Pillars of Eternity: The Infinity Engine revival

Pillars of Eternity raised close to four million dollars on Kickstarter in 2012, shipped in March 2015, and did something no marketing department could have engineered on its own: it proved there was still a paying audience for isometric, real-time-with-pause CRPGs a full decade after the format had been declared commercially dead. Obsidian Entertainment couldn’t use the Dungeons & Dragons licence, so they built their own system, Eora, from the rules up, and the result reads less like nostalgia bait and more like a studio finally getting to make the game several of its senior staff had wanted to make since working on Icewind Dale at Black Isle in the early 2000s.

Nearly ten years on, what holds up isn’t the combat, which was rough at launch and only fully found its footing in the sequel. What holds up is the writing, the setting, and a central mechanical idea — the Watcher’s curse — that ties the game’s themes directly to its systems in a way few CRPGs before or since have managed.

The Watcher mechanic makes the lore playable

The player character is a Watcher, someone who can see and communicate with the souls of the dead and the reincarnated — a condition triggered by a cataclysmic ritual gone wrong called the Sacking of Old Vailia, referenced throughout the world as ancient history the player is now uniquely equipped to investigate. Practically, this means dialogue trees regularly open unique options where the Watcher can read a dying soul’s last memory, or sense a lie in the layered way a soul’s history contradicts what someone is saying aloud. It isn’t a passive lore flourish; it’s baked into how conversations play out, so the central mystery of the game — why souls are vanishing from Eora’s cycle of reincarnation, a phenomenon called Hollowborn where children are born without souls at all — is something the player is mechanically equipped to investigate rather than simply told about in cutscenes.

This matters because it solves a problem CRPGs have wrestled with since the genre began: how does a silent, customisable protagonist earn narrative authority in a world with established lore and established NPCs. Pillars answers by making the protagonist’s specific power the lens through which the entire plot is legible, so a Watcher who ignores every soul-reading prompt is missing content, not missing flavour text.

Combat: real-time-with-pause done with modern quality-of-life

Pillars deliberately revives the real-time-with-pause combat structure of Baldur’s Gate and Icewind Dale rather than adopting the turn-based systems that Divinity: Original Sin 2 would popularise two years later. Combat runs continuously, with the player pausing to issue orders, queue abilities, and reposition before letting time resume — a structure that rewards a player who can read six overlapping health bars and status icons at once, and punishes one who can’t.

The launch version was criticised for endurance and stamina systems that made healing feel fiddly and combat pacing that dragged in mid-game dungeons; a series of post-launch patches and the White March expansions reworked encounter design considerably, and the direct sequel, Pillars of Eternity II: Deadfire, replaced the whole health model with a cleaner system. Taken as the original 2015 release, the combat is the weakest system in the game — serviceable, occasionally clever in boss design, never the reason to play it.

The Stronghold and the pacing of ownership

Caed Nua, the ruined keep the player restores across the campaign, gives Pillars a slower rhythm than most CRPGs attempt: hiring a steward, fending off sieges, and sending adventuring parties out on text-based missions that resolve while the player is off doing something else. It’s a lighter version of the base-building loop Fallout 4 would push much further the same year, and it works better here specifically because it stays optional and small — a background hum of ownership rather than a competing game bolted onto the RPG.

Companions with real, occasionally ugly, convictions

Eder the ex-soldier, Aloth the anxious wizard literally sharing his mind with another personality, Durance the priest whose faith in his god curdles across the game into something closer to trauma processing — Pillars’ companion writing avoids the trap of making every party member instantly likeable. Durance in particular is written to be genuinely difficult company for long stretches, his fanaticism played straight rather than softened for player comfort, and the game trusts that a difficult companion with a coherent internal logic is more interesting than an agreeable one with none. It’s a similar philosophy to what makes Dragon Age: Origins companions memorable — conviction over likeability — filtered through Obsidian’s house style of moral scepticism rather than BioWare’s romance-forward structure.

A class list built to reward system literacy

Pillars’ eleven classes lean hard into Eora’s own cosmology rather than reskinning familiar D&D archetypes: the Cipher channels other people’s souls to fuel psychic abilities and is functionally useless against constructs and the undead, which have no soul to draw on; the Chanter builds up a queue of sung invocations that only trigger after enough phrases have accumulated, turning a single character into a slow-building area-control engine rather than a reactive damage dealer. Multiclassing wasn’t in the original release — that came with Deadfire — so the 2015 game asks a player to commit to a single class’s rhythm for the whole campaign, which makes early-game choices carry more long-term weight than the genre usually demands.

The Cipher is the clearest example of Pillars building a class around its lore rather than around combat roles inherited from tabletop convention: because Focus, the resource that fuels Cipher abilities, only generates from dealing or taking damage, a Cipher standing at the back of a fight generating nothing is playing the class wrong, and the game nudges players towards an aggressive positioning that feels genuinely novel compared to the caster-in-the-backline default most CRPGs assume.

The White March expansions found the combat the base game was missing

The White March, released in two parts across 2015, is where Pillars’ combat design actually matures — tighter encounter scripting, a proper raid-scale dungeon in Durgan’s Battery, and the introduction of the Devil of Caroc as a boss fight built around forcing the player to solve a specific tactical puzzle rather than simply attrition a health bar. It also adds the soulbound weapons system, letting a handful of legendary items grow new abilities as the wielding character gains levels, which gives the mid-to-late game a sense of build progression the base campaign’s itemisation mostly lacked. Anyone bouncing off the base game’s combat around the midpoint owes it to themselves to push through to White March before writing the system off, because it’s a meaningfully better version of the same combat engine.

The density of the prose is a real access question

Pillars inherited more from Planescape-style CRPGs than just the isometric camera — its codex entries, item descriptions and dialogue trees are dense with invented terminology (Adra, Engwithans, the Leaden Key, animancy) delivered with minimal hand-holding, on the assumption that a player willing to read will piece the setting together. That’s a defensible authorial choice and it rewards patience, but it’s also a genuine barrier: a player skimming dialogue the way most modern RPGs are built to tolerate will lose track of the plot’s moving parts by the midgame, because unlike a Bethesda quest marker, Pillars rarely repeats critical information for a player who wasn’t paying attention the first time. It’s a design decision that trusts the audience more than it accommodates them, and whether that reads as respect or friction says as much about the player as it does about the game.

Where the ambition outpaces the execution

The game’s biggest structural problem is that its best writing is concentrated in the capital city of Defiance Bay and the companion quests, while a meaningful stretch of the midgame is spent in dungeon crawls — Od Nua’s megadungeon chief among them — that are mechanically fine but thematically inert compared to the surrounding narrative. A quarter of the playtime is spent somewhere the plot has nothing to say, which is a familiar Kickstarter-CRPG problem: backers were promised scope, and scope sometimes means content that exists because the budget stretched to it rather than because the story needed it.

Where it sits now against the games it inspired

Pillars didn’t just recreate the Infinity Engine feel; it directly seeded the wave of CRPGs that followed it. Obsidian’s own later Kickstarter-funded work carries the same DNA, and the studio’s willingness to let a game’s cosmology dictate its class design rather than the reverse is visible again in everything from Tyranny’s reputation system to the Deadfire sequel’s overhauled subclass options. Playing Pillars now, after Baldur’s Gate 3 has reset expectations for what a big-budget CRPG can render and voice, means accepting a much smaller production: sprite-scale characters on painted backdrops, a handful of fully voiced lines per companion rather than complete voiced dialogue, and load screens between areas rather than a seamless world. None of that undercuts the writing, but it does mean the game asks for patience with its presentation in a way a first-time CRPG player raised on more recent releases may not expect.

What it earns back for that patience is a setting built with unusual internal consistency — Eora’s gods, its soul economy, its Engwithan ruins, all interlock rather than existing as separate lore boxes to tick — and a willingness to let its themes (faith, colonialism, the ethics of managed belief) sit unresolved rather than resolved for player comfort. That’s a rarer commodity than good combat, and it’s the reason Pillars is still worth returning to nearly a decade on rather than filed away as a Kickstarter curiosity that time overtook.

Spoilers below

The endgame reveals that the Hollowborn crisis was engineered — Thaos, the immortal priest manipulating events from behind a dozen aliases across the campaign, deliberately caused the soul-severing catastrophe centuries earlier and has been managing its consequences ever since to preserve a religious order built on a controlled lie about the gods’ true nature. The player discovers, through the Watcher’s own soul-reading, that the gods of Eora are constructs built by an ancient civilisation called the Engwithans, engineered rather than the cosmic beings their worshippers believe — a reveal the game handles with real care, letting NPCs react to the news with everything from denial to relief rather than treating it as a simple twist.

The final choice — whether to reveal this truth to the world, understanding it will likely collapse the social order built on faith in the gods, or suppress it to preserve stability — doesn’t resolve into an obviously correct answer, and the epilogue slides play out consequences on both branches that are uncomfortable rather than triumphant either way. It’s a quieter, more melancholy ending than the genre typically reaches for, and it’s the strongest evidence that Obsidian used the Kickstarter money to build the CRPG they actually wanted rather than the one focus-tested demand would have produced.