in these discussions - but they are there and on official record, so that's that.
All of that combines to form a combat system that, quite frankly, isn't compelling at all. It's telling that most people on the more populated corners of the internet would rather make fanart and argue about romances rather than talk about the actual gameplay. And the few times that the gameplay is brought up, people get into a huge fight about what does and doesn't work. It is especially egregious when the Larian homebrew-favoring side essentially mansplains to everyone about how the already established rules aren't 'fun', wouldn't work in an actual video game, and/or it would be too much work and impossible to implement everything by the book, when it was likely already more dev work to create these new homebrew concepts and balance entire encounters around them to begin with. Larian more or less made problems within their combat balancing where there previously were none. And no, 'get gud' isn't an actual adult response to these concerns.
What would I do?
Change high ground advantage/low ground disadvantage to +/- 2 AC. The same thing that has likely been suggested numerous times before. It would go from being an overwhelming factor in a fight to being a replacement for the missing cover system. It would give value back to spells that grant the party advantage, it would remove the need to have backstabs exist in order to keep melee somewhat competitive, and every point of AC is suddenly valuable instead of being an exponential stat.
De-couple Jump and Disengage, and make shove a standard action. Make Disengage a standard action so that the player is actually rewarded for having their melee pressure an enemy caster or archer (and thus forcing the enemy to either disengage or do something at disadvantage, instead of the current design of disengaging and then suddenly sniping at Gale in the back with no penalty). This would make it so that maybe using that standard action Dash to run up to them is a better tactical idea than just whacking at the nearest enemy. Jump can remain bonus for tactical combat mobility purposes, but using it to jump away from an enemy should provoke an opportunity attack - and you should be punished for leaving your back line exposed instead of having a free escape card, unless you had already considered other backup tactical options to counter that enemy running up to you in melee, like using a Shocking Grasp to remove their opportunity attack, or maybe you wanted to bait that melee into coming over so you could shove them off that cliff that they are now standing next to.
Barrelmancy should be very sparse and reserved as setpieces for special kinds of encounters. If a player can pick up an explosive barrel of any sort, increase their weight to the point where no one but a high strength character could realistically lug even a single one around. Placing them and other heavy objects down should emit a sound that causes nearby NPCs to investigate, and simply being near a barrel should confer disadvantage on stealth checks, because of course everyone wants to make sure there's no one suspicious next to a freaking explosive. And if you're caught next to that barrel you just put down? Better hope you roll higher than the enemy ranged on initiative, or else they're going to punish you hard by immediately blowing up that barrel in your face. (People carrying around grenades and special arrows are fine though, if only because you can't realistically set them all up to explode immediately on the first turn of combat.)
Food during combat should be shot into the sun. They should just be a way to heal out of combat without using spells, scrolls, and potions (basically acting as this game's bedrolls), and that's it. It's completely absurd that some food heals as much as, if not more than actual potions and spells to begin with, doing it in combat is another even worse thing entirely.
Get the dodge action, ready actions, and reactions in. Reactions are self-explanatory. Dodge action is a major defensive tool in what would otherwise be a heavily offense-leaning game, and can give the player something useful to consider if a character isn't in range to do anything else instead of just having a wasted turn doing nothing at all (and if there's one thing that Solasta does exceptionally well, it's that the implementation of all of these makes it exceedingly rare to have a character end up with a fully wasted turn doing nothing at all). Ready actions would require better planning to use well, but again, it is also something you can do that can potentially let you do something useful during the enemy turn, especially during certain encounters where your options are limited to begin with (such as that tabletop Harpy example mentioned earlier). Maybe ready action a shove to attempt to yeet off the first enemy to climb that ladder or ledge in front of you, perhaps? (Something not even Solasta currently lets you do!)
Or tl;dr: Solasta did everything right in terms of combat design and balance along with actually forcing you to think about what you are doing rather than having a set checklist of priorities that applies to almost every fight, and I see BG3 having very unfortunate post-EA gameplay in the long term just like in DOS2 if the devs don't reverse course on most of the above concerns.