Grim Fandango: The Adventure Game's Last Great Argument

Grim Fandango shipped on 30 October 1998. That autumn also gave us Half-Life, Metal Gear Solid, Ocarina of Time, Thief: The Dark Project, Baldur’s Gate and Resident Evil 2. Every one of those games was an argument for where the medium was going next. Grim Fandango was an argument for where it had already been, made with more craft and more money than anyone had ever spent on the case, and it landed in shops the way a beautifully written letter lands in a house that has already been sold.

It sold badly. LucasArts kept the lights on in the adventure division for one more game and then stopped. Sierra’s adventure line went the same way inside a couple of years. So the temptation is to file this as a tragedy — the last good thing, killed by shooters — and move on. That reading is lazy, and it does the game a disservice. What Grim Fandango actually is: Tim Schafer working out, in public, what the adventure game had been getting wrong for a decade, fixing about seventy per cent of it, and breaking something new in the process.

The premise is a filing system

Manny Calavera is a travel agent for the dead. In the Department of Death in El Marrow, the recently deceased arrive and are sold a travel package for the Four Year Journey of the Soul — the walk to the Ninth Underworld. Live well and you have earned the Number Nine, a luxury train that does the trip in four minutes. Live badly and you get a walking stick and a compass and a very long road. Manny works commission. Manny’s clients keep coming up walking stick.

That is one of the great setups in games, and it is doing several jobs at once. It is a joke about bureaucracy annexing the afterlife. It is a noir engine — a working stiff, a rigged system, a client who deserved better. And it is a moral scoring system that the game refuses to let you see the workings of, which becomes the plot when Manny discovers the numbers are being cooked.

The visual grammar is a three-way fusion that ought to collapse and doesn’t. José Guadalupe Posada’s calaca skeletons give you the cast: everyone is a cheerful bone figure in a suit, which means death is universal, comic and completely undramatic. Art deco gives you the world: El Marrow and Rubacava are all setbacks and zigzags and chrome, a 1930s idea of the future built for people who no longer have one. Film noir gives you the camera and the plot. Peter McConnell’s score sits on top playing big-band swing, bebop and Mexican folk with a straight face.

The reason the fusion holds is that all three sources agree on one thing: style as consolation. Deco, noir and Day of the Dead are each a way of dressing up an ending. Put them in a room and they are talking about the same subject.

Four years is the structure nobody copied

Here is the part that has aged best, and the part almost nobody mentions.

Grim Fandango is split into four acts, and each act is a year. Year One ends, and the game cuts to Year Two, and a year of Manny’s afterlife has happened without you. He is somewhere else. He is doing something else. He has a business, or a boat, or a problem you have to be told about. The people you knew have moved, changed jobs, given up on him.

Adventure games before this were spatial. You were given a map and you exhausted it. Grim Fandango is temporal: it gives you a map, you exhaust it, and then it takes it away and hands you a different one containing three people from the last one, older. The Year Two opening is the single best thing in the game, because you arrive expecting to resume a chase and instead discover that Manny stopped chasing about eleven months ago and built a life out of waiting.

That structure buys the game its emotional range for free. A four-year gap does character work that no dialogue tree can. It is also brutally efficient design: each year is effectively a self-contained game with its own cast, its own locations and its own puzzle vocabulary, which means the pacing can reset four times and never sag the way a thirty-hour flat world sags. Full Throttle, three years earlier, had already been reaching for cinematic act structure and ran out of game before it ran out of ambition. Four years is the answer to that problem. It is the closest an adventure game has come to the shape of a novel, and I cannot name a modern narrative game that has taken it.

The engine gave, and the engine took

LucasArts retired SCUMM for this. GrimE — the Grim Edition engine — was built on the technology behind Jedi Knight, with the game logic scripted in Lua, and it put real-time 3D character models on top of static pre-rendered backgrounds. In 1998 that was the compromise everyone was making, and it was the right one: the backgrounds could be as detailed as an artist could paint, and the characters could move and turn and act.

What it cost was the cursor.

SCUMM games ran on a mouse pointer, and the pointer is an interrogation device. You sweep it over a room and the room answers. Day of the Tentacle is legible in about four seconds per screen because of it. Take the pointer away and the player has no way to ask a room what is in it.

Schafer’s team knew this and built a replacement: Manny turns his head. Walk him around a location and his skull swivels towards whatever he can interact with, and the interaction verbs are attached to him rather than to a menu. It is an elegant idea. It is also a much lower-bandwidth channel than a cursor, because you can only ask about the place you are standing, and standing everywhere is a chore.

Then there are the tank controls. You steer Manny relative to his own facing — press up and he walks forwards, press left and he pivots. On a keyboard, in rooms with fixed cinematic camera angles that cut as you cross a threshold, so that the direction you were holding becomes the wrong one mid-stride. Resident Evil had shipped this scheme two years earlier, and it worked there for a reason that does not apply here: in a survival horror game, clumsy movement under a bad camera is tension. In an adventure game it is just admin. You are not being stalked. You are trying to look at a table.

The 2015 Double Fine remaster bolted point-and-click movement back on, and the game is straightforwardly better with it, which is the whole indictment. The interface was a tax the design was paying to the engine.

The puzzles argue with themselves

The reputation is that Grim Fandango is unfair. That is roughly half right, and the half matters.

The LucasArts ethic held: you cannot die by accident, you cannot lock yourself out of a solution, and you cannot walk into an unwinnable state — a discipline the studio had been enforcing since Monkey Island and which Day of the Tentacle turned into an art form. Grim Fandango keeps all of that. Nothing here punishes you with a lost hour.

What it does instead is hide the question. Several of the celebrated stumpers are perfectly logical once you know what you are being asked, and the failure is that you never knew you were being asked. The inventory is Manny’s coat: you flip it open and page through his possessions as physical objects, which is a lovely bit of characterisation and a genuinely worse list than a grid of icons, because you cannot see everything at once and puzzle-solving is pattern-matching across a set you can see. Combine a coat you have to leaf through with a room you have to walk every inch of, and you get a specific 1998 feeling: knowing exactly what to do and not being able to find where to do it.

The good puzzles are very good, and they are good in a way that is inseparable from the writing. The best of them turn on the world’s own bureaucratic logic — the rules of a system that has an afterlife of paperwork and a black market underneath it — so that solving one teaches you something about how the place is corrupt. That is the technique the genre spent twenty years trying to name: a puzzle should be an argument about the world, and the solution should be evidence.

What it was actually arguing

The wordless-cinema tradition that Another World started in 1991 said games could be films by removing everything a film cannot do. Grim Fandango takes the opposite position and says games can hold a real script — a hundred-plus characters of dialogue with rhythm and cruelty and jokes that land four hours after the setup — if you build a structure strong enough to carry one. Tony Plana gives Manny an actual performance, timing and all. Glottis, the enormous orange demon who exists to drive and grieves when he cannot, is a better-written supporting character than most films manage.

The genre died anyway, and it died for reasons that had little to do with quality: the shelf space went to 3D, the audience went with it, and the adventure game’s economics — enormous art costs, no replay value, a runtime you burn through in a fortnight — stopped adding up. Grim Fandango was the most expensive possible restatement of the case at the exact moment the case was lost.

Which is why it is still worth playing. Play it in the remaster, with the mouse controls on, and skip the purist argument entirely. What you get is a four-act structure nobody has bettered, a world with a coherent moral engine underneath the jokes, and a demonstration of the thing the adventure game was uniquely good at: a place you learn by asking it questions. Twenty-eight years on, the answer the industry gave was that this was a dead end. The game itself is the counter-evidence, and it is still sitting there, unrefuted.

Spoilers below

The reason the four-year structure works is that Manny fails, repeatedly, on a schedule.

Year One is a heist: he steals a client, Meche Colomar, from his rival Domino Hurley, discovers she should have had a Number Nine ticket and did not get one, and runs. Year Two is where the game becomes something else. He is in Rubacava, the port town at the edge of everything, running the Calavera Cafe, and he has been there for a year because he missed her. The whole act is a man waiting in a coastal town for a boat that already left. That is the noir move — the plot stops and the character keeps paying.

Hector LeMans, the villain, has been intercepting Number Nine tickets and selling salvation. The moral system Manny worked for was rigged before he arrived, and his own career of shifting walking sticks to people who deserved trains was the fraud’s delivery mechanism. He was not a bystander. The four-year journey he has been selling is a punishment invented by a crook.

The sprouting is the game’s cruellest invention: a sprout gun fires seeds into a skeleton and flowers bloom out through the bone. It is death for the dead, and it is played as horticulture. Salvador Limones — the fanatic running the Lost Souls’ Alliance out of the sewers, all hat and no jaw — goes that way, and the game lets him make his speech while he goes.

Year Four is the Temple, and the ending is the Number Nine finally leaving with the right people on it. What earns it is four years of Manny doing the arithmetic: he cannot undo the tickets he sold, he cannot give those souls their four minutes back, and the best available outcome is that he stops working for the thing that took them. Then he gets on the train himself, which the game has spent forty hours convincing you he has not earned, and the last shot lets you decide whether he has. The game declines to say either way, and lets the platform empty out.