The Stealth Letters, Part One

Stealth game fans pay heed. Over the next two days RPS hosts a conversation between Nels Anderson, Lead Design of Mark Of The Ninja, and a number of other stealth-game luminaries, as they discuss matters of of sneaking and hiding in videogame form. Anderson talks to Patrick Redding, Game Director on Splinter Cell: Blacklist, Andy Schatz, creator of Monaco, and Raphael Colantonio, co-creative director of Dishonored.

This is part one, part two will appear tomorrow. Onwards! (But stay out of sight…)

Nels Anderson, Lead Designer of Mark of the Ninja


Thanks for taking the time out of your undoubtedly busy schedules to talk, although given the subject material, I don’t think I had to twist any of your arms too hard. Given that we’re all just finished or are in the midst of making a stealth game, it seems like it would be mad not to take the opportunity to discuss this style/genre/whatever we call it. Plus I thought it would be interesting to get a couple of different perspectives, comparing the experiences of smaller folks like myself and Andy to the much larger scale you work in, Pat and Raph. Also, maybe Pat could offer some insight into what it’s like to work in a long-established series like Splinter Cell, vs. the rest of us who are working on games without direct predecessors.

To start, what is it that you all find interesting about this type of game? Or more simply, why make these games? Beyond their surface trappings, they’ve always felt … distinct, I guess could be the word. They felt different from other types of character-based games, but I was never able to put my finger on exactly why until I started really digging into how they work so I could, you know, design one. So what seemed to emerge, and I’m really curious about the rest of your take on this, is the flow of their gameplay is fundamentally “pull” where nearly all other character games are about “push.” So by “push” I mean the gameplay is mostly taken on the game’s (read: designer’s) terms. The player enters some area crossing a trigger, enemies spawn, they come charging at the player and they have to react. Even something that’s more of a puzzle is still about the player encountering said obstacle and trying to derive how they must get past it. I’d call this sort of thing very designer-centric. And to be clear, that’s not a value judgement- great games have been and will continue to be made in this vein.

But stealth games are interesting to me because they’re not this. As the default state of the game world is to be ignorant of the player (because, well, they’re hidden), it must be able to operate independently of the player. This seems to result in stealth games being more systems-oriented, with lots of interconnecting gears and pulleys, so to speak. The players wait and observe, and then perturbed the system in a very considered way. They basically pull the world into some new state that suits their purposes to interact with its systems. While success is still defined by the designer on a very high-level, almost anything else is at the player’s pace and on their terms. The context of play is no longer just about reaction and survival. I think it’s fair to call these games more player-centric. And that’s really interesting to me.

Especially because the consequence of this seems to be that play in these games tends to be far more intentional than other types of character-based games. Thinking tends to take place at a higher level that’s more about chains of cause and effect, rather than discrete moments of immediate reaction. The players are able to manipulate the game’s systems to get the output they want. There is a dependency that emerges from this though- players need to be able to understand the game’s systems to bend them to their will. I think this is a factor in why some folks don’t really connect with stealth games. The core stealth systems are generally quite opaque and they require a bit of risk and experimentation to parse out. But given the power dynamics of most of these games (i.e. the player is quite vulnerable in many circumstances), I can see how it could be a bit schizophrenic. I think some folks might be feeling that friction when they try stealth games and don’t really connect with them.

And that’s the long, circumtuitious path to the guiding principles I had when designing Mark of the Ninja. Because 2D side-scrolling stealth hadn’t really been done before (Tom Francis didn’t make the excellent Gunpoint known until about 6 months after we started Ninja, ditto on Stealth Bastard), the design process ended up being a lot about deconstructing what works in 3D stealth and figuring out how to provide gameplay that *felt* similar, but actually was achieved through rather different means. I’ll spare the details for now, lest I bang on too long before we even get going, but suffice to say, we ended up having to do a lot of things that were pretty counter-intuitive. Andy, even though Monaco is top-down and it seems to have a little more in common with 3D stealth games (as in you have, you know, corners to hide around if guards are approaching and such), I imagine there were some differences, yes?
Pat, I wanted to turn things over to you first though. Even though Far Cry 2 isn’t explicitly a “stealth game,” I think it does have a lot of those player-centric dynamics that I mentioned above. And I know I spent a lot of time skulking around in tall savannah grass, waiting for the right moment. Was that an intentional decision you all made or was it just a subconscious thing with Clint and so many other folks having spent time on Splinter Cell? And what kind of things did you all discover while working on FC2 that you’ve brought back to Splinter Cell?


Patrick Redding, Game Director on Splinter Cell: Blacklist


You brought up an interesting point about the underlying philosophy of Far Cry 2 that sort of exposes its stealth roots.

Around the time work started on FC2, we were really preoccupied with two key ideas that Clint had explored in Splinter Cell: Chaos Theory, the first being design for intentional play (which was literally the title of his GDC talk that year), the other being reconciling dynamic meaning with narrative theme. Both those interests strongly informed the design culture on FC2, but we definitely succeeded more at the former than the latter. If you talk to Jon Morin (who was the Level Design Director on FC2 and is now the Creative Director on Watch_Dogs), he’ll tell you about the challenge of identifying parameters for an open world that allowed the level designers to organically deliver recognizable gun combat encounters, in a space where both the player and the systemic AI have 360 degrees to maneuver. His solution was to look at the spatial density of game ingredients in different combinations — so attacking from low density to high, or high to medium, or medium to low, etc. — because anything more pattern-driven started to expose the hand of the level designer in a really ugly way.

That approach had a surprising output, which Clint talked about after FC2’s release: The defining moment of the mid-level experience became that instant when the player has observed the tactical situation near an objective, developed a plan, and was just about to execute that plan when the game world systemically burps up something utterly unpredictable to monkey-wrench everything. If FC2 succeeded at nothing else, it took that notion of a systems-oriented, independent, objective universe and gave it permission to explode into anarchy, which nicely aligned with our larger intended meaning, I think. That dynamic is ALSO very present in systemically deep stealth games. Splinter Cell always featured fairly linear environments, but as the series progressed it started opening up the levels to different player approaches, letting you tackle the objectives in different order and experiment with how that altered the conditions in each section as you completed everything. That meant you weren’t just playing cat-and-mouse with an individual guard in a corridor, but with the entire security apparatus of the map.

Something I’ve been thinking about a lot recently: When stealth games had their “golden age” from Thief and MGS to roughly the first Splinter, it felt like they hit a sweet spot between player accessibility and production values that were possible for that generation of tech. It was okay for the game’s metaphor to be a bit gamey if it was readable and affordant and rewarded exploration.

Somewhere along the line it became too expensive to make games like that because the world needed to be “cinematic” and that’s antithetical to very high levels of interactivity. It became too expensive to create a fully-realized AI that might never know the player was there or a section of the world that the player might never see.

The other challenge was that as blockbuster games became more immersive and realistic, the tools they gave to the player for reading the state of the game – which really hinge on the simulation of human reactions that are at the far end of nuance scale – seemed increasingly crude. In real life, if I’m hiding in a dumpster to avoid the cops, and I’m peering through a tiny crack at some scared rookie who’s slowly advancing down a dark alley with his gun and maglight, I’m parsing non-verbal cues (to tell me whether I should hold my breath or make a run for it) that are as old as time and yet a thousand times subtler than anything we ask AI to do in games.

So the prevailing view in our industry has been that players won’t tolerate having to process a ton of second-to-second information about the game-state, making pure stealth play a tough sell. But when I look at Mark of the Ninja or Monaco, what stands out is that the more symbolic/illustrative metaphor allows players to process a LOT of information. They’re immersed systemically, and it doesn’t pose a problem at all.

Andy, I know that Monaco went through some fairly significant changes in terms of graphics and presentation direction between the version you showed at IGF and what we’re seeing now. How much of that you trying to optimize the amount of game-state intel you’re feeding the player, vs the visual polish that goes along with finishing the game? By contrast, did the player’s low-level actions change much at all?


Andy Schatz, creator of Monaco

First off, let me thank Nels for hosting this conversation, and all of you for including me. It’s an honor to be included in such an illustrious crowd.

Pat, you brought up an interesting point: that in the real world we can use subtle clues to understand human behavior, whereas in a game, NPC behaviors are more mechanical, limited, and far less subtle. I don’t have a problem with this. The ideal AI — in a stealth game — is one that follows predictable but complex rules, immersion be damned.

The purpose of infusing a humanity into stealth AI is NOT to make the player feel “in the game”. The purpose of realism in AI is so that the player can more easily intuit the rules that govern their behavior. Immersion, in my opinion, is a canard, given that we are asking players to accept all sorts of weird things bout the game world and its rule-based limitations. Players accept the fact that the guard doesn’t check the closet after the fifth time he heard a suspicious sound, and that doesn’t in any way reduce the fear of being caught or the thrill of escape.

To that end, Monaco really throws immersion out the window. Instead, the game relies on the player’s ability to interpret symbology in order to connect emotionally with the state of the game world. As a side note, this is one of the reasons that Monaco is set to a silent film score piano soundtrack — because the game very explicitly asks players to use their imagination to fill in the blanks between what they are seeing and the reality of the game world.

This leads me to Nels’ comparison of stealth games being mostly “pull” while other games are “push”. Pat’s description of the Far Cry world being almost fractal in it’s self-similar interactivity reminded me of how difficult it is to design a game where the world simulation acts as the main character of the game, rather than the player character himself. And when the game world is your main character, the player becomes a hero of the negative space, rather than the positive. Living in the negative space of the game world is disconcerting, and is one of the primary drivers of the tension and fear in a good stealth game. In games that provide a stealth option as a predecessor to combat, players use the negative space to choose their entry point into the positive space.

Of course, one of the things I think stealth game designers often struggle with is that it’s hard to avoid punishing the player when they fail to remain in the negative space. In many stealth games, players only have a choice between perfection and failure. I’m curious if games like Far Cry and Dishonored, which seem to live in BOTH the positive and negative space, deal with rewarding the player for less than perfect stealth strategies. During development, did you approach level design in binary terms (“the stealth option” and “the combat option”), or did you approach things from a more analog perspective? Are you relying on the player to make choices based upon preferred play style or strategic analysis? And do find that players are predictable in those choices?

Raphael Colantonio, co-creative director of Dishonored

Hi All,

Great question Andy.

Harvey and I often talk about this very point with the team: We don’t have “the combat path” or “the stealth path”, we like our environments to support a variety of play styles at any point: so we design our missions and locations in a way that the same spaces can be used for both, including a player’s sudden change of strategy: there’s always a place to hide, etc… so the player really makes his own path as he plays.

In our games, stealth is not enforced, it is encouraged in different ways: you save resources, there is the emotional pay off of sparing someone’s life, also you’re more lethal when you take people by surprise… but we have to take into account that the player can fall back to combat at any time, and we must make sure he still has fun if he does so and doesn’t feel punished for falling out of stealth. This specific dynamic has really been one of our hardest game design challenges in Dishonored: to encourage stealth, but not punish the more direct player or make him feel like he’s failing in any way. We want to make sure he’s still having fun no matter what. To meet that challenge, we did a lot of micro tuning, including making the combat very hard, while keeping it fair and rewarding: the player might die but he had fun and he felt powerful on the way.

Nels, I really liked what you said about what attracts you to stealth games. I share your thoughts, specifically the part where you describe the world living on its own. In addition, I would say that the feeling of being somewhere forbidden, and the feeling of being vulnerable if I get caught are two things that really work for me.


Next time: multiplayer stealth?


  1. Hoaxfish says:

    While it doesn’t seem directly related to the people in this article, it still seems appropriate to link to

    link to

  2. BrendanJB says:

    This was an awesome insight in to the minds of some very creative people; I like the candid-yet-technical nature of their responses. I’ve always had a penchant for stealth games and their thematic and mechanical designs, and how designers try to facilitate those ideas in different ways, so this was a very satisfying read. Cheers to all who gave their time!

    • Randomer says:

      Indeed! That said – Dear Andy: Less insightful discussion, more of releasing the wonderful game I have been waiting to play since that fateful PAX so many years ago! Please. Just a little playful teasing from the heart, of course. Take as many years as you need! We’ll be waiting here for you after you’ve had time to figure things out.

  3. DiamondDog says:

    I imagine this might be down to what I personally find interesting, but it seems like RPS has gone up a gear in features this past week or so. Some fantastic articles to read.

    So erm, well done!

    • grundus says:

      I would also like to congratulate erm for fixing The Features Machine.

  4. Justin Keverne says:

    I’m not really sure what Andy means by “negative space”, is he suggesting that within any richly simulated world stealth players become passive existing within the simulation but without influencing it? That’s the only way I can see that making sense if combat is considered the “positive space” due to combat requiring action and manipulation of the simulation itself.

    If that is what Andy means by negative space then I have to disagree. You can play Thief without leaving the negative space, by moving between existing shadows not using any of your tools and barely interacting with the simulation, but that’s only one approach. You can also take an active role in it and manipulate the simulation to your own ends, Mark of the Ninja similarly offers such possibilities.

    Interestingly though Dishonored doesn’t offer those same possibilities, there are few direct ways by which to interact with the simulated world for a stealth player. You can alter the state of certain security devices and of NPCs, but you can’t directly influence the simulation like you can in Thief, there are few options to change the properties of a particular aspect of the simulation, no water arrows or moss arrows to change the properties of the environment for example.

    When it comes to tension and fear in a stealth game, I also think Andy’s got that backward as well. The best stealth games make the player powerful when hidden and vulnerable when not. The tension comes not from being an outsider to the system, but from the knowledge that at any time the power granted by being hidden can be removed either forcibly or simply because you can’t progress if you stay put. The thrill of stealth in that sense is in having to wilfully move from a position of strength to one of weakness simply to progress. In an interesting way that’s part of why Return to the Haunted Cathedral and Robbing the Cradle are such scary levels, they play on that stealth specific need to wilfully make yourself powerless.

    • Outright Villainy says:

      With regard to your point on Dishonored: It did flirt with a bit with rewiring the walls of light or switching off the alarms, but I agree that environmental manipulation was perhaps a bit limited. That said, the Blink power would seem to negate a real need for changing your environment, since it’s basically your playground anyway; making the environment easier to traverse is never much of a problem when you can reach almost anywhere with ease, instead the focus becomes more on awareness and timing, which is plenty satisfying in its own right.

    • Jason Moyer says:

      “The best stealth games make the player powerful when hidden and vulnerable when not.”

      Agreed 100%.

      • Mungrul says:

        Which is one of the reasons why I absolutely adored the Cradle in Deadly Shadows.
        All of a sudden, the world and its rules are inverted. You’ve been skulking in the shadows, an omnipotent voyeur, then the Cradle exploits that and turns it on its head.
        It truly is a genius piece of level design.

        Also, I think it’s one of the great shames that community tools were never released for Far Cry 2. It’s a better game than the first one, and I’d argue it’s also better than Crysis, but it is severely flawed, and those flaws could have been fixed by the community if the game were as open as say an Elder Scrolls game.

        • VelvetFistIronGlove says:

          Not at all. The Cradle is still played in the same way as other Thief levels—avoiding (or disposing of) hostile AI, collecting items. The only thing it really subverted was your expectations in the first half—and that was a trick that only works the first time you play it.

          It’s a great level, but not because it alters the way you play.

          • Justin Keverne says:

            I disagree. The Cradle doesn’t change any of your abilities or the tools at your disposal (outside of the flashback sequence) instead it makes most of them useless. None of the walls within are climbable, there are no torches to extinguish, the few hostile creatures you do encounter are not invulnerable but their behaviour, senses and weaknesses are unlike any you’ve encountered before. If we are to follow Andy’s establish metaphor then the Cradle forces you utterly into the negative space by rendering impotent most of the tools you have. Tools that, given it’s late appearance in the game, you will likely have begun to develop a mastery of.

            It combines this with a narrative that introduces particular areas as dangerous, through aesthetics and explicit story beats, only to then require you to visit them to finish the level.

          • VelvetFistIronGlove says:

            That’s simply not the case. The puppets’ behaviour is the same as all the other AI. Yes, their animations are bizarre, inhuman, and (for me, certainly still) chilling. And they make the lights flicker, which is a wonderful addition to the atmosphere. But they patrol the same; they have the same alert states as an ordinary guard. And they have the same weaknesses as zombies (which are my second-most feared enemies in the Thief series): holy water, and flash bombs.

            When you get to the memory part, then the NPCs there have no weaknesses (as far as I know), but in their behaviour and silhouette (if not their sounds) they are just like other guards. Because you can’t defeat them, the level gives you many chances to practice avoiding them in relative safety before you have to ghost (or simply run) your way past them in the end.

            That there are no torches in the level to extinguish is unusual (although if I recall correctly you can turn off some of the electric lights), but that’s just upping the overall hostility of the level. It’s by no means an inversion of anything you’ve encountered so far.

            The Cradle’s strengths are in narrative and atmosphere; it doesn’t do much of interest with stealth mechanics.

          • Mungrul says:

            Well I dunno bout you lot, but once the atmosphere of the place started to cement itself in my consciousness (and I’m talking about the first half here), it made me want to stay out of the shadows; this is what I meant by turning everything on its head.

            My first time through, I had no clue what to expect.
            I didn’t know there were no enemies in the first half. I just knew that weird things were happening in the dark, scary things, and it made me want to stick to the light. This was further compounded by an experience I suspect was a bug, but I’ve never heard anyone else relay.
            I saw a fireplace that looked like I could sneak inside it, so sneak inside I did, only to take damage along with a terrifying shriek piercing the underlying menacing sussurations of the building. I’ve never jumped so much playing a game in my life, and it set my nerves on edge, sending a strong, violent message of “Shadows were good. Now they’re bad”.

            To clarify, I was slightly disappointed with the second half of the level, and don’t really count it as my Cradle. That first half is mine. That’s the genius piece of level design conveying strong messages through the removal of everything familiar.
            And that’s why it’s best experienced the first time with no prior knowledge.

    • AndySchatz says:

      I *think* we are on the same page. I consider any interaction with the simulation a foray into the positive space. In an FPS, the player is CONSTANTLY in the positive space, in a stealth game, the game asks the players to carefully choose when to interact with the positive space, because one wrong move can thrust you fully into the positive space, which is where you are vulnerable. Which leads me to the second, point: I agree with your elegant description! What I was trying to point out is that stealth games are the reverse of normal games: the player is powerful when the world is quiet. And this creates the tension!

      Sorry if my argument didn’t communicate well… but I do believe we’re in agreement!

      • Justin Keverne says:

        Ah, I see. I simply misunderstood the metaphor, we are in agreement then yes. If I differ it’s than I feel indirect influence upon the simulation is a form of influence and could arguable constitute an entry into the positive space.

        I do thought think it’s slightly misleading to say that FPS games have the player constantly in the positive space. One of the strengths of Halo and the original FEAR, and what makes them markedly different at the design level from the likes of Call of Duty is that they allow the player much greater freedom of when to initiate action. Look at the encounters in Halo, in the majority the player will start undetected, combat will only be initiated when they take action. The start of each encounter is with the player in the negative space. What sets it apart from stealth games is that Halo players have zero in-game power while in that negative space, they can plan and wait but they can’t take action even indirectly.

        • AndySchatz says:

          In my head the metaphor was a little longer: In many stealth games the player lives in the negative space, but he chips away at the positive space, one piece at a time. Little forays into the positive space allow him to create more areas of safety. It’s the difference between being a painter and a sculptor.

    • VelvetFistIronGlove says:

      I understood him to be borrowing “positive space” and “negative space” from visual design terminology as a loose metaphor. In visual design, positive space is the shape that is drawn; negative space is the shape that is formed around it. In Andy’s metaphor as I understood it, positive space is the “shape” of the active agents in the world, the patrolling guard NPCs; and negative space is the “shape” around them. The “hero of the negative space” is rarely interacting directly with these other agents, but instead avoiding them, or interacting with them indirectly (e.g. distracting with a thrown object).

      So Frogger would be an example of a game with this “hero of a negative space”.

      Manipulating the environment, whether by silently disposing of guards, turning off lights, etc. does not really change this, it just chisels away parts of the positive space. The ideal solution in all stealth games involves not setting off any alarms, or having any hostile NPCs become alerted to your presence—“ghosting” (as it’s generally understood, rather than the stricter meaning it has in Thief circles). The ghosting player still remains entirely in the negative space—the world simulation carries on, unaware of their presence.

      Edit: I see Andy himself replied before I finished writing this. Oh well.

      • AndySchatz says:

        No, but actually your description is exactly what I meant (I do have a graphic design/art background). So thank you for stating more explicitly what i was implying with that terminology!

      • Justin Keverne says:

        I can’t agree with your concept of the “ideal solution”, but other than that I think I follow the metaphor better now.

        • VelvetFistIronGlove says:

          I mean as an ideal that a player, trying to be stealthy, would strive for. From a design perspective, the failures to maintain that ideal are at least as interesting and important.

    • Bart Stewart says:

      My impression is that another way of describing positive and negative space is the gameplay-centric terms “engagement” and “avoidance,” respectively.

      An engagement-focused design defines rewards and penalties favoring interaction with the gameworld. Andy’s example of FPSs is exactly the one I was going to use. When you hit a trigger, scenario elements fire that basically force you to engage with the world. Engagement play is a natural approach for both game designers and biz folks: you want players to see all the cool content you spent money and creative effort to produce.

      Avoidance-focused gameplay demands a very different way of designing for a different playstyle. You’re basically making the whole world dangerous (player: “should I try to grab that shiny thing? will that person attack me if I say the wrong thing to him?”), then you’re rewarding the player for successfully avoiding negative interactions with that world.

      That’s probably why stealth games are seen as boring by many gamers, who’ve been conditioned to believe that if the player is not frantically mashing buttons all the time, then nothing is happening and the player is “passive.” But it’s not true that nothing is happening at some moment of a well-designed stealth game. It’s more likely that the player is feverishly trying to perceive multiple bits of tactical state information by closely observing visible and audible cues. Watching a guard’s patrol pattern; listening for whistling or footsteps; assessing whether sneaking speed will get you back to cover before you’re visible to the next guard — doing well at these tasks demands a nimble brain as well as nimble fingers.

      The fact that there’s less overt action/interaction in an avoidance game doesn’t mean there’s not a lot of gameplay happening. Avoidance play is simply a different, less outwardly visible form of play than engagement play. Designed and implemented well, both kinds of play can be very satisfying.

    • Dan Hindes says:

      That’s an interesting point about the lack of ways to interact with Dishonored’s simulation in relation to Thief. Much of those ways don’t matter, however, given that stealth mode is silent on all surfaces, and lights cannot be put out (with the exception of candles) despite shadows only mattering at distance.

      Would Dishonored have been a better game for stealth players with these factors? It certainly would have been more Thief-life. But the point was to make Thief fast, which I think the game does quite admirably.

      Perhaps there’s something to be said about catering for slow-stealth, as well as fast-stealth, on top of combat, and everything in between. The nature of Blink would certainly change if you were blinking over metal grating, from carpeted surface to carpeted surface, rather than from rafter to rafter.

    • Post-Internet Syndrome says:

      Hah, I came down here to say that I really liked the positive/negative space idea and that it neatly encapsulated the difference between stealth games and “normal” games, but I see that not everyone found it as neat as I did. :P

      I agree that games like Halo contain gameplay within the negative space, but it’s a spectrum isn’t it? Probably most games make use of the negative space; what makes a stealth game is just that a very large portion of the game takes place there, and that entry into the positive space is mostly something to be avoided, or at least strictly controlled, because of the player character’s weakness there. Inversely, I would say a game like Dustforce compels you to stay within the positive space at all times (since non-interaction resets your combo bar) while Halo, FEAR, Far Cry and other games with “complementary” stealth is somewhere in between.

  5. db1331 says:

    I thought we had blacklisted Blacklist.

  6. Caenorhabditis says:

    Most games will make you (your avatar) feel more powerful than the other (A.I) characters. One of the main differences between stealth games and, let’s say, a traditional FPS or adventure game is that the latter mostly make you feel faster and stronger than your opponents, while the former make you feel smarter. I like feeling smart.

  7. Ljud says:

    I would like to raise one question: Should we have the option of seeing trough walls in stealth games?
    The interesting thing about stealth )for me) is not knowing. It has some kind of survival horror charm to it. You are usually far more underpowered in stealth games than in shoot em ups, so you have to hide, peak around corners, avoid cameras etc. This was part of the charm of Thief and Splinter Cell series.
    But now you have this x-ray vision in all the stealth games. Batman: Arkham Asylum, Deus Ex: Human revolution, Hitman: Absolution, Splinter Cell: Conviction, Dishonored etc.
    What is the point of having the option of leaning and peaking trough keyholes in Dishonored, if you can use dark vision and you can see all the enemies on the map?

    • Snidesworth says:

      After picking up Dark Sight on my first run through Dishonored I’ve sworn off using it again. I spent half my time with it activated because it’s so incredibly powerful from a stealth perspective. It does have limits of course, being somewhat short ranged, but sneaking about is a lot more tense if you can’t simply fall back on it to tell you where where nearby enemies and which way they’re looking without exposing yourself.

      • SkittleDiddler says:

        My experience with Dark Sight was a bit different: I found that having it turned on for extended periods made the game more difficult because of the way the color scheme affects terrain and opponents that are outside of the range of the “glow effect”. I ended up using Dark Sight sporadically and only in areas where precise navigation wasn’t necessary.

        After finishing Dishonored, I was left with the impression that Dark Sight didn’t quite fit into the game the way the developers wanted it to. It’s a bit of an oddball when compared to the rest of the powers.

      • Highstorm says:

        I’ve had a hard time getting into Dishonored so far, and I think Dark Sight might be to blame for some of that (though XCOM takes most of the blame). It just seems to make it too easy. Guess I should muster the willpower to simply not use it, and try again.

        • VelvetFistIronGlove says:

          Dark Sight is too explicit: you see NPCs as clearly as if there wasn’t a wall between you, and you see a very explicit indication of their vision cone. It’s revealing the man behind the curtain a little too much, especially on level 2 when it shows security systems and their details too.

          I greatly preferred the way Oblivion’s Detect Life spell worked to the same end: you get a vague, hazy blob—which you can see consistently on all life whether behind walls or not, and whether hostile or not—that gets fainter the more distant an NPC is. It still reveals where potential enemies are, and which way they’re moving (if they are moving), but nothing more. It’s a great help, but doesn’t become a crutch.

      • Justin Keverne says:

        It’s biggest drawback is range, but I don’t think that’s a good enough balance for what can become “Matrix vision” for AI awareness. :(

    • AndrewC says:

      Until games can do positional audio good enough for us to get information on where guards are, construct levels that enable hiding while not restricting our sight, or make hiding in a closet for 60 seconds at a time fun, such mechanical ‘cheats’ will be necessary.

      And, as far as ‘cheats’ go, it is a super cool one!

      • Jason Moyer says:

        Thief did all of those things. Nearly 15 years ago, even.

        • Kilometrik says:

          pffffft. Iook, i liked Thief, but you are kidding yourself it thief had a good sound imaging system (no game has) and the enemies communicated well enough that they saw you or didn’t. THe AI had the same frigging position for when it was alert and when it was ACTIVELY looking for you! No different face, no different phrases. So you never knew if they were just looking for SOMETHING or specifically looking for YOU unitl they found you. Hell they didn’t even had a state which communicated to the player that he had been heard. It was Normal standing or Cautious walking. And cautious walking was shared by about 3 or 4 (THAT’S WAY TOO MUCH) AI states of awareness. You could become a master with the system by experimenting enough, but that involved WAY too much Trial and Error. Much more than Metal Gear, Deus Ex or Arkham City, which relied on interface conventions or empowered the player to enhance the communication of it’s rules and game systems. In Deus Ex you had nano augmentations that enabled to become hidden again with some modicum of ease. You could be killed if you lacked energy (which happened very often) or if you just weren’t quick enough, but that’s it. Metal Gear was great at communicating the state of the AI but very gamey (as the guy behind monaco said: IMMERSION BE DAMNED!).

          • Justin Keverne says:

            Did we play the same Thief? O_o

            It’s was one of the first games to model sound propagation and occlusion, though it did have problems with sound being audible between different floors of a building. The use of different surface materials combined with different footstep sounds meant you could observe a space once and then calculate where any AI was within it by the sound of their footsteps.

            AI had two standard walk cycles true, but one was explicitly reserved for when they were actively searching. Each AI state, and there were several, had banks of different phrases related to them, there were specific lines for whether the AI had seen something, heard something, spotted a body, was actively searching for the player or had stopped searching for the player.

            Is all that difficult to learn? It can be certainly but not because it was lacking in feedback.

          • Jason Moyer says:

            “you are kidding yourself it thief had a good sound imaging system (no game has)”

            The Dark Engine did sound propagation better than anything that’s come since. You can tell exactly where NPC’s are in the first two Thief games just by listening to them. You can get distance/direction from the player, general direction they’re walking in (i.e. you can tell if they’re getting closer or further away), and the materials they’re walking on.

            The excellent use of sound in Thief is one of the reasons why it’s still the best stealth game ever made. If someone made a modern game that used sound that effectively and combined it with forced non-lethality and the player being completely vulnerable when not hidden I’d explode with joy.

      • Dervish says:

        “Necessary.” Yeah man, it’s totally impossible to play well without them. And probably impossible to think of alternative mechanics, too. And there are certainly no stealth games that don’t have such abilities, even though Thief was already mentioned.

      • Ljud says:

        Its interesting that in games like Mark of the Ninja and Monaco you don’t have the ability to see everything, but you think it can’t be done in AAA games.

    • AJ_Wings says:

      Probably because of a lack of a true dynamic shadow/light mechanic ala Thief 2 and Splinter Cell Chaos Theory. Wallhack mechanic weren’t necessary in those games because you were given plenty of dark areas to hide and blend in. If timed right, it was relatively easy to go from one hiding spot to another in those games. Modern stealth games rely a lot on bits of cover in the environment for hiding and a mostly binary shadow/light system like Conviction.

      • KenTWOu says:

        Probably because of a lack of a true dynamic shadow/light…

        The answer is so obvious and simple, those games have more enemies around you! It’s much harder to observe three or five enemies simultaneously.

    • Kilometrik says:

      And in regards with Dark Vision. I think the tradeoff is interesting, because you can’t see the awareness of enemies beyond 5 meters. This has made me die on countless ocasion since enemies that are further away than 5 meters can stil (and usually WILL) see you. An interesting compromise would be that you not only cannot see their awareness, but you just cannot see them AT ALL beyond 5 meters, that way Dark vision could be only used to deal with really close enemies but you need to turn it off to plan a longer route, which is when leaning comes into place. Perhaps that could be modded in. Also, in a game with that MUCH verticality you cannot expect Lean to be useful in every ocasion. Such as hanging from a lamp and stuff. What’s more, maybe not Dark Vision, but that mechanic serves another purpose as well which is to be able to see your soundwaves and distract enemies with efficiency. Perhaps that could have been it’s own mechanic.

    • Naum says:

      I also tend to find those wallhack mechanics somewhat dull and therefore didn’t use them in Dishonored and DX:HR. However, that’s not because I’d like the thrill of blindly moving into a potentially hostile space — in fact, I hate it — but because it makes avoiding that very situation too easy. A good part of the joy in stealth games for me comes from collecting knowledge about a section of the map before moving into it in relative safety, and if that becomes trivially easy, it takes away from the gameplay.

      By the way, that’s also the reason why I dislike the lack of a vertical equivalent for the leaning mechanic in Dishonored. It occasionally makes blind upward blinking more or less necessary, giving me no option to find out what nasty creatures may be on the next floor except for trying it out (or using Dark Vision).

      • Ljud says:

        Agreed. There would be more interesting to have some sort of camera for vertical sections like the one Thief 2.

    • tomeoftom says:

      I just chose not to get Dark Vision on my first playthrough (which was a nonlethal almost-ghost run). It worked perfectly for making me feel disempowered and vulnerable at all times. Every enemy was a genuinely scary thing.

      For my next playthrough, I was a hyperviolent predator and the Dark Vision actually made it a lot more fun – being able to plan slick combos on guards based upon where the are, and being able to surprise them from behind, worked with the fiction of the character I was playing.

      Perhaps the solution would just be for the game to emphasize on its part that Dark Vision is the aggressive form of stealth?

  8. AlwaysRight says:

    Next time: multiplayer stealth?

    I hope you mention Splinter Cell: Chaos Theory’s co-op multiplayer, I’d LOVE to see more of something like that.

    • TrueSpite says:

      Apart from the netcode and the lag it caused SCCT’s versus multiplayer was some of the most engaging MP I’ve ever enjoyed. Asymmetric multiplayer has always fascinated me, and SCCT was very much asymmetric. The armed-to-the-teeth mercenaries vs almost non-lethal acrobat ninjas dynamic hosts this great relationship, where it’s beyond pure twitch skills (though those did help a lot on the merc team… insta-scoping and such) but more of a mind game, where even the hunter can get his neck broken if he slips up.

  9. ocelot113 says:

    a number of other stealth-game luminaries: Patrick Redding, Game Director on Splinter Cell: Blacklist, HAHA riiiiiiight

    Mark of the Ninja is a much closer representation of stealth, the old SC games had, than Blacklist. Actual stealth was shot in the face and dumped in a ditch for Blacklist, which is clearly a straight forward action game.

    I judge you RPS for even allowing someone that is on the Blacklist project comment on stealth. Blacklist is the last of the decay in stealth gaming. MoTN is what stealth action IS.

    Challenge, problem solving, planning, timing, and manual execution is what creates the payoff.

    • MondSemmel says:

      Did you realize that the guy (partly?) responsible for your epitome of a stealth action game, Mark of the Ninja, was the one who asked the Splinter Cell: Blacklist dev for his opinion? I don’t see how RPS deserves any blame here. I imagine RPS could refuse to host such an interview (but why would they?), but I’m quite sure they wouldn’t censor it – and if they had tried, I don’t see why anyone would agree to that, rather than find another gaming site.

      • ocelot113 says:

        Yeah, just my fanboyism coming out. It’s really nothing against Redding I was just devistated at the turns that SC has taken. I’m sure the guy is brilliant, I’m just not really qualified to talk about any current SC games i guess because they all feel tainted. I loved SC:CT and too see how the series has turned from stealth especially Double Agent (which I feel is the conceptual predecessor [stealth in the open] to Blacklist).

        So again, I really don’t have anything against Redding just LOVED the SC series and to see it, imo mutilated makes me writhe.

    • Henke says:

      Blacklist looks like it’ll play pretty much like Conviction, which may not have been a straight up stealth game but I thought it blended stealth and action very well. Honestly it may well have been my favourite entry in the Splinter Cell series.

    • ffordesoon says:

      As with stealth games themselves, what makes for an interesting roundtable feature is tension. Maybe you would read an article where everyone’s patting each other on the back for being so great, but most people wouldn’t. The only thing you would learn from such a piece is that everybody who makes stealth games thinks everybody else who makes stealth games is just terrific. Nobody would have to argue for the superiority or validity of their approach, which would make for a dull feature indeed. Here, we have an intriguing range of approaches to the genre.

      Redding, being a representative of the New Stealth (which I’ll call Immersionism, because that looks to be the key schism between the two groups in this series) while also having had a hand in some of the great Old Stealth (which I’ll call Symbolic) games, is an excellent choice, because he offers an Immersionist counterpoint to the two (if I may coin a term) Symbolicists’ views. Colantonio, meanwhile, is simultaneously Immersionist and Symbolicist, which means he acts as an interesting mediator between the two schools.

      I don’t expect Redding to challenge Schatz to pistols at dawn or anything (it’s obvious that they’re fans of each other’s work), but since everyone’s natural inclination will be to defend/champion his approach to stealth design, the discussion will be more interesting than one of those tedious lovefests you occasionally see Hollywood actors participate in (“…And then I realized that we both like memorizing our lines! Ha ha ha, how great is that!?”).

      P.S. If you’re wondering why I didn’t call it “Symbolism” instead of “Symbolicism,” it’s just because something about “Symbolicism” sounds cooler to me. You all may shorten it to Symbolism if you wish.

      • Nelsormensch says:

        Also, really glad you appreciated that, because I very deliberately tried to get a diverse group of voices and opinions.

    • Nelsormensch says:

      Yeah, I invited Pat (and set the whole thing up really), so don’t degrade RPS for that. But I did invite Pat very much on purpose, because he’s a really smart dude that thinks deeply about this stuff and I respect his opinions a lot. His work on Far Cry 2 (still one of my favourite games) was tremendous and the bit of SC: Conviction that he was responsible for (the co-op) was the best part of that game, hands down.

      I haven’t played Blacklist (not sure if anyone not at Ubisoft has), but I know that Pat’s opinions and thoughts are top-shelf and that’s why I wanted him involved. If you ain’t hot on Blacklist, that’s fine, but I hope you’ll at least give what the guy has to say a fair shake.

      • VelvetFistIronGlove says:

        Good, flowing stealth design and good, flowing combat design have a great deal in common—particularly in the way they use movement through space. Far Cry 2’s combat (not always, but often) walked the line between stealth and combat really well. It led to a highly dynamic and reactive use of cover, concealment, diversion, and environmental tools that smoothly segues between stealth and combat.

        Which is to say, that any other games aside, Far Cry 2 alone is a good enough (if nonobvious) pedigree in stealth.

      • KenTWOu says:

        …the bit of SC: Conviction that he was responsible for (the co-op) was the best part of that game, hands down.

        SC:Conviction fans truly believe the game was saved from oblivion by Patrick Redding and his team.

  10. tomeoftom says:

    This was great! Keen for part 2.

  11. Hmm-Hmm. says:

    A great read altogether. Thanks RPS and devs! Looking forward to the next part.

  12. MrUnimport says:

    I find myself disagreeing with the suggestion that “humanlike” AI is something to be avoided in stealth games. When I think of the most memorable experiences I had with, say, Splinter Cell, I think of the times I was crouched in a shadowy corner, pistol out and pointed square at a terrified guard’s chest as he blindly probed the darkness in front of him. The thrill of seeing if he’d stumble just close enough to detect me, or if he’d lose his nerve and go back to his patrol instead. I don’t remember all the times I chucked bottles at empty bits of courtyard so I could lead a guard on a merry chase far away from his position.

    If you make guard AI more mechanical, more communicative to the player, more easily manipulable, you reduce the game from an infiltration against impossible odds to a big puzzle, and the satisfaction of cracking a puzzle isn’t quite the same kind of payoff. This is the same reason why I think quicksaves remove a lot of the fun of a sneaking game: the unwillingness of a player to settle for anything less than a perfect run means he’s interacting with the game on a purely systematic basis, instead of weighing stealth vs safety, lives vs. the threat of losing progress, it just becomes an exercise in trial and error until the “correct” solution is found.

    • Kilometrik says:

      Except that the thrill you speak off is not to be found in any of the game’s system as an objective property. It was your subjective, personal and, quiet frankly, impossible to reproduce experience. Most of the times what people feel if the AI is not communicating or being systemic is that the game just arbitrarliy decide to dick you over. You didn’t lost due to your mistakes (and therfore you don’t know how to master the system) but due to the game being an asshole. Also, a good systemic, communicative, game is not a puzzle because there isn’t a solution. There’s just a goal. Being found, activating alarms, having to kill some dudes. That’s all part of the game, it’s not a fail state. It’s not about “How to pass around these guys without being detected” But “how to get to the other side of the room without being killed” If you are extremely vulnerable and fragile and the game gives you tools to manipulate and use subterfuge to your advantage, then it’s a stealth game. If the game gives a bunch of guns, a ton of health and says “go gung ho” it’s an action game

      • MrUnimport says:

        I suppose the difference is whether you want atmosphere and immersion from a stealth game, or you want a playground in which you can outsmart AI.

        • Sheng-ji says:

          I think what they mean is that the game systems while based on human bahaviour are not trying to simulate it: take the example of a guard who hears a suspicious noise in the corner: In the game his alert state is raised according to how the game has dictated and he follows his scripts – for example looking in that direction, perhaps searching.

          In real life a guard may ignore a suspicious sound the first few times he hears it or he may poke around. He may obsess about it all night, sporadically poking around, or it may be forgotten in a second. That kind of unpredictability while more realistic that the concept that you can know what is going to happen if you make a noise does not make necessarily for a fun game. I think, at this stage, if you made a game like that, it would frustrate too many players as they are very much acclimatised to the idea that AI’s in a game are predictable.

  13. Henson says:

    I’d say that stealth games are less “pull” and more “balance”. Trying to carry an overfull glass of water without spilling a drop.

  14. Radiant says:

    Redding run for your life; you still have time!

  15. says:

    Great read. I feel smarter just reading the words of those smart, creative people!
    Love the notion of game’s negative (and/or “pull”) space – the beautiful unsung hero of most stealth games – a dark corner!

  16. fish99 says:

    As it happens I’m currently replaying Thief Gold, with the new 1.19 patch and also I have it working in stereo 3D (nvidia 3D Vision) for the first time, which definitely adds something. Then I’ll be moving on to Thief 2 and System Shock 2.

  17. ffordesoon says:

    Fascinating stuff!

  18. Suits says:

    Splinter Cell Chaos Theory..if only *sigh*

  19. ocelot113 says:

    “…but we have to take into account that the player can fall back to combat at any time, and we must make sure he still has fun if he does so and doesn’t feel punished for falling out of stealth.” – Raphael Colantonio

    While I do agree that is what modern games try to do (all fun at the cost of achievement) I believe it makes the hardcore feel like there are no good games out there. The lack of punishment for doing something incorrect is great for most people that like casual games, but for those that want a sense of accomplishment from their gaming, they expect that line to matter. I guess it’s mostly a balancing act that isn’t usually met and one style is usually superior pushing you toward it. I think stealth and action compliment each other very well, however I don’t think they can blend evenly while still creating the same tension and excitement.

    IMO part of the fun of stealth games is the “OH shit there is nothing I can do about this situation” and taking time to invent the master plan and quite literally orchestrating a masterpiece of diversion, destruction, and timing to succeed. When you add an equal part action you can just drop out of stealth and borderlands 2 fools, who’s payoff is not as rewarding IMO.

    I think SC Conviction tried to blend them better with the mark and execute but I think stopped short of the real goal. Personally I think it should have been more focused on diversion and manipulation over just flat out killing power. What made SC games great was the luring of AI into the perfect kill and with mark and execute it eliminates all the manipulation that stealth fans thrive off of.

    I guess it boils down to pure stealth fans don’t want a bailout. They want choices to matter and understanding a failure spurs on perfection. Achievement.

    MOST satisfying moment in gaming in a while was with Mark of the Ninja. I was in a room that was slowly filling with poisonous gas. Guards were pouring in and I was hiding in a doorway watching them all fill the room where I needed to escape. Knowing that an all out brawl was suicide I tossed out a spike trap and waited for one of them to hit it. When they did I jumped out of cover and onto the ceiling where i would not be hit by the ensuing massacre being carried out by 6-7 panicked guards. Dropped down and off the last survivor. Casually walked out with bodies all over the floor. Lol the best.

    • Naum says:

      Can’t the ‘hardcore’ players inflict that punishment upon themselves fairly easily when it comes to stealth? If you just reload whenever you’re spotted and/or kill an enemy, that’s the most severe punishment the game could come up with (excluding permadeath). Of course this is not an ideal solution because it circumvents the game’s own definition of a failure state, but I find it a very simple yet effective compromise between ‘hardcore’ and ‘casual’.

      Come to think of it, wouldn’t it be nice if stealth games offered some options to increase the difficulty by adding custom failure states, like a ‘ghost’ option that ‘kills’ you when you’re spotted?

      • ocelot113 says:

        A punishment is not only death. Making the game harder because you have been spotted is a penalty. Locking a door because you’ve been spotted. Sounding an alarm because you’ve been spotted. Punishment doesn’t have to be so extreme as permadeath. All I’m saying is there is less achievement with less punishment. That is my game play mindset. It might not be for everyone but then neither is causal play for everyone.

        And you missed the point. Implementing your own punishment and resetting every time is not engaging. We don’t play to beat the game, we play to perfect it, and resetting or doing things like that are a form of “cheating” if it isn’t a part of the coded game play. Being perfect within the rules makes the game fun, not just finishing it.

        It’s the difference between figuring out a pattern of numbers and just guessing the right number.

        • Justin Keverne says:

          Maybe I’m missing something but isn’t there a contradiction there? If your ultimate desire is perfection within the rules of the system wouldn’t that preclude the type of dramatic improvisation that you describe? I’m just not sure I see what you mean. Surely if you perfect the system you’d never get into situations that would require improvisation as you’d know how to avoid such situations? Failure states exist because player’s haven’t perfected the system. Or is it simply a case of continuly striving for perfection and never achieving it?

          Either way I wouldn’t recommend Dishonored to you, it’s not the type of stealth game you’re looking for.

          • ocelot113 says:

            I think perfection was too strong a word. When I think perfect I don’t think 1 singular way to play something but your optimal way of running it. I guess for me, well when I played SC:CT, I would run the boards over and over to find every nook and tunnel to play every possible outcome so that I could run through the game, that use to be a challenge, in a breeze. Making spontaneity, an easy decision because you know the system and platform so well. That’s what I think of as “perfecting” the game. But that requires the game to be a challenge in the first place, so you can see the improvement, which I feel most games are not doing, seems how dying is a punishment, and designers shy from punishing.

    • Post-Internet Syndrome says:

      Well it’s not a pure stealth game. It’s a game in which you are able to progress in different ways, stealth being one. It’s designed to make both fun. You are not required to like it, but criticizing it for not being another game is a bit silly.

  20. crinkles esq. says:

    Although I loved the MGS series (a shame Hideo Kojima isn’t in this roundtable), I didn’t often use stealth unless I had to. In general for me, the stealth option in games too often feels like a puzzle game, in that “hiding” from AI units feels like a logic problem to solve, and not a reaction to a situation. And the logic problem of being stealthy is too often something like Portal — where there is one and only one solution — due to the limited number of options the player is given to avoid being detected by the AI. I think this ties in to what has been said about the evolution of games being ‘necessarily’ cinematic, which limits the breadth of interaction.

    I think another problem with attempts at superficial realism is that there is an ‘uncanny valley’ not only in the visuals, but in the AI brain. Especially as AI hasn’t kept pace with visual advancement. When this high-poly person acts like a bunch of AI scripts instead of acting unpredictably, my desire to interact with them plummets. There’s a cognitive dissonance there. Whereas with abstracted characters as in Mark of the Ninja, my brain understands it is a game character, and I am fine with the limitations of its actions.

    On a side note, reading that much sans-serif, italicized text really hurts my eyes.

    • ocelot113 says:

      Stealth games are a puzzle game for 2 reasons. AI is not complex enough to clone reality and people talk about not wanting to punish gamers…. a real unorganized non-system driven AI would be so much more unpredictable than an easy to comprehend and reliable system of AI rules.

      People say they want real AI for stealth games. The prepare to die and not know why.

      Part of the problem with stealth games is that we have a limited number of inputs to the gaming platform. When using real stealth there is sooo many context sensitive things that we just don’t currently have the technology to reliably control. Voice, breathing, and walking volume, dynamic camouflage in lighting, step placement with full context object interaction.

      For instance in SC games you could use bottles to cause distractions and you relied on the guards to go check it out. But what if the guard AI decides that it saw the bottle and knew right where it was thrown from. You have a 50-50 shot to move on or be dead, and that’s not what I’d consider fun. Stealth games require the knowledge of patterns and systems to be playable. You could hide the systems more I suppose, but implementing a dynamic and evolving AI without giving the player the ability to be dynamic and evolving in the interaction and input would just frustrate people, imo. It would require 100% simulation, allowing you to control and effect everything in the world in any way, which we are no where near.

  21. Jamesworkshop says:

    Rocksteadys batman games worked perfectly in regards to the problems of requiring perfect stealth, I can’t think of almost any other game not just stealth ones with that many options that all felt perfectly balanced in their usage.

  22. Bart Stewart says:

    To what extent does the baroque geometry of a stealth-enabled game fit the recent RPS observation that PC games are all about terrain?