What Works And Why: Unfair intel in stealth games

What Works And Why is a monthly column where Gunpoint and Heat Signature designer Tom Francis digs into the design of a game or mechanic and analyses what makes it good.

Games about one player character against hundreds of enemies generally have to give you some kind of unfair advantage. In action games, it’s usually resilience: getting shot in Call of Duty covers you in jam for 3 seconds but leaves you otherwise unharmed, gunshots in Wolfenstein can be fixed with chicken dinners, and in Doom 2016 punching a demon feels so good it physically mends you.

Stealth games need a different solution, because the fun part is generally over by the time you get shot. That’s good – they don’t need jam vision or dinner magic. Instead they need a crutch that helps you before things get that bad. And in games about hiding from everyone, that’s usually intelligence. Information is power. To evade improbable odds, you need to know more than you reasonably should.

There are loads of forms this unfair intelligence advantage can take, but they all operate on a similar principle. When you know more about the enemy than they know about you, you can factor them into your thinking. Information is the raw material plans are made of, and plans are the heart of my favourite games. So I’m going to sing the praises of some of my favourite intel advantages, and dig into why they work.

dxhr cover

Deus Ex: Human Revolution: third-person cover

Human Revolution is a first-person game, but when you hold a button to hug cover, it switches to a third-person camera. I don’t think most players even recognise this as an intel mechanic, which is great. The external perspective obviously helps you see to what extent you’re hidden by the cover you’re in, but more importantly, it lets you see the enemies beyond it without any risk of them seeing you. It’s a magic, invisible eye giving you vital, real-time info a normal human would never have in this situation.

It lets Human Revolution support stealth approaches just as well as Deus Ex 1 without resorting to the rather silly-looking way the original game achieved it: short-sighted enemies. When you play Deus Ex 1 today, enemies being blind beyond 30 metres feels like a charmingly archaic quirk. It would have been a little harder to swallow in a slick 2011 revival.

Far cry tagging

Far Cry: binocular tagging

In most Far Cry games, you can peer at enemies from a distance to ‘tag’ them. Once tagged, they’re tracked on your minimap in realtime, and in the latter games, you can see their silhouettes through walls. I think this is one of the most elegant and effective mechanics in stealth games.

It makes no sense, of course, but it manages to achieve two almost contradictory things at once. Firstly, it gives you a huge intel advantage, which as I say, is vital for fun intentional stealth play. But a lot of ways of giving the player extra information relieve them of the need to behave like a hunted, outnumbered survivor would. If you can see through walls, for example, you don’t need to peek before you round a corner. Far Cry’s tagging is genius because it both gives you an intel advantage and increases your motivation to scout.

In real life, if you had to infiltrate a camp of enemies, you’d absolutely want to get a distant vantage point and watch them before you approach. The information you gained would be partially out of date by the time you snuck in, but it’s so life-or-death that it would be worth hours of uneventful scouting to have some idea of numbers and usual locations. In games, if observing enemies doesn’t tag them, it’s usually not worth your time to do that kind of extensive recon – you’re rarely invested enough that outdated information is worth the time investment.

But because Far Cry rewards your scouting with a special kind of super-intel – real-time tracking of every enemy you know about for the rest of their life – it’s profoundly worth it. It’s also faster and more fun that it would normally be – the tagging is automatic and rather satisfying. And it still leaves gaps for surprises: you might have missed someone who was hidden from your scouting angle, so you still need to be on your toes as you sneak in.

Invisible Inc Vision

Invisible Inc: visible vision

Plenty of games show enemy vision cones. In Invisible Inc, a turn-based stealth game, you can tell whether an area is being watched by an enemy even if you can’t see the enemy themselves. Areas being seen by guards or cameras are tinted red, and if you go there, you’ll be seen. More importantly, the reverse is true: if the square’s not red, you can be absolutely sure it’s safe to move there. That saves endless tedious scouting and moments of brutal bad luck – looking at you, XCOM. And it still stops short of perfect information: you can only tell if an area’s being watched if you can see the area itself. You’re still discovering the level room by room as you progress, and it’s still dangerous to open a door into an unseen room.

Hotline Miami Information

Hotline Miami: perfect intel

I think of Hotline Miami as an ambush game, which is different to a pure stealth game, but it has the same requirement for an intel advantage: imagine bursting into any room in this game blind. Hotline Miami’s solution is simple and effective but surprisingly rare: you just see everything. Enemies are always visible, walls don’t block your vision, there’s no fog of war. That lets you come up with a plan to kill everyone in a room before you set foot in it. And it doesn’t rob the game of surprise or challenge, because it’s so skill based: Hotline Miami lives in the gap between your plan and your ability to execute it.

There is one limit to your intel, of course, which is the screen. You can’t zoom out, and you only pan the camera a little. Personally I think that’s a flaw – it always sucks to get shot from off-screen – but I appreciate it’s not easy to solve. In Heat Signature, we give you perfect intel and let you zoom out as far as you like, but it creates an unhappy tension between wanting to see all the enemies but not wanting everything so small on screen. We ended up having to change the way levels are generated to ensure all the enemies you need to worry about are within a neat rectangle, so you can stay relatively zoomed in and see them all, but it’s not a perfect solution.

splinter cell thermal

Splinter Cell: thermal vision

‘See through walls’ vision modes have a long and patchy history in stealth games: it’s an incredibly useful power, but it’s hard to make it work in the game’s existing art style. So either you spend all your time in neon clown vision, or the game makes it so awkward or costly that using it is no fun.

Splinter Cell’s thermal vision is one of my favourites, not because of the information it gets you, but because of the information you lose. It’s beautiful in a surreal way, but I don’t spend all my time in it because frankly I can’t see what the hell I’m doing. Basic level geometry and important detail is lost, so it’s natural to flick back to a more conventional mode when you don’t need it. And when you do, hunting oblivious enemies by their heat has obvious Predator appeal.

This shouldn’t need a conclusion but it looked weird without one

In my own games I tend to start by giving the player perfect information, and only hide some if I see a specific reason it will make things interesting. In Gunpoint, I never did – you can see everything at all times, because the more information you have, the more intricate and interesting a plan you can form. Any unknowns hinder that, make it less worthwhile to think ahead.

In Heat Signature, too, you can see all enemies at all times. But we do hide what’s in weapon crates, just because that makes all of them worth getting to. If you could see what’s in them, any that aren’t useful to your current character would be dead weight, and all of the interesting things you might do to try to get to them would never happen.

Writing this, picking apart the information systems I love most, I realise I do like games that hide information from you, but only if they successfully make gaining intel a fun game in itself. In Far Cry games and in Human Revolution, I nearly always end up with perfect information on the situation I’m in – I hide and scout and peek and track until I know everything I need to. So hiding that info doesn’t make the ‘action’ part any more interesting for me. But those games make gathering intel so fun, and fit so well into the fantasy of who you are in this world, that hiding it initially is worthwhile.

I’m probably not going to copy that. As an indie, any sentence of the form “make X a fun game in itself” strikes a chill through my heart. Making one game fun takes me three fucking years. But it’s good to understand why it works, and it gets me wondering about a game where intel gathering is the game.

Tom Francis is the designer of Gunpoint and Heat Signature and a former games journalist. You can find more of his thoughts on making games on his blog.


  1. John Vakmos says:

    Here’s one you missed – sound in the original Thief. The sound is good enough that you can hear where enemies are by carefully listening to the sound of their feet walking. It’s a great game to play with headphones.

    • Premium User Badge

      subdog says:

      Right, but this is about *unfair* intelligence in stealth games. In Thief, sound was your primary tool for tracking enemy locations- but it was also their primary tool for tracking you.

      Asymmetric advantages are certainly present in Thief (predictable AI paths, tools, the ability to disappear in darkness) – but they weren’t primarily about giving the player intel. At least not until later in the series with the mechanical eye and scout orbs.

      • Mi-24 says:

        I remember reading about stealth games (I’m pretty sure on Tom’s site) that they are based on 2 approaches / emotions: the fear of being caught and the power you have over the situation. Most stealth games have a bit of both, if you are caught you are at a disadvantage but whilst you remain in the shadows you have the upper hand over the AI.
        But the original Thief definitely falls into the first category more, and maybe this is because of the intel. If you had a game with full x-ray vision and radar you would feel less cautious and more in control, reducing your information to a first person viewpoint and relying on sound limits your confidence when making decisions and forces you to be careful.

        • Sarfrin says:

          But you don’t rely entirely on sound in the original Thief. The light gem lets you know exactly how visible you are. Guards could pass within inches of you when you were in the shadows and never see you.

          • Mi-24 says:

            yes, although I feel that’s more for a usability point of view, it would be really frustrating having to guess whether this darkness did or didn’t hide you when the game is so dependent on it

          • Premium User Badge

            Ninja Dodo says:

            It feels like a concession to UI that replaces an awareness you would have IRL. Your peripheral vision is such that you would probably be aware if, say, your arm was in the light, but in a virtual first-person view that peripheral vision is missing, so it’s replaced by the gem. It’s more precise than reality maybe but it’s a decent abstraction.

  2. Sin Vega says:

    That last one is so important. Vision Modes are a terrible thing and have seriously weakened far too many games (even Crusader Kings 2 completely wasted its pretty map because it was so useless compared to spending 98% of the game in politican map mode). But done right they obviously work. Rainbow Six Vegas also did both thermal vision and third person cover really well.

    • R. Totale says:

      Hi, is Unknown Pleasures coming back anytime soon? I haven’t known pleasure in weeks.

    • grimdanfango says:

      The thing that Splinter Cell did so well with the thermal vision wasn’t just that it took away information, it was that it wasn’t all-powerful either… they consistently accounted for material types and thickness throughout all of the levels, so you could see easily through tent-canvas, but heavy concrete walls would block most/all heat.
      It was a beautifully implemented effect, and felt like it blended thematically into the world entirely, rather than feeling like a cheap hack to make the game more approachable, as most “see through walls” systems tend to.

      • haldolium says:

        And then, over 15 years later, along comes Ghost Recon: Wildlands where they entirely screwed thermal vision in making it simply a shader/postfx that is being placed on top of models and ignores any materials or thickness.

        Most disappointing game design failure given the outstanding implementation of thermal vision in the early Splinter Cell gaems.

      • Mi-24 says:

        I also really loved how it wasn’t just a simple see enemies mechanic: it would reveal laser tripwires and mines, lights and hidden electronic devices, and my personal favourite: if you view a combination lock shortly after somebody used it some heat residue will be left on the keys of the code.

  3. Benkyo says:

    I think it is worth mentioning that Invisible Inc. also works fine (better!) when you ramp up the difficulty and remove the ability to see the sight cones of guards and cameras that you can’t see.

    I’d say the real “unfair intel” is the fact that sight cones exist in the first place.

    • Wormerine says:

      True. I am surprised that vision cones were a focus of the unfair super intel in INV. Inc. you can predict what your enemies will exactly do next turn by spending an action point!

      Wonderful game.

  4. Mi-24 says:

    A game revolving entirely around intel gathering would be interesting, unfortunately I’m not sure it would work unless it involved other humans. Say you had a game where you did the spotting far cry style of an outpost and then an AI attempted an assault, based on your findings. I don’t know how exciting this would be for a player, since the attacking AI would have to be designed to react appropriately to your intel, e.g if you spot a sniper then the AI will take out the sniper from range maybe. However this would simply mean you slightly controlled a flowchart: did player spot sniper, if yes then take out sniper, otherwise get shot by sniper. Even worse would be an example where you got some text description of the result of your intel, effectively there would be a one-to-one relationship between what you found and the result.

    I think this could be fixed by making the story the focus of the game in the style of virginia / telltale in terms of your actions having the effect of impacting the overall story. As i wrote this I remembered ORWELL, not played but it seems that it does exactly this, what info you find and relay changes how events play out. I can imagine an interesting spy type game set in the cold war without weapons, where you take photos, sneak into buildings, communicate with informants and steal documents.

    It would probably work better with coop, I can think of Arma, where playing mulitplayer with one player as spotter / drone recon adds an interesting aspect to teamwork. Maybe though its a bit similar, playing entirely as observation might make you feel frustrated at not taking part in the action, but the human interaction between the source and destination of the intelligence would be much much stronger than any artificial mechanism.

    • Frank says:

      But there already is a game (almost) all about intel gathering where you play vs AI: link to en.wikipedia.org

      In contrast with the example you proposed, in Oasis, you gather intel and than plan a defense, not an attack.

  5. Artea says:

    “Plenty of games show enemy vision cones. In Invisible Inc, a turn-based stealth game, you can tell whether an area is being watched by an enemy even if you can’t see the enemy themselves”

    This is only true if you play on the lower difficulties. On the highest difficulty, you can only see the vision cones if you also see their source and I would argue that this is the best way to play the game, since it forces you to use every tool at your disposal to be alert of your surroundings.

    “Human Revolution is a first-person game, but when you hold a button to hug cover, it switches to a third-person camera. I don’t think most players even recognise this as an intel mechanic, which is great. The external perspective obviously helps you see to what extent you’re hidden by the cover you’re in, but more importantly, it lets you see the enemies beyond it without any risk of them seeing you. It’s a magic, invisible eye giving you vital, real-time info a normal human would never have in this situation.”

    This is a terrible design decision and is one of the reasons why Human Revolution is inferior to the original, where learning about your environment was done organically through leaning around cover and listening to enemies without ever abandoning the first-person perspective and without an omniscient 3rd person camera that shows you things your character shouldn’t be able to see.

    • Mi-24 says:

      I personally do dislike the 3rd person mechanic, but I feel the alternative is also not a great choice, I remember on the ship section of the DLC in narrow corridors, having to walk right into the middle of the corridor to get that info feels silly, any person with functioning eyes would see you. I think a better solution is the lean mechanic, or having some kind of periscope gadget

  6. Baines says:

    There is one thing that is so important that it deserves to be mentioned even if it isn’t what the article meant by unfair intelligence, and that is… uhm… unfair intelligence.

    Specifically, that the AI in stealth games is almost always intentionally restricted to give various advantages to the player. Guards often are restricted to specific patterns, goons in Arkham are discouraged from doing 180 degree turns in order to make it easier to sneak up on them, etc.

    This often even creates an article-friendly version of “unfair intelligence”. These AIs, even the “smart” ones, are intentionally predictable. This awards the player with the “unfair” advantage of knowing with a fair degree of certainty (in some games this can be 100% certainty) how the AIs will act and react.

    • Mi-24 says:

      A good point and a great play on words.

      I guess as well as being intentional this is due to the restrictions on AI, because unless you have a really interesting decision making model actions will be deterministic: AI’s will always react the same way (or at least predictably) in the same situation. Until this can be overcome all game AI’s will be flawed in this way.

      Also this can be reduced partially by numbers, as naive as that sounds, in MGSV for instance you can outsmart and predict a single soldier, but an entire bases worth of guards interacting, sharing sight-lines and communicating when investigating is a much more interesting challenge.

      • zabieru says:

        “Deterministic” and “player-predictable” aren’t always the same thing, though. For instance, while the AI’s actions might be fairly simple and deterministic, if the player doesn’t know what a particular AI soldier’s goal is, the soldier’s left turn towards the barracks will be a complete surprise.

        We can see the inverse of this pretty easily: look at all the effort stealth games put into telegraphing alert levels and the like. Some of that is pretty natural: the sirens and searchlights, and even telling your buddy/CO “hey, think I saw something, gonna check it out” makes a lot of sense. But stuff like “Gonna get some chow and turn in,” not so much: there are twelve of us at this outpost, we all know the watch schedule. At my work, you make sure whoever’s responsible sees you heading for the door and out of uniform, and that’s enough: they know you’re on lunch/going home, and given the four-hour difference between those events, they know which you’re doing. A ninja hiding in the ducts, though, would need a much clearer cue as to whether this particular guard was going to be coming back with a sandwich any second or is gone for the day.

        You can make a simple AI less predictable without making it random if you give it divergent responses to different inputs, and preserve a little mystery about what stimulus it’s responding to. Maybe the response pattern for “heard something in the bushes” is a little different from the pattern for “saw a flash,” which in turn is different from “saw movement.” Tie in your stealth level, too: “saw someone walking” merits a different response from “something moving at knee height.”

        Since you don’t necessarily know whether the guard saw you or heard you, you can’t be totally sure what he’ll do. You could even manipulate this: maybe if you’re good enough to make sure you are heard without being seen, you can draw a certain response.

        One of my minor pet peeves with MGSV, among other games, was that none of the guards would do recon by fire. It boggles my mind that a soviet conscript in Afghanistan or a merc in Africa, on becoming aware of a wild animal (he thinks) in the bushes… Would wander over to check. I guess if you’re hoping to get bitten and be sent home on medical, maybe it makes sense? But in any other circumstance, given the rules of engagement they use in the rest of the game, they’d put a couple rounds into the bushes to scare the damn dog off. This would be a great mechanic! It adds a little difficulty (makes it harder to choke out a base one man at a time) but also the player can use it (if you can trick a guard into thinking there’s a dog and time it right, you’ve got a bit of cover if you need to take an unsilenced shot yourself. Not exactly realistic, as a pistol shot from the East is still going to stand out from rifle shots up North… But close enough for vidya games.)

        • Mi-24 says:

          those are some really good points, especially about variation of responses, I was just kind of assuming that any player would play the game long enough to learn all the responses, but if like you say you don’t telegraph them so much this wouldn’t be the case.

          I guess the “guard doing stupid thing to investigate a mouse” problem is one that is deliberately designed in for fun / making the game easier. And to be fair an outpost in which the guards stayed exactly at their posts and shot you from range wouldn’t be much fun, unless you had some further way of luring them out like impersonating an officer / playing dead.

          That said the guard shooting over your head would be a cool mechanic and would be nice if it happened occasionally to surprise you. I just realised the guards sometimes do it in far cry 5, I was recently doing an outpost and they did shoot into the bushes when they got to a high enough alert state without actually seeing me (I think seeing a dead colleague and then hearing a thrown stone did it).

          • zabieru says:

            Yeah, totally: “Guard checking out a mouse” will always be with us and for good reason. If everybody was a perfectly-disciplined genius telepath, there’d be no need for player characters! And to be entirely clear, I’m not really arguing for naturalistic-but-mysterious guard barks: I think “Gonna grab some chow” serves a useful purpose even if, in real life, at least some of the time close-knit groups will communicate in ways that aren’t legible to a guy watching from the bushes. I just wanted to point out that there is a level of artifice (and a lot of effort and testing) that goes into making those sorts of movements legible to the player.*

            I was personally perturbed by the absence of “check out a mouse with short, controlled bursts” in MGSV because it seemed like such a natural, obvious thing to include, doubly so in a game that actually cares about the effects of war on the natural and civilian environment.

            You can also do some personality stuff to add weightings: maybe some guards are new and scared, which makes them less likely to shoot at low alert levels but more likely as they get nervous. Others are four-year veterans and know that wasting five rounds on a mouse or civilian costs them nothing, but holding fire could get you killed. Or some guards like dogs but others hate ’em.

            Or… Good troops don’t react to a threat by all focusing on it, they react by picking up their own sectors. That makes stealth hard, of course… but it could be used as part of a useful diversity-of-response scheme: the FNG will look wherever you make a noise, but the vets will keep watch on their sectors. Observation or eavesdropping might tell you which is which and give you an exploitable weakness, but if you’re stumbling in you have an unpredictable response.

            There are other ways to skew the results too: maybe guards get tired, so the best times to move are an hour before shift change. But fresh guards just coming on shift change are more likely to be diligent and actually go check out that sound instead of just going “nah, fuck, ain’t leaving the OP just for a mouse.”

            You don’t want to do this all at once, of course: It would be a totally illegible mess. The player would have no idea what caused the reaction she just observed: did that guard react differently from the last time I tried this because I did something different? Or because it’s a different time of day? Or because he hates dogs more than the other guy? WHO KNOWS, fuck stealth, guess I’ll need a bigger gun.

            But a selected subset of these different responses, with appropriate measures to teach them to the player (especially things like shift changes: personality differences you can probably leave more vague, load-screen hints about some guards being more nervous than others might be enough) could give you a lot of interesting design space.

            Ultimately, that’s basically how a lot of MGSV works in terms of more guards=bigger threat, as you pointed out. I really really liked that model and I’d love to see it extended in different directions.

            *Real-ass bank robbers/KGB commandos/whatever spend days, not minutes, watching their targets to build guard schedules and the like. That’s the sort of patience and time it takes to turn mostly-scheduled but unexplained human actions into a legible set of routines. It’s also a really hard task which very few people can do even when it’s their actual job (which is one reason covert ops is an elite field and successful career bank robbers are rare). Doesn’t make a good game. Producing a realistic-looking set of guard actions that nevertheless allows a lurking player to work out what the guards are doing in minutes is an impressive achievement.

          • Mi-24 says:

            Maybe you just need the right combination. As you say too much variation with not enough explanation becomes unreadable and frustrating, but too much telegraphing without variation becomes boring. Maybe adding readability in different ways could work; instead of behviour always being dependent on a bark it could have some visual attribute. This works at least for the separation between newbies and well trained soldiers / officers / ncos / regulars. Like in far cry 3 the 2 islands had different enemy contingents, with different uniforms, equipment (and I cant remember if they had different AI). personally I really loved this, the feeling of meeting your match again was refreshing.

        • Baines says:


          A simple way to make a patrol route more effective would be for the guards to look around at unpredictable intervals. Eliminating those guaranteed windows of safety would not only be logical intelligent guard behavior, it would take almost no effort to implement.

          So why don’t more games do it, and why are the games that do do it often looked down upon for it? Because from a player perspective, it simply isn’t fun to have your carefully planned stealthy approach sunk simply because unpredictable random chance caused a guard to turn around and see you. You might see an entire level’s run scrapped because of one random number that you could not even affect. As mentioned earlier, this is why the Arkham devs went out of their way to prevent unalerted goons from doing 180 degree turns, because otherwise any attempt to move behind (either to bypass or ambush) a goon would have its fate determined by random chance regardless of skill or planning.

          And thus, the AI becomes more predictable for the sake of gameplay. The player is given the knowledge that, barring his own mistakes, there are guaranteed (or near guaranteed) safe windows of movement behind guards.

          Similar goes for stuff like AI agents calling out their actions. The more intelligent behavior would be to use hand signals, radio, mouthing commands, or other more private methods of communication, not to shout so that everyone hears when you need to reload or when you are attempting to sneak up behind your target. Developers intentionally effectively make the AI less intelligent to create the illusion of more intelligence, and to reduce player frustration, by having the AI system constantly inform the player of its actions.

          • Kitsunin says:

            And the only reasonable alternative I can see, would be to assume the player character is prepared enough to understand such signals, and predict unpredictable behavior. But for instance, to have a UI pop-up tell you that you know a guard is heading to the break room is basically just a really clunky version of the stupid “gosh I’m hungry” call-out, which would make things feel less realistic despite being more realistic.

        • Malleus says:

          Splinter Cell Chaos Theory did that though. If you triggered a guard enough times (with noises, distraction) without getting caught, next time you made a noise, he just started shooting in that direction.

    • jeppic says:

      An excellent point.

      One of my favorite solutions to this is surprisingly from the cooperative heist board game, “Burgle Bros.”. From a design perspective, not only do you need to balance this exact problem of predictability vs surprise, but you also, because it is a coop board game without a computer to control it, make the “AI” easy to calculate for the humans running the game.

      In the game, you draw cards representing one of the locations on the floor you are on, and then the guard moves towards that location. So once the card is revealed, the guard becomes somewhat predictable in that it will move at a set rate towards that location. However, you never know where the guard will go next until you flip the next card. Since guard movement is resolved after each player’s turn, and the guard’s movement can increase as the game progresses or certain events occur, you can usually predict what the guard will do on your turn, but not if he’ll discover you on subsequent turns.

      I know it’s not a video game, but it’s one where I really appreciate how tightly the mechanics and theme are tied together given the limited fidelity of a board game. I also think there’s some value to understanding what works with such tight restrictions, especially with regards to the design of Indie games that are forced to emphasize clarity over complexity in most cases. (It’s also what make the successful ones such satisfyingly tight packages)

  7. poliovaccine says:

    Nice article, some good points here. Thanks for that.

    I’d say that “unfair” is relative, though, and “unrealistic” is maybe closer. Like in DX:HR, for example – I’ve seen a good number of people complain about that same cover mechanic you praise, calling it “dumbed down” and so forth – but to me it’s not so much dumbing down as it is approximating other, subtler informational cues games can’t simulate.

    Like say, in real life we all have a certain somatic or kinetic sense, a sense of ourselves and our bodies – and obviously we don’t have that in games, at least not til the Vive releases its neural interface USB adapter and allows us to feel simulated pain. But my point is, in real life, you don’t *need* a third person view to know whether the top of your head is poking out of cover or not. Or like the “visible sightlines” in Invisible, Inc – you’re still close enough that they’re basically approximating stuff too subtle for games to represent, because in reality, if you’re in one room and someone who doesn’t know you’re there is in the next, they will make little noises and scuffles as they move around, the leather of their chair will groan as they adjust themselves, the pages of their newspaper will rustle as they leaf through… all stuff which functions essentially the same as these shorthand visual representations, because games just can’t simulate that level of subtlety to perception – yet.

    I mean, have you ever noticed that you can sort of tell the difference between a TV on mute and a TV that’s off, even from another room? The only way I can think to explain it is that the TV gives off a high, near-inaudible whine when it’s on and power’s running through it, and that’s somehow perceptible even if it’s not on a wholly conscious level – because that’s basically how it seems whenever I notice that difference. Stuff like that, when translated into games, effectively looks like the stealth indicators for modern Fallouts: [HIDDEN], somehow you know you’re not being watched, your unconscious has picked up on all the cues it can and has delivered the all-around “okay for now” signal to your primary consciousness -[CAUTION], the fine hairs prick up on the back of your neck, and you have this uneasy feeling, like people tend to do when they feel like they’re being watched… etc, etc.

    Basically, games don’t have human intuition, and they have to represent more senses visually as a result. Touch is nonexistent – you can’t feel a draft and realize the door was opened downstairs. Sound in games is rudimentary when compared with the many details of actual audio input we perceive every day – which we then mostly narrow down to those few details which are relevant to us at the time, and that makes up the level of what we’re *consciously* aware.

    And have you ever been in serious danger which made you genuinely believe you were going to die? People say time slows down in those conditions but that’s not exactly right – you just think faster, and in retrospect it seems like slow motion, but really it’s just extra density of details. We say time slows down because it’s the closest way we have to approximate it, but really it’s just that somehow we have time to consider all of our relationships and regrets in that brief moment when the ground is hurtling towards us. We’re capable of that level of totally unfiltered, unrestricted perception, but it would be exhausting for everyday life, and needlessly distracting (incidentally, this is sort of why nobody gets addicted to psychedelic drugs haha… no matter how much you like em, you can only handle so much hyperstimulus).

    So I think of stuff like the “light gem” in Thief or even the bulletproof protagonists of FPSes to somewhat approximate actual information we’d be innately aware of in real life. Like even the Call of Duty/FPS thing – it doesn’t seem very realistic to be so forgiving, when in reality even “just” a flesh wound, like a single shot in your thigh, would be pretty goddamn distracting and debilitating – however, the *real* thing it’s approximating is the fact that, if I were really at war, instead of playing Call of Duty, *I’d be fucking scared,* I would understand implicitly that there are no respawns after death (or at least I know I can’t rely on that prospect turning out true), and as a result of the reality of the situation, *I would be way more slow and careful than I’ve ever been in an FPS.* So let’s just pretend that, instead of time-rewinding 30 seconds before my death via quicksaves, what really happened was that I never got shot in the first place cus in real life I didn’t actually just dash up from cover to go running around the game world and picking up enemy’s ammo by running over it and therefore hearing a satisfying ratcheting click noise with every stray magazine I reclaim.

    That is, that’s what I would be thinking if realism were all that important to me, haha.

    • Sin Vega says:

      The only way I can think to explain it is that the TV gives off a high, near-inaudible whine when it’s on and power’s running through it, and that’s somehow perceptible even if it’s not on a wholly conscious level

      It does, it’s a very high frequency that some people (mostly people under about 25) can consciously hear. You can do a fun experiment with a group of people about that age to discover which of them are insecure enough about their age to pretend.

    • April March says:

      Excellent points. The article mentions how the enemy tagging in Far Cry is, from a realistic standpoint, ridiculous. But. In Far Cry 2, your map was very clearly a paper map that your character would hold up. This paper map, however, had a continuously updating arrow indicating your position. This reading, for me, meant that my character was good enough that he could look at a map and instantly figure out his position, at all times, even though I, the player, certainly can’t. It follows from there that, if I tag the enemy, it’s not that I start magically knowing his exact position; it’s just that my character is good enough to infer all that.

      I wonder if there are doors to be opened in stealth games by making this connection more explicit to the player.

      I once imagined making a stealth game in which guards were people infected with an alien virus, or just very poorly made alien clones. They’d follow in repetive, circuituous routes because of their obsessive behaviour, and they’d shout out their every thought because of their verbal tics.

    • Premium User Badge

      Ninja Dodo says:

      Ah, I just got to reading this comment but I totally agree (and tried to make a similar point about Thief’s light gem above).

  8. Mouse_of_Dunwall says:

    Er, I don’t really have anything to add, but I wanted to say that I really enjoy this type of analysis of stealth games. It’s too bad Sneaky Bastards hasn’t been updated in a while, they used to write some great stuff like this.

  9. Aetylus says:

    Interesting article. I just finished Watchdogs 2, enjoying it far more that I expected… and in reflection a significant part of that can be attributed to the system of hacking security cameras to provide progressively more advance intel, including a big red x-ray silhouette of enemies.

    There is an added bonus of these ‘enhanced vision’ red silhouettes… they make the game much more playable for those of use who are vision impaired… trying to spot some dark green pixels in a lighter green background is nigh impossible for me… spotting an always-on red silhouette I can do.

  10. indigochill says:

    I liked Alien Isolation’s unfair intelligence feature: the motion tracker. It comes straight out of the movies, and lets you see which direction any nearby entities are regardless of any obstacles, but not whether it’s above or below you or whether it’s friendly or not, or even what direction its facing. It’s really necessary because you’re stuck always in first person and there’s no predicting the alien’s patrol pattern when it appears (although it does still have some predictable behaviors that can be exploited to get around it, which is also thematic since it’s essentially an animal with certain instincts, just as the android patrol patterns are predictable because they’re robots). It does a great job of giving the player just enough information to make informed decisions while still feeling completely vulnerable. Just overall a superb reinforcement of the theme through game mechanics.

  11. LennyLeonardo says:

    This is a really great article. Personally I think the absolute epitome of the super-info stealth game is the Silent Hunter series. All the info-gathering tools you have (sonar, radar, periscope, deck watch, radio messages) feel totally integrated into the world, with the usual abstractions being hidden or done away with entirely. You still have intelligence-gathering superpowers, but they don’t feel artificial like they do in most games. Not that abstract is bad, it’s just the connectedness in Silent Hunter feels so good.

  12. caff says:

    Great article, and I’ve just realised I’ve missed the previous two articles by Tom F. I do like how he relates to the mechanics built into his own games.

    Regarding the duck and cover mechanic, my brain automatically assumes it is fair – by imagining that either my character is peeking over the top, or made a mental map before taking cover.

  13. y0LAs says:

    Well, in Hotline Miami, you can look away, with Shift on a keyboard and LB on a gamepad.

  14. Chacaceiro says:

    I am an avid player of Invisible Inc. and for the higher levels of play (Expert Plus difficulty) you are unable to see the red squares without seeing the subject which is looking at you. However, the game also has a way to “Predict” the enemy movements and to safely peek before bursting into the room. It’s all about playing safely and knowing about some game mechanics like “if someone comes inside the room from that way, what spots are still hidden from sight?”