The detective genre in games has spent thirty years trying to make deduction happen and mostly producing its opposite. You know the pattern: you walk into a room, press the button on every glowing object, and once the counter reads 6/6 the detective announces the solution he worked out without consulting you. The game calls this an investigation. What it actually is is a search-and-collect with a lecture at the end.

The Case of the Golden Idol, released in October 2022 by the small Latvian studio Color Gray Games and published by Playstack, does the obvious thing that almost nobody does. It gives you the evidence and then makes you say what it means. If you’re wrong, it says no. It does not say why.

That “it does not say why” is the entire product.

The mechanism

Each of the eleven scenes is a single tableau: a frozen moment, hand-drawn in a style somewhere between Hogarth and a bad dream, populated by grotesques mid-crime. Somebody is falling off a cliff. Somebody is being poisoned. You click around the scene — pockets, letters, ledgers, signage, faces — and every clickable thing yieldswords. Names. Occupations. Verbs.

The words go into a bank. Then you open the thinking panel, which is a page of sentences with holes in them, and you drag words into holes until the sentences describe what happened. Who is who. Who did what to whom, and with what, and why.

That’s it. There’s no dialogue. There’s no interrogation, no timeline scrubber, no notebook that fills itself in. Two verbs: look, and assert.

Why the word bank is smarter than a dialogue tree

Here’s the design problem every detective game hits. Deduction is internal. It happens in a head. To make it a mechanic, you have to externalise it, and the moment you externalise it you risk turning “I worked it out” into “I picked the right option from three”.

The word bank solves this by making the answer spacecombinatorial and hostile. When a puzzle has forty available nouns and eleven slots, brute force isn’t a strategy — it’s a punishment. You can’t guess your way through, because the possibility space is too wide to walk and too narrow to fluke. So you’re pushed back into the only remaining approach: actually thinking about it.

And the game refuses to grade partially in a way that would let you triangulate. This is where it separates from its most obvious relative.Return of the Obra Dinn confirms your fates in batches of three, which is a genuinely brilliant compromise — it stops the game being unwinnable while making you commit to trios. It also means a canny player can farm it: lock two you’re sure of, cycle the third.Golden Idol declines the compromise. Submit an imperfect answer and you learn that it’s imperfect, and you go back to the tableau with your ego intact and your theory in pieces.

The result is that the moment of solving is undiluted. Nothing helped you. The game withheld everything except the facts, and the facts were sufficient, and you found them sufficient. I can’t think of a cleaner delivery of that feeling in the medium.

The other trick: the story is in the ledger

The eleven cases run across decades, and the plot — a cursed golden idol, an inheritance, a family, a great deal of murder — is never narrated to you. It’s assembled from the same nouns you’re using as puzzle pieces. You learn the dynasty’s shape because you keep filling in surnames. You work out the political situation because a scene requires you to identify who signed a document.

This is a genuinely rare thing: exposition that costs the player effort and therefore sticks. Nobody remembers a cutscene. Everybody remembers a name they had to earn. It’s the same economicsTunic runs when it makes the manual pages both the lore and the solution, and it’s why both games feel dense at a fraction of the word count of a proper RPG.

The art carries more of this than it gets credit for. The figures are ugly on purpose — pop-eyed, jowly, caught mid-gesture — and the ugliness is functional, because you need to distinguish nine strangers at a glance across ten scenes with no name tags. A realistic style would have made them a soup. Caricature is a legibility tool that happens to also be a tone.

Where it fights itself

Two honest complaints.

The scenes are static, which means the tableau has to carry both the puzzle and the drama, and occasionally the drama loses. A frozen frame is a fantastic puzzle substrate and a limited storytelling one, and a couple of the mid-game cases feel like admin — identify eight people at a party — rather than a crime you care about.

And the difficulty is uneven in the way hand-built puzzle games always are. Most of the eleven land beautifully. One or two hinge on a single obscure noun in a corner, and if you don’t click that corner you’re not stuck on logic, you’re stuck on pixel hunting, which is a different and worse kind of stuck. The game has a hint system for exactly this, and using it feels like a small defeat, which is arguably correct and definitely annoying.

The 2023 DLC chapters — The Spider of Lanka and The Lemurian Vampire — are tighter than the base game on both counts, which is a good sign about what the studio learned. The 2024 sequel,The Rise of the Golden Idol, moves the whole apparatus forward a couple of centuries and adds quality-of-life the original lacked.

The bit about being wrong

I want to dwell on failure, because it’s the least discussed part of this design and the most radical.

Modern games treat a wrong answer as a UX problem. Something must happen: a hint surfaces, a difficulty slider quietly nudges, an NPC wanders over to helpfully observe that the lever looks operable. The industry spent twenty years engineering frustration out, and in the process engineered out the state that precedes insight. You can’t have the click if nothing was stuck.

Golden Idol lets you be stuck. Properly, unproductively, for a quarter of an hour, staring at a picture of a man in a wig. And the reason this is tolerable rather than infuriating is a quiet piece of craft: the scene is always complete. Everything you need is on screen. There’s no second location, no locked area, no character who’ll say the missing thing on Tuesday. So when you’re stuck, you know with certainty that the failure is comprehension. That certainty is what makes persistence rational.

This is the oldest lesson in the medium and it keeps getting mislaid. The C64 adventures I grew up on were frequently stuck-forever affairs, and the good ones differed from the bad ones on precisely this axis: whether the puzzle was closed. A closed puzzle you can’t solve is a challenge. An open one is a guess. Color Gray have simply remembered which is which.

The verdict

Golden Idol is the rare game that respects you by ignoring you. It won’t encourage you. It won’t nudge. It has no interest in your session length or your completion funnel. It puts a horrible little painting in front of you and waits.

The genre lesson underneath it is worth naming: detective games have been adding features — timelines, reconstructions, deduction boards with animated string — when the missing ingredient was always subtraction. Take away the confirmation and the thinking arrives on its own. Every mechanic Color Graydidn’t build is why the one they did build works.

It’s on PC, Switch, PlayStation, Xbox and phones. The phone version is better than it has any right to be — the whole game is clicking and dragging, and a tableau sits fine on a tablet. Play it in single-case sittings with a real pen if you’re that way inclined. Most people won’t be. Most people will find they need to be by case seven.

Where next:Obra Dinn is the sibling and the better game overall, though not the purer one. If you want deduction with an actual world to walk around in, and a game that will happily let you be catastrophically wrong,Paradise Killer is the other end of the same argument.

Spoilers below

The idol itself is the best-kept structural joke in the game. For eleven scenes you’re doing forensic work on a series of murders, and the object motivating all of them has a power that is never explained by any mechanism and never needs to be, because the game has correctly identified that its supernatural MacGuffin is doing zero puzzle work. The idol is a reason for people to be greedy. Greed is legible. Curses are furniture.

The dynasty structure — the way the same family line keeps regenerating the same crime across generations — pays off because you built the family tree yourself, one dragged surname at a time. When the last case asks you to name a relationship you established four scenes ago, it’s checking whether you were investigating or just solving. Those turn out to be different activities, and it’s the only game I know that can tell the difference.

In 1991 I could not play Monkey Island without the cardboard wheel. You’d line up the pirate’s hair with the pirate’s chin, read off a date, type it in, and the game would let you past. Lose the wheel and you owned a coaster. Every Amiga in the country had a shoebox of this stuff: code wheels, Lenslok, and the lookup tables that made you find page 14, line 3, word 6 in a manual you’d otherwise never open.

Publishers did that to stop people copying disks. What it actually did, by accident, was make the manual part of the machine. The game was on the disk and the game was also on the kitchen table, and you couldn’t run one without the other.

Tunic is what happens when someone takes that accident seriously.

What it is

Andrew Shouldice released Tunic in March 2022 after roughly seven years of work, published by Finji. It came to Xbox and PC first and reached PlayStation and Switch in September of the same year. It has spent the time since drifting through subscription services and sales, which is how most people meet it now — a small isometric action-adventure with a fox in a green tunic, obviously wearing the first Zelda’s clothes.

That’s the trap. It’s dressed as an homage and it’s actually an argument.

The manual

Scattered through the world are pages of an instruction booklet. Not lore fragments — an actual manual, laid out like something that fell out of a 1987 box: glossy illustrated spreads, a map, diagrams, a bestiary, arrows pointing at buttons. You collect the pages and you can open the booklet at any time, and the booklet is where the game keeps everything it hasn’t told you.

The manual is almost entirely written in a language you cannot read.

There’s an invented script — Trunic — running through every page, and the first time you open it your eye slides off it and lands on the pictures. Which is the point. You start reading the manual the way a nine-year-old reads a manual for a game they don’t own yet: guessing from diagrams, inferring from arrows, building a theory of what the game must contain out of pure iconography.

And then a page teaches you something real. There’s a spread that shows you how to dodge-roll, and until you find it, you do not dodge-roll — the button worked all along, and you simply did not know the verb existed. Later there’s a page that shows you a mechanic so fundamental that finding it retroactively re-explains everything you’ve been walking past for six hours.

This is progression made out ofknowledge rather than items. The character never gets a new ability. You do.

Why it works

Metroidvanias gate you with objects, and objects are honest but inert: the game withholds the double jump, gives you the double jump, and the space of what you can do expands by exactly one predictable increment. You knew the double jump was coming. You’ve played this before.

Knowledge gating has a different shape. When Tunic hands you a page, your entire back catalogue of memories re-sorts at once. Every strange wall, every suspicious statue, every geometric thing you clocked as decoration — the page doesn’t open one door. It opens all the doors of that type, everywhere, retroactively, and the game didn’t have to build a single new room to do it.

That is enormously efficient design, and it’s also the reason the game can’t be patched into being easier. You cannot hint your way around it, because the thing being withheld isn’t in the save file.

The related trick is that Trunic is not decoration. It’s a real cipher — a consistent mapping to English phonemes, learnable, and people did learn it, sitting down with the pages and cracking the script like a philology homework. The manual is fully readable if you do the work. Shouldice built an entire functioning writing system and then made almost nobody need it, which is the single most confident act of restraint in the medium.

Chants of Sennaar went at the same problem from the front, making the decipherment the loop and giving you a notebook to be wrong in. Tunic buries it and lets you walk past. Both work. Sennaar is the better teacher; Tunic is the better ambush.

The manual as an artefact

Look at the pages themselves and you find the second layer of the joke. They’re faithful to a specific era of print — the slightly off registration, the airbrushed box-art idiom, the bilingual clutter, the way European manuals crammed six languages into a booklet none of us read. There are hand-scrawled annotations in the margins in biro, because of course there are; every used manual in every second-hand game I ever bought had somebody’s map of level three in the back.

Those annotations do heavy lifting. They’re the previous owner. Somebody was here, they figured some of this out, and they left you circled hints in a hand that isn’t the manual’s. It gives the game a social texture without a single line of multiplayer code — the same trick Souls messages pull, achieved with a pen.

None of this is nostalgia bait, and I’m allergic to nostalgia bait. The point isn’t that manuals were nice. The point is that manuals were asecond information channel the game couldn’t see, and when that channel died — when everything moved in-game, into tutorials and tooltips and quest markers — designers lost the ability to withhold. If the game must teach you everything it can do, then everything it can do is a checklist. Tunic reopened the channel and immediately used it to lie to you.

Where it fights itself

The combat is the weak link and always was. It’s a slow, stamina-gated, roll-and-poke system with a shield, and it wants to be taken seriously enough to be tuned but loose enough to be forgiving, and it settles in an unhappy middle. The bosses have real teeth and the moment-to-moment fighting doesn’t have the precision to make that teeth-baring feel fair. There’s an option to switch off damage entirely, and I don’t think it’s a defeat to use it; the game’s actual content is above the neck.

The isometric camera is the other tax. Tunic hides things behind geometry deliberately — an entire class of secret depends on a path being invisible from your fixed angle — and this is genuinely clever the first six times. After that it’s a game where you occasionally walk into walls hoping. Depth ambiguity as a puzzle mechanic has a low ceiling, and the game finds it.

And the mid-game asks you to do a lot of running. The world folds beautifully and the shortcuts open, and there’s still a stretch around the halfway mark where you’re crossing three biomes because a page told you something and the thing it told you about is a long way away.

The verdict

Tunic is a game about the pleasure of not being told, and it holds a position almost nothing else in the medium is willing to hold: that the player is capable of figuring it out and will enjoy the figuring more than the finding. It spends its first hours letting you believe it’s a small polite Zelda tribute so that the betrayal has somewhere to stand.

Play it anywhere — it’s on everything now, and it’s light enough to run on a toaster. Play it with a notebook and a pen, actually, and be ready to be embarrassed by how long it takes you to notice you’re allowed to draw things.

The pairing I’d suggest isReturn of the Obra Dinn, which runs the other great knowledge-progression system of the era, andInscryption if what you liked was being lied to about what kind of game you’d bought.

Spoilers below

The Holy Cross is the whole thesis, and it’s the best thing I’ve seen a game do with an input.

There is no new item. There’s a sequence of directional presses, and the manual has been showing you where to use it the entire time — hidden in the page borders, in the decorative frames, in the fold. The instruction was in your hands from the first page you picked up. You looked at it a hundred times. It was printed on the paper.

And so the game’s real final ability isliteracy. Once you see the border code, you go back through every page you’ve collected and read the game’s own documentation as a walkthrough, which is precisely the ritual the code wheel and the lookup table trained a generation to perform — go to the manual, find the page, read off the answer, come back. Shouldice took the most hated piece of 1980s anti-piracy friction and rebuilt it as the reward.

The two endings sharpen it. You can beat the Heir with a sword, which is the answer the game’s combat has been training, and it’s the lesser ending. Or you can gather the pages, understand what the fox has been doing to the previous heirs, and end it another way entirely — an ending available only to a player who read the paperwork. One route is reflexes. The other is attention. The game knows exactly which one it thinks is worth more, and it never once says so out loud.

There is a specific feeling you get about ninety minutes intoChants of Sennaar, and I’ve been trying to work out how a puzzle game manufactured it. You are standing in front of a person who has been shouting the same glyph at you since you arrived. You have finally worked out what it means. And the glyph, it turns out, meanswelcome.

Rundisc’s game came out on 5 September this year, published by Focus Entertainment, and it is built on the Tower of Babel — five peoples stacked up a single structure, each speaking a language the others have forgotten, each convinced the floor above them is a threat. You climb. The only thing you carry is a notebook.

How the validation page works

The mechanic is simple to describe and much harder to build. You see glyphs — on signs, in speech bubbles, carved on doors. You have a journal, and against each glyph you write your best guess at the meaning, in plain French or English or whatever the game is running in. You can write anything. Nothing is checked.

Then, periodically, the journal produces a page with several illustrated panels on it and a slot beside each. Drag the right glyph onto the right drawing for every panel on the page, and the page locks: those meanings are now confirmed, and from then on the game silently prints them as words wherever they appear.

That page is the entire design. Three things fall out of it.

It defeats guessing without punishing being wrong. A page only validates as a set, so you cannot brute-force one slot at a time and read the tick. Getting five of six right tells you nothing, in the same way a nearly-correct hypothesis tells a scientist nothing. But a wrong guess in your notebook costs you nothing either — no penalty, no lockout — so you’re free to be confidently wrong for hours, which is how actual decipherment works.

It separates two different pleasures. Working out that a glyph means “door” is one kind of fun: you noticed it above every doorway. Working out that a glyph means “forbidden” is a much better kind: you noticed it abovesome doorways, and beside a guard, and on a sign the guard was pointing at while making an unmistakable gesture. The page validation lets both count, and then rewards the second by turning a wall of nonsense into readable prose.

It makes the world the dictionary. No character teaches you anything. The language is only ever learned from context — a hand gesture, a repeated juxtaposition, the layout of a room, the fact that this word appears exclusively in a kitchen. Rundisc had to author every glyph’s meaning into thelevel design, which is a properly enormous amount of invisible work, and it’s why the tower feels lived-in. Every prop is a lexical entry.

Why the translation reads as empathy

The Warriors are the point at which the game reveals its argument, and it does it without a line of dialogue you could quote.

When you first reach their floor you cannot read a syllable of it. They are armoured, they shout, they hunt you through corridors in genuinely tense stealth sections, and they are — obviously, self-evidently — hostile. That’s not an interpretation. That is the raw sensory data.

A floor’s worth of glyphs later, you can read the signs on their walls. And the signs say things about duty and about fear of what’s above. The stealth section doesn’t change. The level layout doesn’t change. The soldiers still chase you. What changes is that you now understand that they are frightened people doing a job in a building whose upper floors they believe are full of monsters, and you have just come down from those floors, and you are the monster.

The game achieves this with zero exposition, because it made you do the work. This is the whole reason “empathy machine” fits here rather than being marketing: comprehension arrives asyour achievement, at a moment ofyour choosing, and it retroactively rewrites everything you already saw. A cutscene could tell you the Warriors are afraid. Only a language puzzle can make yourealise it.

The later floors take it further, when the game starts asking you to translatebetween the peoples — using one language you’ve cracked to bootstrap another, and eventually acting as an interpreter for two groups who have spent generations inventing reasons to hate a noise they can’t parse. The mechanic and the theme are the same object. That’s the highest thing a game system can do and it happens about four times a decade.

Where it fights itself

The stealth is the seam, and everyone says so for good reason. The Warriors' floor and a later sequence swap deduction for patrol routes and hiding spots, and Rundisc are plainly better at semiotics than at line-of-sight cones. It’s not punishing — a fail sends you back a few metres — and it is a different, lesser game wedged into a great one, presumably because someone worried the middle act needed a pulse. It didn’t.

The second issue is subtler. Because the validation pages arrive at authored moments, the game’s difficulty is partly a function ofwhich meanings Rundisc decided to check. There are glyphs you will crack instantly and glyphs you’ll carry for two floors, and occasionally a page will ask you to distinguish two concepts on evidence that is thinner than the rest of the game’s standard. Those moments feel like a design slip in a piece of work that’s otherwise extraordinarily precise about what it has shown you.

And it’s short — a weekend — with no second run in it, since you can’t un-know a vocabulary. That’s the same taxObra Dinn pays, and it’s the correct price for both.

The ancestors

Heaven’s Vault (2019) is the obvious comparison — Inkle’s game also has you translating an ancient script — and the two differ in one instructive way.Heaven’s Vault lets you be wrong and keep walking; it will accept your bad translation and build on it, and the ambiguity is deliberate and, for a lot of players, maddening. Sennaar validates. You always eventuallyknow. Inkle made a game about interpretation; Rundisc made a game about decipherment, and decipherment has a right answer, which is why Sennaar’s dopamine hits harder.

But the real ancestor isCaptain Blood, 1988, and I say that with the confidence of someone who owned it. ERE Informatique’s oddity — on the Amiga, the ST and the C64 — put you in a spaceship talking to aliens through UPCOM, an icon language of pictograms. You built sentences out of those symbols and the aliens replied in pictograms, and the whole game was a negotiation conducted in a vocabulary you had to assemble by trial. It was unfriendly and slightly broken and completely unforgettable, and it understood in 1988 the thing Sennaar understands now: that the moment a stranger’s grammar clicks is a bigger event than any firefight.

Tunic is the modern cousin — an invented script and an in-game manual you learn to read — though Tunic’s language is mostly a wrapper on its secrets, where Sennaar’s languageis the secret.

The verdict

Chants of Sennaar is the best thing this genre has produced sinceObra Dinn, and it gets there by a completely different route: where Pope built a forensic machine, Rundisc built a social one. The validation page is a small, elegant answer to the verification problem, the art — cel-shaded, Moebius-flavoured, all flat colour and clean line — does real work in making glyphs legible as glyphs, and the central trick of making comprehension feel like forgiveness is genuinely new.

The stealth is a wart. Ignore it; it’s over in twenty minutes and the game on either side of it is close to flawless.

It’s on PC, Switch, PlayStation and Xbox, and the Switch version is the one I’d choose, because this is a game you play with a real pen and paper next to you and a handheld leaves the desk free. Guess wildly, and be wrong in ink.

If you want another game where the systems carry the argument instead of the script,Cocoon came out a fortnight ago and has no words in it at all.

Spoilers below

The Anchorites are where the structure pays off. By the time you reach the top of the tower you are carrying four vocabularies, and the final floors stop being about learning a fifth and start being aboutarbitration — the game hands you the ability to move meaning between peoples who cannot address each other, and every puzzle after that is a diplomatic act.

It is also Babel told properly. The myth’s usual reading has the confusion of tongues arrive as a punishment dropped from above; Sennaar puts the whole catastrophe on the ground floor, where it belongs — generations of people directly above and below each other, each treating an unparsed noise as a threat and building a theology out of the misunderstanding. Rundisc never state that. They let you assemble it, glyph by glyph, and then hand you the job of undoing it.

And the descent at the end is the finest thing in it. You go back down through floors you crossed in fear, and you can read all of it, and the tower that spent the whole game being a hostile puzzle box is now simply a building full of people who have been shouting explanations at you from the start.

Almost every detective game lies to you about what detection is. It gives you a magnifying glass, highlights the clue in yellow, plays a sound when you’ve found enough of them, and then has a character explain the conclusion you were never allowed to reach. The player’s job is attendance.Return of the Obra Dinn, made almost entirely by Lucas Pope and released on PC on 18 October 2018 with console versions exactly a year later, is the game that finally handed the job over.

Five years on it has no real rival, and the reason is a single rule about how it checks your homework.

The rule of three

The setup: it’s 1807, you’re an insurance adjustor for the East India Company, and a ship that sailed in 1802 and vanished has drifted back into Falmouth with nobody aboard. You have a ledger with sketches of all sixty souls who sailed, and a pocket watch — the Memento Mortem — that lets you touch a corpse, hear the last few seconds before it died, and then walk around a frozen three-dimensional tableau of the instant of death.

For each of the sixty, you must fill in three things: who they are, what happened to them, and, where relevant, who did it. Get all sixty right and you’ve finished. The obvious question is why you can’t simply guess. Sixty names, a couple of dozen fates — a determined idiot with a spreadsheet could brute-force it in a weekend.

Pope’s answer is the best design decision in the game. The book confirms nothing until you havethree fates correct simultaneously, at which point it locks all three in and says so. Nothing else. No per-entry tick, no “warmer”, no partial credit.

Sit with what that does. Guessing is now useless, because a wrong guess is indistinguishable from a right one until it’s part of a correct trio, and the combinatorics of finding three simultaneous correct answers by accident are hopeless. Meanwhileactual deduction is rewarded generously, because the moment you nail a cluster — the three Formosan passengers, say, or a knot of Russians who only ever appear together — the game confirms them and hands you a foothold.

The rule solves the deduction game’s oldest problem: how do you verify without giving the answer away? Every game before this either verified per-item (so you grind guesses) or verified only at the end (so a single error hides in eighty hours of work and you never find it). Three-at-a-time is a checksum. It tells you that abody of reasoning is sound without telling you which part of it did the work.

The tableau is a document

The Memento Mortem scenes are the other half. You hear a few seconds of audio — often a shout, a name, a foreign language you don’t speak — and then you’re standing in a still photograph you can walk around.

The craft here is in what Pope refuses to do with the camera. You are never shown anything. The tableau has no framing and no emphasis; it is a volume of frozen space, and every fact in it has been left lying about at the level of a shoe. Who is wearing an officer’s coat. Who is holding the knife. Who is looking at whom. Which hammock a man sleeps in, three chapters earlier, in a scene you didn’t think mattered.

That last one is the mechanic that makes the game sing. Identification almost never comes from the death scene itself. It comes from cross-referencing: a face in the background of scene four is the same face in the foreground of scene eleven, where somebody says a name aloud. The ship’s crew manifest lists ranks and nationalities. A sketch shows where people stood at the moment of the ship’s departure. The information is distributed, andyou are the index.

This is why the one-bit presentation is load-bearing rather than a style choice. Pope dithers everything to two colours, with a menu of palettes named for old monitors, and spent a chunk of his four-and-a-half-year development wrestling publicly with the dithering algorithm — early builds had the pattern swimming as the camera moved, which he documented on the TIGSource forums as he solved it. The payoff: at one bit there is no texture detail to gawp at and no lighting to admire. Everything that survives the dither isshape — silhouette, posture, the angle of a head. Which is exactly the register the puzzle runs in. The art style is the difficulty setting.

Where it fights itself

Two honest complaints.

The first is the tail. Roughly forty of the sixty come apart beautifully, in cascading clusters, in the best deductive week you’ll have. The last handful are genuinely underdetermined — a few crewmen are distinguishable only by a hammock position or by process of elimination, and the game’s final act asks you to identify people whose faces you have seen in a crowd twice. It’s fair. It is also where a proportion of players start guessing, which the three-rule punishes by withholding everything, so the last hour can curdle into an inventory of grudges.

The second is that it’s unrepeatable in a way that feels like a small loss. Knowledge is the only progression here, and knowledge doesn’t unlearn. You get one run at this game per lifetime. That’s the price of the design being honest, and worth paying, though I’d note it’s the reason it will never build the audience it deserves — the second-hand experience of watching someone else play is a shadow of the thing.

The ancestors

Her Story (2015) is the usual cross-reference and it earns it: Sam Barlow’s search-box archive is the other game where the deduction happens outside the software. Fine. The real ancestor is older, and it’s not a video game at all.

I grew up with a C64 and then an Amiga, in the era when the box was a large cardboard object and half of it was paper. Infocom shippedfeelies — a physical map, a fake newspaper, a matchbook — partly to defeat piracy and mostly because the fiction needed a place outside the machine.The Secret of Monkey Island had a cardboard code wheel in 1990.Zak McKracken made you read a real magazine. The assumption underneath all of it was that the player had a desk, and a pen, and the willingness to use them.

Then hard drives got big and manuals died, and games spent two decades assuming the opposite: that any information not held by the software would be lost.Obra Dinn’s notebook is a feelie brought inside the executable. It is the first game in a long while to assume you have a memory and are prepared to use it, and the generation raised on quest markers found that either exhilarating or offensive, with very little in between.

The verdict

Return of the Obra Dinn is the best deduction game ever made, and the margin isn’t close. Its greatness is a rules problem solved cleanly: the three-fate checksum, the un-framed tableau, and a refusal to help that never tips into obstruction. Pope built the only mystery where the detective is definitively the person holding the controller, and then had the nerve to make it look like a 1984 Macintosh so you’d have nothing to look at except the evidence.

The last hour sags. Everything before it is close to perfect, and the moment when three names lock at once — that specific rising chord — is one of the few genuine inventions in the medium’s last decade.

It’s on PC, PlayStation, Xbox and Switch, and it plays fine on all of them, though a mouse and a real notepad is the intended posture. Do it without a guide. There is no version of this game with a guide.

Afterwards,The Case of the Golden Idol is the closest anyone has come to the same discipline in a different frame, andPentiment is what happens when you keep the research and give the player the burden of being wrong forever.

Spoilers below

The kraken is the moment the game admits what it is. For six hours you’re doing sober insurance work — accident, murder, disease, the grim ordinary arithmetic of a sailing ship — and then a chapter arrives with a sea monster in it, and the astonishing thing is that it changes nothing procedurally. The tableau still just sits there. You still have to work out which of the men being dragged over the side is the boatswain. The supernatural gets exactly the same forensic treatment as a fall from the rigging, which is funnier and colder than any reaction shot could be.

The shells and the mermaids are the real engine of the plot, and Pope’s decision to reveal the ship’s fate backwards — you meet the consequence in chapter one and the cause in the last one you unlock — means thestory assembles by the same mechanism as the crew list. Nobody ever tells you what happened to theObra Dinn. You work it out, and the working out is a single continuous act from the first corpse to the last.

And the ending is a gut-punch precisely because you’ve spent the entire game treating sixty human beings as a logic problem. The game hands you the ledger, you assign the payouts, and the widow’s column is the first moment anyone asks you to feel anything. It’s earned, because you did the arithmetic yourself.