Make Things Worse: Enabling Setbacks for Consequential Play

[Music] hi everybody thanks for coming down I'm Patrick reading before I launch into this I'm supposed to inform all of you I get five o'clock on day three that you're supposed to turn your mobile devices off at this point assume most of you've already done that no no videotaping a recording of this that's a prerogative reserved for the folks at GDC who are going to be doing that and I think other than that if you have questions at the end there are microphones that are set up in the aisles so you're encouraged to use those okay so I want to talk about a problem we have with lost time and to be clear I'm not talking about this kind of lost time this only Wednesday maybe by Friday this will be kind of more relevant I'm actually I'm talking about the players time and Wow okay in the spirit of this I'm going to back up a bit so when a developer you know has put on a few

years and has maybe shipped a few mass-market games they will at some point invariably have some version of this conversation why do we have the genres that we have you know why have mass-market games kind of settled into this particular landscape and the answer to that is some variation on this it's to maximize the production values and constrain the design space okay so Triple A games which also need to be legitimately interactive are expected to me kind of steadily escalating expectations in terms of sensory immersion in terms of accessibility in terms of visual quality and the like designers also need to be able to limit the range of the games outputs and minimize the amount of energy they spend dealing with edge cases you know dealing with the player colliding with the edges of the simulation and so there are only really kind of a few sort of games that meet both those criteria but to be fair

there are also plenty of games and quite a few developers including I imagine a few in this room who are who have managed to challenge that point challenge the first point kind of say that the triple a triple a Emperor is wearing clothes etc but what about that second point so if you haven't seen it in the sort of critical path segment of many kinds of combat I won't touch and sin talks about the fact that you know one of the reasons why game French honors had settled into the kind of preoccupation that they have with combat and life and death struggles is because we can't underestimate the value and taking people out of play and taking people off the board when we when you let the losing guys stick around it generates this kind of cascading series of problems that you then have to manage like well then what happens after that what happens next you know whereas when we we go with kind of existential conflicts we're able to really limit and constrain the scale of that problem and give it a give us the designers but also the player kind of a reset button there's another facet to it

this is Jesper Yul who wrote a book called the art of failure has spoken extensively about this and he points out that you know over the years we've also kind of trained up a generation of players for whom this is fun right like dashing yourself at the kind of Groundhog Day cycle of of trial and error is a kind of fun as long as the refractory period is short enough right and so it were at the stage now or even starting to make games that are offering a commentary about the fact that people find that sort of thing fun so this is the prevailing model of failure that we have right now I'd like to tell you that this is some kind of elegant step function but actually this is just a cliff you fall off the failure cliff right and and it's it's a frictionless unclaimed Abul cliff and the game state collapses into something that the player is incapable of reversing alright so so

let's say that this weird shape represents the kind of sum total of the possible states supported by the game simulation right now this isn't really a very rigorous or scientific model so light is simple every time every enough time passes and the player will eventually cross the magic threshold that tells the game that it should return a failed state so there's an interrupt the game resets itself and the player end up and the player responds now what's interesting about that is that that big orange dot that you see over there on the far left sits outside of that state space all right why is that interesting well because what it means is that ultimately the cost of failures external to the game right think about that I'm gonna be willing to bet that most of the people in this room put some kind of premium on the consistency of the player experience but what we're saying is that this one very critical recurring aspect of the player experience namely failure incurs a cost that can't possibly be measured because the value of a person's time is

completely dependent on who they are in the kind of life that they live right it depends on a number of things but it's safe to say that even if they're playing the same game but the cost for this guy is going to be different from the cost for this guy all right but that's what we have right now and implicit in that as a kind of bargain that exists between the player and the game right which says that in return for submitting themselves to this binary failure model players have an expectation that they can evaluate the risks associated with the actions that they're about to take and also that they have enough of an understanding of the game state that they can kind of manage its complexity can plan for it in other words the player says to the game says to the player I'm winnable now we also have to acknowledge and this is great right that in recent years there's been a shift you know we've seen new games all the time that are challenging the established assumption about what happens when a failure of player fails but we can think

of these as approaches that are trying to kind of sweeten the deal for the player all right they're still dealing with the currency of player time but doing it in a way of that's kind of it's because it's a way of offloading the problem they're trying to apply some multiplier to the failures effects on the game and make sure that it's worth more than just lost time and also at the same time start to bring back some of those consequences into the game state so what are some examples of that so we see in kind of persistent world games and in MMOs this kind of commitment to the idea that you know your resources are recoverable you know your stuff has a persistence in the game and that generates its own kind of tension you respawn and it immediately turns into a race to try to recover what you may have dropped before it slips into the lava as it were alright some games give the player a mechanic for managing the cost of their own potential failure while still staying within the kind of narrative logic of the game or the metaphor of the game and then tweak it to make sure it doesn't become an

exploit you know and we've seen an interesting evolution that you saw that between Bioshock 1 and Bioshock 2 something happens with the vitae chambers you know so when you use them there's this kind of reset factor to what's going on with the rest of the world and it becomes a way of preventing the kind of grind of going up against Big Daddies and they sort of boss like moments in sports games where you have an ongoing dynasty mode or in any squad game where the player is managing and investing in a rotating roster of player characters there's this interesting effect where as a particular character gets better and better and becomes more of an asset to the larger goal the players own sense of investment causes them to second-guess their use right so it's that problem of you know you should become overly protective of the one guy that can make a difference in the battle you know you don't want them to get killed you don't want them to end up on the injured list or dead it's a very strange very strange phenomenon with that I think it's also

interesting we have games which are very transparent about the fact that when the player fails it's kicking off a completely new mode of play right it's basically you know if you're playing Max Payne 3 or you're playing Borderlands for example and you get killed under certain circumstances the game simply stops everything and presents you with a mini game of trying to track down the person responsible for your death and and take them out in order to kind of recover the right to continue with the game and game like SSX or Forza there's a rewind feature which you know becomes increasingly valuable to further into the end of the session you are and potentially allows you to recover that lost about having to completely restart everything and then still other games establish an entire resource practically around failure alright the idea that as you're grinding your way through things you experience this kind of specific loss loss of humanity and Dark Souls for example and of course we're seeing games like rogue legacy which kind of bring combinations of these strategies together right this sort of recoverable resources linked

with the the kind of upgrade loss in the form of your heirs who inherit some combination of these things in a modified way is kind of the very essence of that game but these are all strategies ultimately for keeping the player from running away with a game right keeping them from colliding with the limits of the simulation you know but hard failure conditions no matter how you slice it or kind of inelegant I mean there's specific circumstances which they just flat-out don't feel that great like for example in multiplayer games the session must go on it's not going to stop just because one person dies so guess what you're gonna have to deal with that so that's why we would like to look at how to bring that player mitigation back into the fold of the core game experience we want to expose the player to survivable changes in their immediate situation from which they can then find their way back to something stronger or something more sustainable okay so think about why a

player spins out in the middle of a race it's because they're doing the thing they're supposed to be doing they're driving fast they're driving as fast as they possibly can they're driving too fast which is really what they should be doing and and what they're doing is they're going so fast that they're exceeding the performance envelope of the vehicle that they're driving or their own skills you know and so depending on some other related driving skills that might be fine that might be the opportunity for them to kind of rest control back of the vehicle and get things back on track with a minimal loss of momentum sure some players depending on what their motivations are trying to beat their best time they're gonna they're gonna stop and reload and go back to the beginning of that but for a lot for a certain core group of players the important thing is that they've maintained the continuity of that experience this this is the function of a setback so when we talk about setbacks just to be clear you know these are arising out of the game's low-level mechanics it's not something we design

into the map right we're talking about something that is not a forced mode change it's not an exotic gameplay sequence it's not a scripted moment where Nathan Drake is tossed out of the back of a cargo plane in midair even though that obviously fits the narrative definition of a setback that's not what we mean we're talking about a prolonged period of gameplay and it arises from the same systems so the players using the rest of the time it's preventing runaway behavior but it's still allowing play to continue survivable and it's perceptibly altering the state of the game in some meaningful way it's interesting so this experience if we want to think of it aesthetically is directly analogous actually to going through a reversal in the kind of traditional narrative sense it yields that water-cooler anecdote moment that's completely the player zone so there I

was but dynamically what it means is that the system defined by the player and the game interacting at runtime is able to absorb serious shocks and recover without interruption it's resilient in the way that economists typically use the term so what we want to do is we want to replace the failure cliff with something a little softer and a little more reversible all right something that falls a bit more of the kind of stimulus response curve and that kind of behavior emerges from having a feedback system that prevents the player from progressing unimpeded into meaningless degenerate end state while at the same time plucking them out of that kicking them back into a more volatile game condition in which success is by no means ensured but without driving them to the point where they rage quit now this isn't my idea this isn't new it was established by a lot of different people including notably Clint Hawking in his

2009 talk on improvisational play which was entitled appropriately enough fault tolerance and in describing the players experience on far cry 2 for example which was you know this kind of friction generated by systems like malaria Max is laughing at me weapons jamming you know fire probably getting fire all these various other functions you know the player is getting hit with these kind of small inputs and they rarely die from these events but it's unpredictable right it forces them to adapt I mean it's exactly because they are small and unpredictable that the player doesn't feel this kind of compulsion to to quit and reload so just to quickly illustrate the difference between what a failure is and a setback is if you're flying in a helicopter and your main rotor goes out that's a fail if you're flying in a helicopter and your tail rotor goes out that's a setback so basically you now need to look for a place to land as soon as possible you need to modify the way you're using the helicopter to sort of rely on autorotation and you need to somehow get the thing down without loot

loss of craft or crew ok well that's cute but we need a more formal definition than that so a setback needs to disrupt the players focus on their current low-level activity it also needs to revise the players awareness of the games state or context in other words a setback can be defined as that moment when the players immediate goal changes unexpectedly that says that setbacks are built out of the switching off between what the player thought they were going to be doing and a new priority that's absolutely critical for the continuation of play so if we can work with that definition right away we can start to understand the scope of potential setbacks in terms of a broader system of player goals so luckily there's a lot of work out there that establishes general taxonomy of player goals and the good news is they're mostly pretty finite

lots of people are already familiar with York and helping in patterns in game design so I've kind of adapted the list that they use there's 13 goals that they talked about yeah return them a little bit to put them in verb form but out of these 13 what's interesting is if you just start with the transitions that go from a single goal a to a single goal B and from that same goal B back to goal a because for our purposes those are distinct out of those 13 goals you're gonna get about 650 unique transitions a little more a little less I don't know that's not too many and out of those we only really care about the subset that's systemic recoverable and interesting but also the ones that keep the player in the sweet spot between wanting to quit the game and trade it in or being so inconsequential is to get lost in the noise of a moment-to-moment challenge all right so that's an interesting model but it's still too abstract I want to try applying it to the types of games I'm the most familiar with so I want to

look at the stealth genre I want to look at failure and stealth games which typically looks like this so classically stealth games have relied on kind of four key pillars you got the fragility of the game state the fact that you know the game starts the player out undetected by the game's ecosystems and the players challenge is to try to preserve that state moving forward there's the fact that that ecosystem is a simulation right it exists independently of the player it's analog for the most part so the player needs to use their intuition their observational skills in order to read the AI in order to read the environment and understand the gameplay potential there and the player is really incentivized to do that by the fact that they're not very powerful typically in the kind of classical model the game imposes often survival-horror levels of scarcity in terms of resources in terms of ammunition and health and and all of the other resources at the player's disposal and if they do get into a fight they're

better off running away typically now the flip side to that is that if these games are richly systemic as they often are they're typically pretty high agency right players are afforded a lot of tools and opportunities for playing with the AI and finding creative solutions to get through a problem now the classical hardship of stealth though was always that almost anything the player does disturbs the system right and once the player falls off that that's called the detection cliff it's hard to support many ways for them to continue moving forward it forces the player ultimately to be a perfectionist sometimes literally by so returning the game a game over of an alarmist trip or at best forcing the player to make a run for it so the early solution of that has been to go through some kind of strange contortions to limit the players exposure that leads to some sort of arbitrary level design solutions like

resetting the AI after each section and that doesn't really encourage experimentation so that model of stealth maps to these kinds of Gold's transitions so this is this is classical stealth in terms of gold transitions the way I was describing so from this initial state the player either needs to explore the world in order to reveal the various objectives available to them or maybe because it's on some kind of formal mission structure they just need to get from A to B they just need to get into the mission space and then from there they need to align themselves to be able to go through a collection of objectives some which are probably mandatory some of which may be maybe optional and the main disruption occurs when they come in the first contact with the enemy and have to get past them undetected and you do paint some kind of picture of the state of the game and understand where the enemy is at and what they're doing which leads to usually some fairly trivial goal shifting either to realign or to maybe get to some kind of safe haven and this

is notable because for much of play in this kind of game doing nothing is actually a really good strategy and the player has these kind of extended periods of stimulated planning not exactly action-packed and when there is a major setback frankly it almost collapses the game entirely and assuming that it's not a straight-up game over the players probably booking it okay so what we'd like to do is let the player recover from those kinds of situations and restore the game to a condition where they're hidden or they're back to being stealthy and they're doing what they like to do and we want them to be able to do that systemically so jamming open the detection window by preventing a single instance of the AI from wrecking everything and from deep propagating the detection to the rest of the world we want to assume that if you're some kind of elite spy or thief or assassin or hitman that you have it in your ability to take out that one stupid guy that's gonna

sound the alarm and kind of climb your way back up the detection cliff in order to get into a state of stealth all right so for the rest of the talk what we're gonna do is we're going to go through and look at kinds of gameplay models that are doing what I'm talking about we're gonna look at recovery dynamics and there are a lot of different ways that players try to modulate their ability that sort of the games try to modulate the player's ability to recover and when we look at them in terms of gold transitions it starts to become clear which strategies do a good job of exposing the player to good setbacks and which ones don't so the save anywhere model is going to be familiar to anybody who's been playing a lot of Skyrim on PC all right you've been riding the quicksave quickload buttons you know this can be satisfying I mean it's in its own way but make no mistake it degenerates the game right in terms of our model it effectively reduces any situation in the game to incremental experimentation and

that is a valid way to play that game but it does quickly hollow out whatever said some context the player has over time and it renders it mostly inconsequential in the context of the game unfortunately the the console friendly flipside of that doesn't fare a whole lot better right in this case the player and in corridor shooter simply has to endure the designers intentions and reach the next trigger volume in fact they might not actually have to do anything else they might not have to fire a shot again that act of traversal to get from point A to point B effectively collapses the game into a challenge of that totally undermines in mechanical tension so another solution is for games to deliver what's really kind of a thousand foot view on the game world and a company by kind of a rich

information layer if you compare clays mark of the ninja with for example Dishonored two games that share several core mechanics the side-scrolling format means players can see exactly what part of the world they can disturb and they have a lot of UI tools for monitoring the state of the AI even when they're obscured when they're hidden behind a wall or a door and when something does happen that disrupts the player's intentions it slips the game back into a goal of alignment and so this has a kind of interesting effect it's sort of having such a complete tactical picture of the world around you effectively turns the game into kind of a strategy game even though it's you know it's a 2d side-scrolling platform stealth experience and you know mark of the ninja is a hugely satisfying game it just it happens to rely on a different mechanism in order to deliver this that sense of setback it's not coming from necessarily a lack of information the way it would in a more immersive style of stealth game and Plus which because

of the format that it uses there is a more rapid turnaround when you fail and need to reload so it isn't it isn't punishing in that respect you know we talked earlier about how we put a premium on the consistency of the player experience but some games will change the rules on the fly as a hedge against players selling short when the chips are down and so they basically stated that that consistency is not as important as trying to keep the player in the game and that's a valid choice you can sort of see the carrot version of this with fighting games that feature damage scaling or some kind of desperation mode and ultimately you know what that what happens is the player learns over time that there that the trick is to absorb a punishment until the game delivers them the means of recovery and the effect of that is again to reduce the sense of consequence to the game now the stick version of that can be seen in games that use some kind of token based damage system so the players that are in an

otherwise skill based game can end up suffering certain kinds of setbacks deterministically all right so to use the stealth example AI enemies in the Batman Arkham games who will start targeting destructible gargoyles and challenge rooms is a way of interfering with the players ability to maneuver and they use the kind of Batman swinging around well the mechanism is the same that I should the players perception perception of is the polar opposite because in their view the game has taken something away from them and now it takes them time to get it back so in that case the perception is that it's a it is that it's pushing against them it's it's frustrating or foiling their attempts to do the thing that they want to do so those kinds of tweaks need to be applied conservatively so I want to talk for a second about the example of Dungeons of the endless this is interesting right we have the sort of roguelike experience where the players need to manage limited resources in order to get to the exit if you will of the current section that they're in that's pretty straightforward what spices it up is the fact that then we have these additional rooms that are

kind of these mystery boxes and that and the risk reward trade-off is by opening some of them up you can potentially add to your resource pool but you can also unleash additional Horrors on top of the ones that are already standing in your path right and so this generates a fair amount of tension in a very simple way now what's interesting is the simplest kind of tension economy is really just in a sports game it's when you have two teams that are both scoring points at the same time so this type of approach applies to a number of different games it's very interesting to see that it crosses a lot of genre boundaries that kind of simple Red Queen scenario with a kind of moving goal posts is forces those are pressures the player to switch goals when the point deficit between them and the other side gets too big right so that would be what we'd call like a single axis tension economy the points can go up the points bread can go down but whether I have four points or twelve points matters only in terms of how many points the other side has right and so that sets the context for the

game what that means in terms of goal transitions is that it forces the player to abandon the kind of luxury objectives you know the collection of objectives you kind of see on the far left over there in favor of the most expedient goals such as taking the opposing force out of play or going all-in on the main prize now you get a more nuanced two output in tension economies when you add a second economy that just consists of time that's slowly chipping away all right so you can take a game like rim capsule which is real-time the player has to watch the clock they're frantically completing all of these critical tasks you know building their bases scanning their monoliths all the stuff that they got to get done within 45 minutes and and then just waiting for that exact moment to like pull back into defensive positions because the next big wave increasingly large wave of enemies is gonna come in and try to mitigate the effects of that it works in a turn-based game as well so you could take invisible ink the game formerly known as incognita and every time the players little team

of spies makes a move it's depleting one little tick from this kind of finite pool of time that's available to them to complete the mission and so when it hits zero the whole map goes into alert you know and suddenly the reinforcements show up hot on your on your heels and and so this puts an interesting counter pressure on the player where they have to decide how greedy they want to be right they have to decide how many of these juicy optional objectives that are available to them they want to try to go after and and of course they're willing to overextend themselves a little bit each time when they know that the SWAT guys are coming in through that through the main doors and I get to chase them down to the exit and so with the addition to that second economy it kind of brings the dynamics of the game more fully in into view and it's what you're seeing is that it adds this layer of suspense to the players decision-making because it's no longer feasible for the player to succeed just through simple attrition all right Wilson that's interesting that's interesting so let's look at games that offer a kind of level playing field and symmetry between opposing

sides this applies to everything from head-to-head fighting games to sports games to some MOBAs and what we can say is that table-turning as we're going to call it here becomes enabled when the sort of mid-level get a minute-to-minute play between the two sides consists of them being engaged in a symmetric back-and-forth contest of control over territory and then that's driven or enabled by this low level or second a second game play it's basically kind of like rock-paper-scissors contests between individual units but it's perturbed or spiked by short-term power-ups all right so that's what's kind of driving this and so and what comes out of that is that the players immediate focus both the player and also his opposition was kind of a mirror image of him is shifting back and forth between goals and counter goals you know goals and the goals that are designed to block those goals and and the opposition is doing the same thing and this is relevant and talking about stealth games because it powers this kind of ongoing role reversal that you get between the hunter

and the hunted particularly in games where the AI is operating with a fairly strong enough perception model and you know the same conditions that allow the player to break line of sight with the AI and stay in remain hidden are also depriving the player of a full picture of the game state of the game world and it's making it risky for them to move out of hiding because the wrong move could bring them into immediate close-quarters contact with an enemy and alert the entire space and cause the situation to flip it so you get interesting setbacks emerging when the game dynamically plays with the amount of space that's actually available to the player on the fly so in GTA this happens because the verbs available to the player when they're driving anyway are all really designed to increase the localized amount of chaos in the world the amount of destruction that they're putting into the world and that becomes a magnet for the sort of aggressive law

enforcement response and we all know what how that ends up right that you get this cop cop card dogpile that results in clogged roads and the roads were already hard enough to drive through it triggers this immediate goal transition with a player you know has to switch from paying attention to all of the various mission objectives they may have had active at that point too realigning themselves around just you know stealth and hiding the car and eliminating enemies and and and trying to overcome the kind of amassing numbers of vehicles that are getting in their way and it's forcing all these snap decisions and maneuvers until things are getting progressively more claustrophobic and klog and klog and klog at which point the player of their own volition entirely elects usually to make a mode switch they abandon their car and they take off on foot just like in real life by leaning heavily on the destructibility of cover battlefield 4 for example can set up conditions where these were the players who were taking the smartest tactical decisions right

they're moving into buildings that afford them a good line-of-sight on their enemies and great creating excellent fire positions for them and so doing are of course becoming huge magnets then for return fire which necessarily creates this disproportionate effect when the entire building comes crashing down so so through that small action that seemingly intelligent move from a tactical point of view that player has set off a chain of events that results in this kind of massive re reformatting or restructuring the battlefield space we look at stealth games we can see that this approach opposes interesting setbacks just by flipping the dynamic lighting state of the game so other words you leave a dead body lying around the AI gets disturbed the next thing you know the lights come on so what's interesting about that is that area denial does not have to rely on literally physically blocking the player it could just simply be taking an area that was dark and now making it light and now it's not opportune for the player to move through that space anymore so with civilization v in the

transition to sieve there were four key rule changes that helped to power some very specific kinds of setback situations one unit production became limited became curtailed to the point where individual units became rather precious the second was that it became possible for an individual unit to lose a skirmish to kind of suffer a defeat without actually being eliminated you know and need to kind of retreat and rest up and recover and in order to remain in play there were also there was also the introduction of the new hex grid which created increased the number of adjacent playable spaces in the map and also the restriction of a single unit per hex so what those four rule changes ultimately did is combined to put a powerful increase in emphasis on laying siege as the primary mechanism of eliminating enemies on the map now as in

real life laying siege is dependent on all of the elements of the siege holding the line and that is jeopardized when a single unit is damaged and the player finds themselves tempted to withdraw that unit because it's so precious to a safe distance in order to be able to recover so the player needs to decide they need to go through this gold transition choice as to whether they are going to focus on a collection of goals that are guaranteed to try to increase the survivability of the individual unit or a collection of goals that are focused on the success of the larger strategic action very much consistent with the fantasy of of playing strategically and it's worth noting that those kinds of sacrifices at a granular level or can also matched in some games by supporting the player having to take much larger sacrifices in other words being willing to lose large swaths of territory ultimately in order to to win the larger war now I mentioned before in Clint's

fault tolerance to talk how exposure to potential setbacks is a characteristic of improvisational games you know the player is kind of constantly moving back and forth between cooking up plans for their next action and then executing those plans and the game features features lots of small sources of friction that combine in order to keep the player out of that execution phase and quickly back into planning and response you know in Far Cry 2 provides an interesting case study on how that unfolds in a stealth game but it was also an explicit part of the design challenges on Splinter Cell Conviction the creative director that game max bland who's actually sitting in the front row if you guys want to wave it him he one of his stated goals was to equip the player with multiple ways out of situations and that led to the establishing of kind of these three gameplay axes with the intention that players can be impeded on one or maybe

even two of these axes and still have it at least one Avenue left for pulling off a clever escape so there's axes are 3d navigation infiltration and combat and so when we look at at the outcome of adding that focus on having ways out and offering multiple axes of both input and output for the player what that does is it keeps the gate the player into gold transitions more frequently and some of those gold transitions exist on top of what we'll probably would consider to be the classical stealth loot so that extended loop is when the player gets into trouble with the AI right the new loop was termed prepare execute and vanish and you can see that at each of these stages prepare execute and vanish we're talking about a specific a set of gold transitions that allow the player to first move from an act of reconnaissance keeping eyes on where the targets are where the enemies are being able to then take action to deal with them either in an aggressive way or in a

more stealth oriented way and then ultimately regaining the state of being the being stealthy or hidden and the team introduced new mechanics specifically aligned with the pev game loop so they let the player mark enemies in order to keep tabs on them as they moved around the environment even if they were obscured it gave them the ability to to eliminate those marks if they earned that token through taking stealth takedowns and it also let them see their last known position giving them insight into the AI is thinking and planning and creating opportunities for them to lay ambushes or set traps it's important to note and I'll talk about this a bit later on as an important feature of setback generating that those mechanics were not simply added wholesale but they were kind of nuanced expansions of existing features of the game that retained that kind of deep rich interconnectedness with the other mechanics of the game so that it wasn't gonna undermine the players ability to use all the various tools of stealth but was designed early to be added on to that now a general lesson from

improvisation is that games were setbacks emerge along multiple axes can also provide players with an opportunity to adapt or adjust their approach along the same axes and we talked about ways to challenge or impede the player along specific axes and that gets into the idea of having orthogonalize din greedy n't sin the game and in this case we want to talk specifically about orthogonalize den amis so like left for dead for example uses kind of specific versions of the infected that are designed to exploit the kind of routine moments of vulnerability bubble up for individual players often creating moments that can draw of jeopardy that can then drive cooperative support very much part of the the kind of co-opt dynamic of that game but also left 4 Dead's AI director creates the right level of unpredictability to the infected hordes and the appearance of specialized infected you know in gears of wars horde mode you can kind of imagine the ideal playthrough for this it's like the players spend just the right amount of time setting up

fortifications they spend just the right amount of time collect running around collecting ammo and picking up guns they think are gonna be good and they get back into their defensive positions adjust the right amount of time sort of plan their strategy and just before the next wave of enemies attack and the reality is is that the appearance surprise appearance of kind of heavily orthogonalize enemies into that mix typically has the effect of tipping that applecart over right you know with suddenly oh okay our plan didn't account for that particular enemies showing up and Burling is way up through our ranks and so now we have to deal with that and then the aftermath of that creates a bunch of knock-on effects the players will then spend the next several waves trying to recover from so to go back to our stealth example you know on the follow-up to conviction Splinter Cell blacklist the team continued to build on gameplay axes that were established on conviction and the design lead Laura malval hit on this idea of using specialized archetypes in order to inhibit the player's freedom of action on purchase specific combinations of

axes and it's it's interesting it's uniquely challenging to use this kind of solution on a realistic game like Splinter Cell because unlike gears or left4dead the differentiation can't be exaggerated to the point of fantasy you know the variations on the gameplay abilities of the AI and even even their appearance and their their silhouette has to remain credible within a real-world context so so the specific archetypes that they came up with were you know the armored centric heavy that would challenge the player on navigation in combat a couple of different units actually that would inhibit the player an infiltration in combat one of which was a dog which had some limitations because he was a dog and it couldn't really communicate tactically with the humans and had this kind of imprecise smell mechanic for finding the player but was coupled with a corresponding human enemy the commando that had a little more on the ball in terms of being able to see in the dark and move quickly a drone operator that would interfere actively with the players technology and with their goggles and their ability to move around in the dark as well as

delivering autonomous drones that could challenge the player along 3d navigational paths and then ultimately kind of the super archetype the sniper which only appears about halfway through the game that can actually challenge the player on all three axes so with that in mind what's interesting then is to go back and revisit that extended progression of gold transitions that we associate with classical stealth plus prepare execute and vanish and you can kind of see where each of these archetypes and motivates the player to switch their focus so with a heavy you can see where the heavy is the heavies focus on navigation and combat is for single switches there between the kind of exploration and traversal stages and realignment and then later on between in within the execution phase of the the prepare execute vanish loop you can look at the dog and see where the dogs still blocking the player on combat is also in addition also blocking the player on the infiltration level so this is pushing the player to have to move into a recon phase trying to understand a bit better

they're the lay of the land their situational awareness to understand where the dogs coming from and what they're going to be able to do to get away from them and then ultimately the sniper hitting them at all three so you can see where the problem potentially lies with that particular architect all right so that's that's them on paper I mean you don't typically deal with them in a vacuum they're almost always presented in combination with kind of the generic grunt AI so it's interesting to look at what happens in the kind of more typical scenario represented representative scenario you know you the player enters the sort of kind of sniff radius the olfactory radius of the dog the the dog raises but it raises a bit of a fuss the human AI treat that is a bit of a disturbance same as like a broken light bulb or a door that shouldn't have been left open and eventually the dog began under begins undertaking a kind of random walk that will and deterministically lead him to the player if the player doesn't do something about it and so the kind of robust move in this case is to take advantage of the fact that the dog isn't really challenging the player on 3d navigation and use that as the

tool for staging and escape but sometimes it doesn't work out and the dog pulls the player up from cover and and the entire map gets alerted and they find themselves exposed now in the case of the sniper you have an archetype that can engage the player at about a hundred and fifty percent the distance to the other AI and is able to see in the dark using the laser on their eyeful and so that can create immediate problems for the player if they're relying solely on darkness as a way of staying hidden it can also create problems of the players relying on using their 3d navigation in order to try to stay away from the usual prying eyes of the AI because the sniper is usually in a position to see them and often if the player just simply gets involved in a normal run-of-the-mill altercation with one of the other AI then the sniper is in a position to put fire on them very effectively right away okay so that's you can start to see right away where that becomes a pretty substantial force multiplier so here's the really scary prospect dog time

sniper and so in this scenario the worst-case scenario is that the dog pulls the player out from cover and the player gets shot in that but just as likely to happen is that the player aware of the fact that there's a dog around and knowing that they need to move quickly in order to avoid that scenario simply runs around the corner and gets shot in that so there's an important lesson there and so I wanna I'm gonna try to wrap things up by talking about the insights that come out of that out of all of the things that we've just talked about the first is that when we use orthogonalization and these kind of defined game axes as a tool for both putting friction on the player but also providing the player with channels for surviving and recovering from the setback it's very important that those axes be agnostic to all of the various scenarios and specialization forces that we normally inflict on the player all right it doesn't do you any good to have three

axes if the player is only ever interested in using one of them at any given time or if the level design of the game blocks the player from using all but one of them in any given time that's not going to yield a setback that's just going to yield frustration so it was important that you know it's important than in a stealth game 3d navigation infiltration combat however you want to define those axes are of interest to a player regardless of whether they're a hyper traditional stealth player or somebody who's a little more action-oriented the other thing lesson which should have been fairly evident from the example that I used admittedly it's a data point of one but I think it's fairly compelling is that second-order combinations of setback ingredients can often generate frustration very very quickly and so they have to be used with a great deal of care I think one interesting observation that we made was that ultimately if you take a classical stealth model and then give the player tools for surviving altercations and a

continuing play and being able to extend that experience into other areas there is a general tendency for the game to push very very quickly into kind of arranged encounters with the AI and if that's not what you want if you don't want the game if you know the game to be telling the player great you're playing a shooter now then there needs to be some specific design around how to prevent that and I have some thoughts on it which I'll get to in a sec and then ultimately there's the danger of additional breadth all right we talked about the idea that you need to have mechanics that will help support these extended loops of play so that players feel like they have tools in order to help take advantage of the opportunities afforded to them by these setbacks and be able to feel clever and use those opportunities to come up with a more successful outcome but all of those features and you still need to be richly interconnected with all of the other mechanics the players using the rest of the time otherwise you end up with a shallow mess that the player somehow has

to learn and then interpret and decide how to use instead of a real opportunity to develop ongoing mastery in the game so it's important that those features be be applied judiciously and not and not in a giant mess of the new new gack okay I haven't had a visual for a while my favorite moment in Far Cry 3 was when I'd get harassed by cassowaries cassowary suck and actually it doesn't matter if it was a cassowary or a cow or you know actually getting attacked by a tiger isn't like the most exciting thing that happens to you in Far Cry 3 it's when you're in the middle of like lining up a target through your sniper scope you're about to take out an outpost and all of a sudden you get head-butted by a cow until you fall off a cliff cliffs again it's the theme I guess so but there's a very important lesson to be learned there right like for most of the rest of the game you're getting into long-range altercations with enemies that fall back behind cover take out a

gun and shoot at you I have never been shot at in my entire life knock on wood right I have however been chased by a dog menaced by a schoolyard bully had a drunk take a swing at me you know but like I know what it feels like to be to be menaced by somebody who's coming after me at close quarters with intent to do bodily harm right most of us know that feeling unfortunately and the reality is is that it's super visceral it's hardwired into our system we respond in a really in a really effective way when we're menaced in that fashion because it's part of our wiring and I think what's interesting and it's an interesting challenge and it's something that I have not seen executed especially well in games but I think it's it it's a clue that points the way forward particularly with stealth games if we want to try to support greater setbacks and richer setbacks is to try to encourage encounters that are like that try to encourage you know try to invest in AI and invest in enemies that want to move in for close-quarter

encounters with the player and menaces in that fashion you know but that means being willing to step away from the model that we use all the time when we're making kind of like modern-day sort of geopolitical military style games which is the guy pulls out a gun and shoots at you because you know everybody's got a gun so I think I think that's an interesting it raises an interesting question it's an interesting opportunity that I'd like to see more people take advantage of so as a quick recap I only have two points the first is you want setbacks in your game you need to figure out how to describe your game in terms of goal transitions it doesn't matter what taxonomy of goals you like or choose to use what flavor you want to go you just need to have a set and you need to be able to look at your game and kind of objectively evaluate what's going on in the game that way and the other part of it is once you've started doing that you're looking for specific setbacks that are gonna keep the player in the sweet spot they can't just be little bumps in the road that they're not going to even notice and they're also not going to make them want to rage quit and that's it that's how it works

we had a lot of time for questions if anybody wants to ask any yo hi very interesting talk and it's great a lot of it seems to really very strongly to strategy games which I work on as well which is very interesting to see the correlations I I think I think strategy games probably have been have been supporting and dealing with setbacks as a thing as a dynamic longer than anybody else has so it's it's it's not surprising I think there's still lessons to be learned from from looking at a lot of strategy games so one thing I'm curious about though is uh you started out describing sort of the danger or the the bulb the bad part about having binary failed conditions in that you're basically costing your players time which is an intrinsic to the game so

it's kind of a frustration point the setbacks you described are all super interesting but they seem to all just be steps along the way to a still binary fail condition so yeah yeah so I'm curious as to whether you have any insight into how you can get those setbacks to expand beyond that into maybe player motivations that exist outside of just straight-up failure so that you can set your player back without the setback just being a step towards fail and then reset right so I I think I it's a super good question and and to be honest I mean one of the things I'm acutely aware of and and I became aware of it as soon as I started looking at the fact that there was this spectrum between or a dynamic range between rage quit and I didn't even notice anything happened and the problem is is as soon as as soon as those are the things that are bracketing the output of the game the reality is is that you you are probably at the end of the day gonna have to fall back on some kind of binary failure state as in order

to have any kind of stakes now that's that's looking at it through the lens of someone who's been working on probably triple-a games that are like you know you shoot at people and they shoot back at you and you die etc so I acknowledge that you know probably a more 5 dimensional thinking about that would yield something that can stay completely within the game state and preserve the game state forever and allow you to continue you know odd infinite in without ever having to get popped out of it and I think there's lots of examples of games out there that do it I'm just not summoning them in the middle of this talk very well but yeah you're absolutely correct at the end of the day this is a it's it's a buffer it's a kind of a kind of a firewall against the the ultimate reality which is at some point the players gonna interrupt what they're doing and have to reload and lose some time we're just trying to make make that happen less often and less easily and not as early I guess awesome thanks thanks you mentioned a couple consequences for failure such as like you just go to a

loading screen come back and then all the enemies are the same or you kind of go it makes you do like a mini game or something that just reminded me of pray I don't know if you yeah then you know that game is actually quite an inspiration for me because I I firmly believe that going into a loading screen is very destructive to the players immersion yeah and so I was but I know some people complain about pray in particular that it it didn't provide enough of a consequence or failure that it didn't feel like they were failing and therefore wasn't satisfying what are your thoughts on that yeah I mean it's funny pray pray pray was actually one of the examples that I that I had and when I was originally working on this talk for for looking at the the kind of sort of Max Payne 3 Borderlands SSX 4 is a

style model where you're kind of introducing a mode change or kind of an alternate reality mode to the game that kicks in I mean pray literally does that I mean it says literally you're going into another universe as a result of failure and you get through that and come out the other side of it I mean I think it's interesting but I think I think maybe that also provides a little bit of a hint to why some people grappled with it it's like modifying the context of the game so radically that it changes the fundamental reality of the fundamental metaphor of what you're going through is is sometimes a risky thing to do you know and it's not like it's not like praise sort initial premise was was particularly accessible in the first place I mean if you were abducted by aliens in the first like five minutes of the game right so and you're in you're in occupying a spacecraft that has portals that let you jump through you know so it's there's a lot of strange things going on in the normal reality of that game and then kind of layering on this additional sort of spirit world on top of that I think was just a hard thing for some people to wrap their heads around but I don't think the concept is intrinsically

flawed I just think that there's there's ways of establishing context for it that probably would have made it easier for people to read and understand okay thanks anybody else thank you everybody thanks for coming