Splinter Cell: Conviction Q&A Feature

At an event showcasing the multiplayer modes within Splinter Cell: Conviction, TVG had the opportunity to sit down and put a few questions to the game's director,
Patrick Redding.
 
Read on to find out his thoughts about AI evolution, the dearth of the pure stealth experience and what Project Natal could bring to the Splinter Cell series.

The multiplayer appears to be staying away from the adversarial side of things apart from one mode, is that because stealth doesn't lend itself to that type of gameplay?
 
Stealth by itself can lend itself really nicely to adversarial gameplay, but it's really what Conviction is specifically trying to do, which is we've given the player these tools like 'Mark & Execute' or 'Last Known Position', which are really designed to allow them to feel predatorial, like they're able to dominate their enemies in situations where they're outnumbered. 
 
The reality is that we carefully looked at all the ways we could do that for multiplayer and it was immediately obvious that co-op would be a better fit, just for the simple reason that you're not going to give players fighting against each other the ability to do Mark & Execute because it would be pretty frustrating and anti-climatic as well.  Similarly the Last Known Position tool is a way of taking something really abstract, which is how does the AI character know where I am, and visualising that isn't necessarily going to work with human players. 
 
What it would turn into is a straight skill versus skill engagement, which certainly Spy vs Spy is fun to do, but I think it would stop being about that predator in the dark and become a lot more about how agile am I, how well do I know the maps.  There's a place for that, it's just that is less of what Splinter Cell: Conviction is really trying to push for.
 
Stealth has traditionally been a single-player experience...
 
Absolutely
 
But co-op is obviously becoming more important in this generation; has it thrown up any issues with the AI?
 
The way I like to put it is that, I have no scientific basis for this beyond my own twisted intuition, but I've always said the challenge of having two players in stealth mode using the tools that Sam Fisher has, is that two players are four times as likely to be exposed.  Because of the complexity of what they're doing, taking actions, and the timing, using different paths, different styles of play, it doesn't just double the amount of exposure it's literally four times the amount of exposure. 
 
That's why you'll notice in our maps, even though they're still very linear as you go from objective to objective, we open things up a little bit, create more multiple paths and choices as to how you're able to advance through a particular zone.  We need to do that because obviously our AI has to not only be able to detect one person, but potentially switch targets from one to another, when one guy disappears and another appears, make sense of that and be able to differentiate between these two characters and not just assume it's one, or in some cases if they don't know it's two characters only assume it's one. 
 
These are the types of complexities we have to deal with if we want the AI to seem realistic and to seem like it's intelligent.  One thing's that great is that we work incrementally. We work in iterations, so we can start out, work on our maps, add our AI, we can see how a certain number of AI functions in a certain kind of space and under certain circumstances whether it's action or stealth.  Then as the AI team is constantly making improvements and making it more efficient, then we can lower the numbers a bit and play it again and reset it.  So we're constantly iterating all of that, that iteration cycle is the most fundamental - that and where we're putting lights. 
 
So what you're saying is absolutely correct.  That is always a challenge, being able to take all of the same advancements we're making for the AI in the single player and make sure it works for two players.  I think that the good news for us though is that two player co-op in this stealth situation is in a way really optimal; it's not three it's not four, it's really two because as soon as you start getting three player co-op in a situation that's as tactical as Splinter Cell it really starts to feel like a squad game.  Suddenly three guys is a squad, two guys is like two guys working as solo operatives working together, whereas three guys is a squad and we're almost getting into Rainbow Six or Ghost Recon territory.  So in this stealth orientated environment two players is an optimal number for us, and it's not as difficult to adapt the AI.
 
The stealth genre was huge back on the previous generation, but doesn't appear to be so prevalent this time around.  How challenging is it for Sam Fisher to survive in the current climate?
 
Do you mean the current market climate or the current game environment?
 
The game environment.
 
It's interesting and you're right, it's not really a coincidence that the timing of this evolution is happening kinda weird in terms of the generation of the consoles and also the titles that are coming out.  Stealth was always a more binary concept on the last gen, you're either hidden or you're not, you're either in light or shadow, they can either hear you or they can't.  As a result the gameplay model could be a lot more binary and a lot less forgiving in some ways.  Here we're creating a world for the player that really has a lot of life and realism, a lot of nuances and suddenly a lot of grey areas, so players have this expectation that there's a bit of fuzziness to the system, that they're able to test the boundaries a bit, back off and not get caught. 
 
I think that's been the challenge, to make sure there's enough of a range of actions in the AI, a range of feedback that we can provide the player as they're moving around the environment.  That's why we shifted towards using the kinds of in-game feedback and in-game presentation that we're using instead of putting things in the HUD.  It's based a lot more on a player's intuition, they can use their eyes, they can use their ears, they can hear whether they're making loud noises, they can hear whether guys are talking and that's kinda the trick, we make sure we're bringing all of those tools. 
 
One of the things about Conviction that we're super proud of is that the whole presentation layer of the game, whether it's the text in the world, using projected video, the camera design that our presentation team has been developing, is really designed to illuminate the stuff that you're talking about.  It helps the player have those skills, that sort of extrasensory awareness of the environment that a guy like Sam Fisher would have but you and I don't have because we haven't been trained in that way.
 
Jumping form that question, everything has moved forward, not only visually but AI such as the flanking in Gears of War or the squad play in Call of Duty, is it hard to follow that in a stealth game?
 
Maybe a little bit.  Having worked on shooters in the past I know it becomes a more pure case of war gaming, because things tend to be more upfront, it's a stand up fight.  You're not trying to create conditions necessarily in which the AI doesn't know the player's are present, so that certainly represents a part of it.
 
What we're doing is having to extend the normal set of AI states, I'm getting a little design geeky here, but extend the set into a bigger range that includes various degrees of knowing something is going on but not sure what, knowing someone is there but not sure who, knowing exactly who it is but not sure where.  All of these additional states have to be modelled, designed and coded and debugged.  So that certainly raises the bar for us, but I think what is cool about it is you can be in a stealth scenario, the requirements in terms of, for example, hitting the player with huge number of enemies for example isn't there as much.  You can do a lot more for less because an individual AI can be plenty dangerous and plenty threatening. 
 
So what it becomes about is slowly stripping away and stripping away until you have the smallest number of the most intelligent AI that you can use for the purpose of supporting stealth gameplay.  Then you can tune that number up if you want to challenge the player a little bit more, if you want to put them into a really stressful situation towards the end of the game or if you want to support a higher difficulty level.

 
The blending of action and stealth was evident in the most recent Metal Gear Solid and seems to be the case with Conviction...
 
Absolutely.
 
Is there room for the absolute pure stealth game anymore?
 
I think that's entirely... if we're doing our jobs correctly then it should be self correcting, in the sense that we should be trying to develop and make sure our gameplay systems support a really broad range of player choices.  That it's possible for an old-school Splinter Cell fan to jump into Conviction and say 'you know what, I'm going to play the way I played Chaos Theory or the way I played Pandora Tomorrow, I'm just going to go through this thing and nobody is going to know I was here,' 'I'm going to wait the necessary time for that patrol to pass, I'm going to very carefully disable the lights and cameras'.
 
I think we tried as much as possible to make sure that option is available to players, but we didn't want to go super hardcore in terms of controls for example.  One thing we do know is that can be a real deterrent for people to get into this kind of game, but I think as much as possible we wanted to retain that as one style of play.  It's not so much as saying 'we're going to make Splinter Cell like Gears of War', it was a lot more like saying 'let's make sure that it's possible for the player to push the envelope a bit, screw up a little bit and bring the full weight of the AI on them but still be able to survive that by disappearing, breaking the line of sight, getting the hell out of there'.    Having that heart-pounding moment of 'that was close' and then slipping back into the shadows again.  I think that kind of dynamic, that's why we talk so much about planning, execution, and vanishing, because if we can constantly make that available to the player then they feel that it isn't so stressful if they get into a firefight once in awhile, it's pretty exciting and helps to break up the monotony.
 
Project Natal is obviously a big thing at the moment; I'm not sure whether you've looked at it a great deal, but can you see any application for that technology in a future Splinter Cell game?
 
You and I can probably spend the next two hours brainstorming stuff like that, absolutely.  Certainly amongst the design community at Ubisoft we're very interested in Natal, we very interested in the Wii and DS, and various other advancements in terms of interface.  I think every designer could come up with a laundry list of interesting ways to apply it to stealth gameplay, Splinter Cell in particular.  I don't know if there's anybody actually thinking about it as their job description at this point, but nobody has asked me yet so you're the first.
 
There seems to be a considerable amount of uncertainty as to whether Natal will be going after the casual Wii market, or a way to ditch the controller for hardcore gamers.
 
Absolutely, as always it's a bit of a Craps shot as to whether a particular technical development is going to end up representing some major sea change in the way players are interacting with games.  I certainly have always hoped that we would start to see steady advancements in those areas.  The reality is the output of our games is getting richer and richer and you hear Yannis Mallat our head at Ubisoft Montreal talking about the convergence of media, not just in terms of how people are watching movies or playing games, but just in terms of production, being able to use a lot of the same assets the movies are using. 
 
Our games are becoming really realistic and getting a lot of fidelity in terms of how immersive they are, I think that the next step, beyond obviously advancements in AI, graphics and so forth, is going to be how do we help the player have a bigger, broader range of input.  Day-to-day the controller that we're using is pretty similar, it's an incremental evolution over the controllers that we had almost twenty years ago.  So I think we can find a way to make input more gestural, more expressive so that players are able to add a bit of subtlety, a little bit of nuance to how they're giving their instructions to the avatar in the game world.  If we can do that then we're opening the door to lots of interesting game developments.  But this is far-future speculation in game development terms; it's more than six weeks from now, it's almost irresponsible for me to speculate.
 
You have the perk system in multiplayer which almost seems to be a necessity these days?
 
What's important for us is there's always that meta-game going on, especially for people who really get into a particular title and are looking for excuses and reasons to fire it up, they're waiting for us to give them that reason.  When we can say to them, 'look you like collecting Achievements that's awesome we've got achievements, you like being able to interact with the community with fellow players, that's awesome we've got that.'  But giving them different layers of challenges to try to complete, different ways of measuring success, their reputation, having an experience point system is something that we can steal from old-school RPGs and apply to a lot of different genres because people like that idea of unlocking.
 
In the context of the Tom Clancy universe it makes a lot of sense to apply that to gear and gadgetry, the technical side of it because it is very high tech.  In terms of changing the appearance of your character, it's nice to... I remember when I became a hardcore GRAW player, playing with my friends doing the co-op campaign it was nice to be able to be the dude with the baseball cap and the sunglasses as opposed to just another guy in a helmet and goggles.  Those little things allow players to develop a personal stake in the game and feel like they want to come back.
 
Jumping off from that, Splinter Cell was one of the first games to support Avatar clothing; Microsoft has spoken about such items being given away with Achievements, is that something you're considering?
 
You know what I don't want to speak about it because it's part of a large strategic thing between the single-player as well as us.  I think the focus from PR and communications will have separate packages.  I've seen some of the other stuff that's already out there on Xbox Live and I've dressed up my avatar in Third Echelon elite suits.  So I think we really believe in that model, we love the idea of being able to add to that with value added elements that are available to the community as a whole, because like I said it gives them more of a stake in the franchise and extends the experience outside of the strict confines of the game.
 
With the other Tom Clancy brands in the pipeline, can you provide any update on what those teams are working on at the moment?
 
There are absolutely teams working on other elements of the Clancy universe but I am totally not at liberty to discuss that.  But I'm sure if you talk to the fine folks either here at the UK or EMEA they'll be happy to send you the package on that.