This game doesn't deserve to be called Baldur's Gate.

It's not a bad game as such, but it comes up short.

There's plenty of positive things to say about this game: The gameplay and combat is pretty good throughout more or less emulating 5e D&D combat with some changes, the game is pretty nice looking and the production values are generally high even if the over-the-top high fantasy aesthetics don't appeal to me personally. There's some good performances in the game as well and it's very impressive that the game is fully voiced (other than your PC in dialogue, which I don't count against the game). Overall the game is impressively ambitious. Some of these things are something of a double edged sword though, more on that later.

So, while most of my negatives will be very much up to taste, it is undeniable that the game is unfinished. The entire game is janky in every aspect of the game, but the further into the game you go, the rougher it gets. Then it almost completely falls apart toward the end. Performance takes a huge hit, weird glitches and bugs start appearing (both visual and gameplay), quests start breaking, dialogue scenes start breaking, enemies see you through walls, people start conversations with you from a mile away. It's really bad.
Your choices end up having very little impact in the end. Especially one persistent, very foreshadowed choice that you choose to do or not do throughout the game, ends up seemingly making no difference at all. Toward the end NPC interactivity drops significantly, wherein almost every NPC becomes an animatronic puppets barking single lines of dialogue, where you cannot interact with them, like you would in any other CRPG. It feels like companions stop participating in the story almost altogether, sans their personal questlines. Some of this is present before the final act as well, but toward the end is where it becomes very jarring. This is part of the aforementioned double edged sword for wanting to voice and animate all dialogue in the game.
It also feels like they cut an entire area out of the final act, an area for which you can see an entrance but can never access. That is fair I suppose, as they didn't manage to populate even what they had with meaningful content, I wouldn't want to see it spread out even further.

But then there's other CRPGs I've loved even though they've had lackluster or unfinished final acts, such as Tyranny. So what are the other problems?

I said that the gameplay is pretty good, but it has its own share of problems. In tabletop D&D 5e, the DM should be careful about making their combat scenarios too large, because the system scales pretty poorly and becomes a slog. Same is true here, and Larian was not careful about scaling their combats. There are many mass combats with way too many participants that end up being such tedious drudgery. It becomes more tedious when the game has its characteristic slowdowns where enemies just stand there for 30 seconds before apparently succumbing to analysis paralysis and skipping their turn. This game is so janky.
Then when it is your turn, you'll have to contend with some very bad user interface and user experience. Targeting your spells becomes maddening when the AoE indicator keeps wiggling and flickering around. Good luck not hitting your allies.
Sometimes the game just kind of freezes for about 5 seconds just to figure out what's going on. Sometimes the game will tell you, you have a 100% chance of hitting... and then you miss 3 times in a row. What?

Then there's the jank of party control, or more accurately lack of control. It's fine most of the time, except when you present me with areas with traps and hazards which are triggered by characters stepping on them, and then have my party step on everything without my say-so. Not to speak of when a party member just decides to stop following you for some reason. Where's Shadowheart again? Oh, she's on the other side of the map standing in front of a chest high wall every other companion jumped on top of no problem. This game is SO janky.
In general the quality of life and usability features of this game are very lacking. Inventory management and looting especially is a bane for the entire game.

I really don't like the camping system. You're just whisked away into a pocket dimension in an instant, a place that is always safe regardless if your camp is in the middle of a dangerous dungeon. No watches, no wandering encounters, no meaningful consideration for resources or safety. This may not be a big deal to many, for me this hurts the immersion of feeling like this is an adventure. For reference, if you want to see this kind of system done right, play Kingmaker. In fact, my general advice is to play Kingmaker instead of this regardless.

Then there's the writing, which is what I think truly makes this game unworthy of having the name "Baldur's Gate". It manages to be kind of entertainingly cartoony at best, and pretty bad at worst. The companions are a very mixed bag with middling highs and steep lows. So many of them have a kind of a "coolest guy ever" syndrome going on, where they have these incredibly over-the-top grandiose backstories. We're level 1 or 2, and my companions are formerly paramours of gods, right hands of an archdevil and the most notorious warlock-batman of the whole region. It's ridiculous, and so lame. There's a couple exceptions, though. Astarion and Shadowheart turned out to be ok as characters.
Oh and everyone wants to have sex with you, for some reason? I think Larian imagines that the end goal of any positive human relationship is to have sex. It gets even worse when all sorts of otherwordly being start wanting to bonk you as well. It's embarrassing, and juvenile.
The story and writing generally runs the gamut from tropey and shallow, to childishly melodramatic. Plenty of ironic detachment, Marvel-style smug quips, squeecore, and scenes where it feels like you're a receptacle for exposition rather than a character. If you're looking for something with depth, maturity, interesting character dynamics, or complexity you won't find it here. I think pretty early on there was a villain who wanted to kill a kid without a good reason other than she was just that bad? You won't find a character like Jon Irenicus in this game, I'm afraid.
Oh and the humor. I was afraid of seeing Larian -style wacky humor, and Larian provided. Comedy is probably more subjective than most other types of writing, but man the sort of 2010s style random internet humor stuff doesn't work for me at all
I'm not pretending that BG 1 or 2 were perfectly written. Both had their quirks and clichés, but it was much more nuanced and complex than this, especially for its time.
What the other Baldur's Gate games did much better as well, was portray a world and place with a reasonable degree of verisimilitude. You felt like a character inhabiting a place in the world. BG3 feels more like walking around a high-fantasy theme park. It seems like Larian really favors having big open maps where everything in the current section in the world is present seamlessly, but it makes plot points like "None of our scouts can find this place that's next door 5 minute walk away from here" feel really ridiculous.

I feel like this property was given to the wrong hands, or maybe shouldn't have been given to anyone at all. The game is fun, and impressive in many ways. However, Baldur's Gate deserved better than a 'just okay' CRPG with a big budget, low artistic ambitions, and all-encompassing jank.

Baldur’s Gate 3 is incapable of escaping comparison to fifth edition Dungeons and Dragons, and it clearly doesn’t want to try. It aestheticizes the act of rolling dice, offers cosmetic skins for when rolling them, and the d20 polyhedron is a core part of its narrative’s iconography. It aims to capture the Dungeons And Dragons Experience, something heavily informed by Critical Role, Community episodes, and broad parodies as much as it aims to capture the 5e player’s handbook.

5e works to take a genre renowned for fiddliness, sand off the rough edges to lower the skill floor even if it harms the ceiling, market it as a beginner-friendly experience even if it lacks common features for any other modern game on the market and has lacking systematization for non-combat sequences, use its unique market position to have eye-catching production values that no other game can attest to, and then have the canny and luck to release right in the middle of a massive upsurge of interest in tabletop roleplay. This, coincidentally, perfectly describes Baldur’s Gate 3 as well.

No game in this genre has had this much effort put into filmic direction and realistic visuals since Dragon Age Origins, and its focus on high-quality cutscenes, voice acting, and motion capture is exactly what was needed to make it sell Elden Ring numbers. Twitter artists’ attention spans have been caught by it in the same personality-altering ways as Mihoyo games and Final Fantasy XIV, which is the true mark of a culturally significant work. Doing mocap for every NPC in the game speaks to its unparalleled ambition and production scope.

Mechanically, the game attempts to offer systemic interactions uncommon to the genre. A confession: I could never get into Divinity Original Sin or its sequel and thus struggle to directly compare its execution. On its own merit, the implementation suffers because of how stifling 5e’s action economy is. Taking two standard actions to throw a barrel and ignite it in a game with a four-man party limit is very rarely the most effective use of a turn, and magic takes forever to start creating interesting environmental effects, by which point it is probably a better use of time to turn one cast Haste on the cracked-out martial class of choice and let them solo the encounter instead.

Martial classes get changes from the 5e PHB which are highly appreciated - general rebalances make some subclasses more compelling, weapon proficiency now grants combat maneuvers with each weapon type that replenish on short rest which allow for debuffs, area-of-effects, and crowd control in ways that are pretty logical for each weapon type, and the oodles of magic items synergize incredibly well with oft-neglected playstyles. An open hand monk with a three-level dip in Thief for the extra bonus action can crack 300 damage per turn by endgame, and if one has, say, a decade of Pathfinder brainrot, that’s deeply satisfying in ways not offered by the uninspiring feat and class feature list.

It’s inherited from the tabletop game, but it’s still disappointing how guided and on-rails character building feels. Feats are hard to come by and are in direct competition with ability score increases, which creates incredible opportunity costs for efforts to go gimmicky. Being guided towards picking one, maybe two build-defining feats and dumping others into ASI feels really bad, and multiclassing means losing out further and subsisting entirely off of choiceless class features or every-so-often subclass bonuses (and 80% of subclasses either have linear progression or are just picking spells with extra steps).

The lack of flavor is reflected in non-combat dialogue choices, which focus on the act of decision-making as a substitute for roleplay. Actual dialogue options are bland, simplistic, and any personality is pre-defined from the player’s chosen origin or class. The focus on full mocap, as much as it means there are truly excellent performances (shoutouts to Astarion), means individual conversations run short and utilitarian. Dialogue often lacks distinct character voice (Astarion and Lae’zel exempted), it instead gaining its sparkle from behavioral tics, quivers in the voice, or sweeping body language. Each line is usually just a short sentence or two, and conversations rarely run for too long.

Similarly, quests are often binary and offer few chances to meaningfully tinker with inputs, outputs, or outcomes, or are obscenely frustrating in the lack of consideration for alternate paths. The quest chain that defines Act 1, a crisis between goblins, druids, and refugees caught in the middle, outright resists any method of play that is not “go to druids, get quest to go to camp, go talk to goblins, kill them where they stand, teleport back to applause.” Narratively, siding with the goblins immediately loses two party members from the player’s ranks and has literally zero in-character justification past moustache-twirling villainy. There is no way to make the tieflings flee ahead of the assault, and they try to kill the PC for suggesting it. Interrupting the druid ritual similarly has no real effect on the outcome. Playing both sides and luring Minthara out into the open grants the player a harder fight with more at risk and less roleplay reason to do it. In the third act, the narrator (assumedly playing the role of game master) drops the artifice several times to clearly explain the binary choice at play and suggest there is no other route, which is, put politely, advice no GM should ever take.

This sense of railroading and resistance to straying from the beaten path is omnipresent, and further hindered by its frankly godawful approach to ability checks. Here is where the fetishization of dice comes to the fore: it outright ignores how 5e is supposed to be played solely to introduce more randomness. The Dungeon Master’s Guide (and 5e’s director) states to use passive checks for rolls that aren’t player-initiated - passive checks working by simply adding ten to the player’s proficiency bonus and deterministically answering whether they succeed. This allows the GM to conceal rolls, a valuable tool, but equally so it being non-random is important. Player’s skill choices and ability spreads affect the outcome when they’re operating on autopilot, not their luck. This is in contrast to ability checks, which are high-stakes and player-initiated.

The d20 is an inherently insanely random and swingy die, and even skilled player characters are often at risk of failure. This is exciting in a tabletop game, where failure can have unintended consequences. 5e’s DMG (and its director) knows that failure despite extensive investment can be frustrating, however, and explicitly rules that critical failures and successes should not exist when rolling ability checks. Literally none of these rulings operate as such in BG3. Passive checks don’t exist; everything is rolled. Critical failures and critical successes both apply to ability checks, and success/failure is either strictly binary (either losing out on quest progression or getting sent into combat, or getting exactly what is desired or and/or avoiding combat) or entirely superfluous (a lengthy series of ability checks in its climax has literally zero difference between success or failure in terms of animations, dialogue, outcomes, bonuses, or penalties).

It is a mantra of most actual play and most dungeon mastering guides, regardless of systems, to have the dice serve to amplify stakes but not define the game - only let the dice come out when the outcome is meaningfully uncertain, and when what is being rolled is clearly defined. There is a place for it, and systematizing uncertainty is a key part of what separates TTRPG from improvised narrative (check out Amber Diceless Roleplaying, though!). However, it defining all aspects of player expression is equally poor in execution - it smudges out roleplay, character building, and simple fun to have a high-level master of the craft still muck up something completely mundane - unless there is a factor in the scene to add tension, which rarely occurs in this game.

This is further compounded by the game embracing save-scumming. Unlike many, many other games in the CRPG space (and obviously unlike TTRPG, which is beholden to linear time) the player can quicksave and quickload at any point in dialogue, including on the ability check menu. The only thing stopping them from constant, eternal success is a belief that failure is interesting (it almost never is) or respect for their own time (an assumption challenged by the game’s mammoth length).

This is a sizable pacebreaker, but it’s mitigable by offering the game respect it doesn’t earn. By far one of the most frustrating and runtime-bloating occurrences is inventory management, a symptom of rough edges and ill-conceived QoL decisions colliding messily. Party member inventories are individualized, and logically are taken with them when dismissed. The player can send items to camp at any time (except for the final dungeon, for some reason, despite there being no reason to ever manage your inventory by that point?) and can similarly teleport to camp at almost any time.

These systems interlink to create a system that is fiddly (individual party members may overcap their inventory at any time, necessitating shuffling and sending to camp, and searching the inventory (an already-onerous task due to poor UX and lacking categorization) does not display items from the inventories of camped PCs) but also entirely superfluous (being able to visit camp at almost any time means the player can swap out party members or access their storage equally at almost any time). It lacks any actual difficult decisionmaking about what to bring, as combat-affecting items like scrolls, potions, and grenades weigh fractions of a pound while the limit even for STR-dumping characters is somewhere around eighty.

This has the side effect of completely eroding the feeling of camp as an actual space that inhabits the world, instead coming across as Fable 3’s inventory dimension. Despite its accessibility and lack of immersion, there is no way to quickly dismiss or replace party members past individually walking to each one (which can take 15+ seconds on larger camp maps) and mashing through dialogue. The low party limit means that there is incentive to do this a lot just to play the game and advance quests, but the completely RNG environmental skill checks means there is a want or need to swap people out for another reroll after all the WIS people in the party chunked a perception or survival save, the presence of locked doors potentially incentivizes a pocket Astarion to teleport in, jimmy a trap, and teleport away, and the ever-present horror of realizing the wrong person has a desired item and the player will thus have to cycle through everybody’s inventories one-by-one until it turns up.

The game is obviously not lacking in redeeming value - as a set of encounters it is unreal. Every single fight in the game has a unique compounding factor, and the infrequent instances when enemies are reused it is in very different compositions and contexts. The acting direction really is good, and Astarion might ultimately unseat Daeran as a new favorite in the CRPG canon of prissy assholes who prove that negging really does work. Some quests are enjoyable, even if many ultimately disappoint or get their conclusion swallowed up in the sea of bugs and inconsistent writing that is Act 3.

What it excels at, notably, has little to do with tabletop roleplay (unless your table has trained thespians) and rarely happens in CRPGs. This speaks to its broad appeal, but more notably gets at the heart of the matter: the commercial ideal Dungeons And Dragons Experience is not actually how almost any tabletop game, 5e or not, is played or performed. It is defined by secondary experience via podcasts and television episodes and broad parodies. This, more than anything, is what Larian offers: the ability to play a game that your favorite voice actors play, or to get the Dungeons And Dragons Experience when you’re not in a position to get a group going. It offers the idealized and aestheticized vision of it, even when that idealization makes the game outright worse.