Pillars of Eternity

46.5% of everyone who owns Pillars of Eternity has completed Act I. About 27% have finished Act II.

For a game with an 88% review score, that's a startling pair of numbers, and they're the most honest description of this game I can give you. Pillars is very good and it is very hard to start.

What Obsidian actually built

The pitch in 2012 was straightforward: give the people who made Fallout: New Vegas and the Infinity Engine classics enough money to make a proper isometric party RPG again, without a licence dictating the rules.

What came back was a world rather than a homage. Eora runs on souls — they're real, they're measurable, they reincarnate, and the setting's central crisis is a theological and biological one about where they go and who has been interfering. Your character can see them. That isn't flavour draped over a fantasy kingdom; it's the engine of the plot, and it produces a story that could not be told in a Forgotten Realms game.

The systems are similarly deliberate. Six attributes that apply to every class, an unusually clean health-and-endurance split that makes attrition matter across a dungeon rather than a fight, and classes with genuinely distinct rhythms. Building a party here is a planning exercise, and the game respects you enough not to explain the interactions.

The stronghold gives the middle of the campaign somewhere for money to go, and the hand-painted backgrounds have aged conspicuously better than the 3D RPGs of the same period. Median recent playtime is eighty hours; one recent reviewer has logged over 2,400.

The systems are the reason to stay

It's worth being specific about what Pillars does mechanically, because "deep CRPG systems" is a phrase that hides the interesting part.

Health and endurance are separate. Endurance is the per-fight pool that drops and regenerates; health is the long pool that only recovers when you rest, and rests are limited by camping supplies you have to buy or find. So winning a fight isn't binary. You can win five encounters and still be forced out of a dungeon because your front line has bled health down to nothing, and pushing one room further is a genuine gamble. It turns a dungeon into a campaign rather than a series of independent battles — the same attrition logic Darkest Dungeon built its whole identity on, arriving here first and with more nuance.

Abilities are split per-encounter and per-rest. Your reliable tools come back every fight, so you're never reduced to plinking with a crossbow because you spent your good spells an hour ago — but the powerful ones are still finite across an expedition. That single division solved the oldest pacing problem in the genre, and modern RPGs have been quietly copying it ever since.

Engagement makes position matter. Melee characters lock enemies in place, and breaking away earns a free hit. It sounds small; it's what stops fights collapsing into a scrum and makes a front line a real thing you build and maintain rather than a formation you draw once.

Put those together and combat rewards preparation over reaction: the right party composition, the right opening, an understanding of what your durability budget looks like across the next hour. That is a genuinely different pleasure from the tactical puzzle of a turn-based game, and it's why people with two thousand hours are still writing reviews of a decade-old RPG.

None of it is explained well. All of it is worth learning.

Why so many people stop

Three reasons, and they're all in the reviews.

The text. This is the most common substantive complaint and it's fair. Pillars introduces an enormous amount of world material — peoples, gods, nations, wars, a whole glossary of terms — early and fast, before you have any reason to care. One reviewer describes it as reading an amateur fantasy novel's appendix; another, who finished the game, says the volume of dialogue and lore is precisely why they'll never replay it. The writing is good. There is simply too much of it arriving too early.

The comparison. Almost every negative review measures this game against Baldur's Gate 2, and it was made to invite exactly that. Reviewers say the world feels smaller, the companions less memorable, the choices less consequential. One calls it a rushed homage rather than a spiritual successor; another diagnoses a fanservice-induced identity crisis, which I think is the sharpest formulation available. Pillars is caught between reproducing a twenty-year-old game and being its own thing, and it doesn't fully commit either way.

Might. The attribute complaint recurs for a decade. Might governs both physical damage and spell power, so your wizard wants the strength stat. There's a coherent design argument — it's soul force, not muscle, and it stops casters dumping physical attributes — but it fights every instinct a D&D-trained player has, and reviewers with a hundred-plus hours still cite it as the thing that made the system feel arbitrary.

The update that changed the answer

Here's the practical thing a 2026 review can tell you that a 2015 one couldn't: there's a turn-based mode now, added long after release.

Real-time-with-pause is the genre's biggest filter. It asks you to manage six characters in continuous time by pausing constantly, and plenty of people who love tactical combat simply cannot enjoy it. One recent reviewer says outright that they bought the game years ago, never got into it because the combat felt as bad as old Baldur's Gate and Icewind Dale, and fell in love with it once turn-based mode existed.

If that's you — and given how many people bounce in Act I, it's some of you — this is a different purchase than it was ten years ago.

8.0 out of 10. Rated 8.0 out of 10.

Pillars of Eternity did the thing it was funded to do: it proved the isometric party RPG was commercially alive, and everything since — Divinity: Original Sin 2, Baldur's Gate 3, a decade of smaller CRPGs — walked through the door it opened. Eora deserves its reputation, the systems are excellent, and eighty hours is a lot of game for the money.

It doesn't reach the top tier because a game whose first act loses more than half its buyers has a structural problem, not a taste problem. Front-load less lore, introduce the world through incident rather than glossary, and this is a 9. As it stands: superb once it opens up, and you have to mean it for the first ten hours.