'Pillars of Eternity' and Proper Attribute Tuning

[Music] hello welcome to the most amazing talk at GDC um this talk is called gods and dumps I'll get to why it's called that in a bit it's about attribute tuning specifically this is very narrowly focused on tuning attributes in Pillars of Eternity Who Am I who are any of us really um I am a game director Josh Sawyer I'm at Obsidian Entertainment where I've been for the past 11 years I've been a tabletop role playing gamer for about 30 years I started with basic Dungeons and Dragons I've been a crpg developer for the past 17 years starting at black Isle Studios actually this month and I started out doing area and narrative design um but I primarily focus on system design and Direction these days

what are we even talking about we're talking about pillars of Eternity if you're not familiar with the game it was a Kickstarter funded game that is inspired by the infinity engine role playing games from the late '90s early 2000s uh that means balers gate icewind Dale and planescape torment and we had 77,000 amazing backers who helped us raise over $4 million and we released the game in 2015 and we just finished releasing the second part of our expansion the white March uh earlier this year we are specifically talking about attributes or what is commonly called ability scores and Dungeons and Dragons this is your strength decks Constitution all that sort of stuff I'm amazed this talk was approved because this is the easily the most narrowly focused talk I've ever given I'm going to be talking about things using the terms viable and optimal uh I'd like to define those just so it's clear what I'm talking about viable means you can actually get

through the game with it uh like your character will live uh and I don't really mean by the skin of your teeth I mean like you can actually do it it's not a nightmare of frustration optal means the sort of highend where a Min maxer will take a character by you know scrimping and scraping and finding every single bonus they can possibly get uh a big Focus that we had on pillars was to kind of try to reduce the gulf between viability and the optimal end uh it makes it a lot easier to balance uh but obviously we want men maxers to be able to go crazy because that's fun even though I hate Fun by the way uh what I mean by gods and dumps is it's kind of common when people are talking about stats in games to talk about things in terms of a god Stat or a dump stat a god stat is a stat that is extremely extremely good for the class or a build uh you always want to Max that out as much as possible a dump stat is a stat that you can just put it in the toilet like nobody cares it's not going to really affect your character how you play the game maybe a class doesn't rely

on that stat at all and that's something that we are watching out for throughout the course of development of pillars so I'm going to go back into the archives because uh pillars of attorney is a very Nostalgia driven game obviously we were inspired by the infinity engine games the infinity engine games were based on Advanced Dungeons and Dragons second edition except for ice ice windale 2 which was Third Edition D and so there's a big Legacy there that we had to pay attention to we didn't have to do exactly what those games did but if we strayed too far away then understandably our backers would be kind of annoyed at a minimum so what we're really talking about when we mean attributes in class-based role playing games is AD and D Dungeons and Dragons and advanced Dungeons and Dragons um a lot of games have their own variations on this but really it started with um pre-basic it started before basic uh there's a version of Dungeons and Dragons that's just after chain mail which was the sort of hardcore uh tactical tabletop war game and even that had stats back then

so they go back a very long way but lots of games have them now Ultima uh Final Fantasy games have them Dark Souls have them lots of games that have classes also have ability scores uh those can come into conflict which I'll be talking about but it's very widespread in role playing games for these things to exist they're used to define and distinguish the fundamental things about a character so how strong they are how tough they are how smart they are uh how well spoken they are these sort of very basic concepts of a character originally in the olden days they were always random actually what I found when I was doing some research is that in the first Dungeons and Dragons before basic the dungeon master actually rolled the stats for the players characters which is kind of crazy um but yeah there's a big emphasis on randomization uh but eventually over time it shifted to point by uh computer role playing games tend to have things as Point by because you know otherwise people just sit there and they roll and Roll and Roll and Roll which some people have fun with but it makes it kind of hard to balance the game that way and uh yeah so the the Trad

stats in Dungeons and Dragons are strength Constitution dexterity intelligence wisdom Charisma and can anyone remember the seventh attribute that was in thank you seventh attribute that was introduced in unar Arcana was comeliness and it went away immediately in second edition um yeah because how beautiful your character was is not necessarily the most important stat to have um in D andd so in the olden times meaning from basic through second edition you roll your dice and then you would figure out what class you could be so it wasn't and it's actually interesting if you go back and you look at the text in the old players handbooks it really does say roll your stats then think about what character you want to make whereas I think now uh especially in the language of players handbooks and things like that you sort of conceive a character and then maybe you roll or you allocate stats but you kind of do it in that order but in the olden days it really was like well maybe you can play a paladin I mean can you get a 17 Charisma because then you can play a paladin if not you're not playing a palad

Paladin uh there were lots of dump stats whole bunch of them so there were plenty of stats that classes could ignore it didn't really matter and there were big dead ranges um I'll talk about scaling of values but uh in a lot of cases D and AD and between 8 and 14 not a lot really changed um or if they did change they were really minor secondary things like carry weight and stuff like that oh you know what I just forgot to ask uh tell you guys we do have review forms I'm sorry I'm completely ruining the flow we do have uh review forms for this please fill out the evaluations at the end and if you have cell phones please silence them I was supposed to say at the beginning but I screwed up anyway uh so because of the fact that you were rolling these and because you had these big dead zones and you have these requirements for how you play the class uh viability was always in question you might not be able to play a character of a certain class or if you allocated stats they might be so bad that that character would either immediately die like in the first session or they would eventually just run out of steam and you couldn't practically speaking play them

anymore I'll give us a few examples of how this might work so here's one the reasonably intelligent wizard in ad andd make a wizard with a 14 intelligence it's a bad idea you might say like well 14 on a scale of 3 to 18 14 sounds pretty high according to the player's handbook I have 140 IQ I'm a genius I'm an amazing wizard wrong you suck because as soon as you advance to 13th level you're going to realize you can't cast seventh level spells or higher so you can keep advancing but you're not doing crap I mean you can find ways to boost your intelligence but if you if you built your character that way you're going to run into some trouble you thought you were going to be this dude you're this dude reasonably strong fighter in ad and D this is not quite as bad um so you start with a 15 strength by the way khed in balers gate had a 15 strength I'm just going to mention that because it's not a very good strength score um and then you take specialization cuz Fighter special and weapons and your damage is

really really bad so if you use a two-handed sword against medium or small creatures just in case anyone's wondering uh you'll do 1 d10 plus two because you have specialization which is 7.5 average damage versus 1 d10 plus 8 which is 13.5 average damage that's pretty bad and you might say well I'm going to get some more stuff later so I can work my way out of this hole so you get up to 13th level you take Grand Mastery you're using combat and tactics rules whatever so and then you get a plus4 weapon and your damage is still terrible and the thing is now it's even worse because this is being multiplied across multiple attacks so this golf is like you can play this way but this character is way worse than the character that sort of maxed out so you thought you were going to be this guy but you're this guy in Third Edition there was a shift uh the explicit class requirements went away so it no longer said hey you have to have a 17 charisma to be a paladin

there were lots of implicit things that you were supposed to do uh usually you could see them if you read the abilities and said like oh wow a lot of my abilities are based off of wisdom or Charisma I should probably focus on that uh bonus XP for your class disappeared so there was a thing in ad andd where if you had higher than a certain thing in your main stats then you'd get more XP which is giving a reward to someone that's already doing really well so they got rid of that because that's weird um also Third Edition got rid of the um the nonlinear progression so they went to every even point every even step you would get plus one so that changed the feeling of it it is much more consistent but that sort of massive ramp up at each end went away there were multiple builds that were vial you could build characters in a few different ways uh not necessarily optimal especially as you get to higher level in Third Edition that stuff gets really tricky some of the classes arguably required only three stats to be pretty good others like monks monks and paladins are very notable because they need four to be at Le least kind of good

or if you do not do that then some of your class abilities will be stinky poo poo uh and then D and D introduced Splat books uh if you're not familiar with the term splatbook it's an expansion book that usually introduces a bunch of crazy stuff um like cool new Prestige classes and insane Feats that your DM should say no to um but they introduce lots of neat gimmick build possibilities because you could do something like make a high intelligence fighter that was actually a good character whereas if you made a high Charisma or a high intelligence fighter in cor Third Edition that's a shitty fighter like that might be a really cool character to role play with that that character is very bad at doing their job which is fighting uh 3.5 or 3E and 3.5 is really A Min maxer's Delight um it clarified a lot of rules but it left a lot of room for optimization so you know it said well here are all the rules and but here are all these bonus types so if you can search through every single Source book

and combine bonuses in the right way you can just really go crazy um people made a lot of character builders for 3 and 3.5 and for Pathfinder uh those character Builders allow you to rapidly see like where all these stats can come from and people are insane on online forums um like they're crazy they'll they'll come up with really really crazy stuff uh as an example of this I like to use the example of Lawler skate um there is a forum something awful and in their traditional games Forum they had this uh ult ra highlevel character contest to see who could make the most insane character and the winner was an insane Australian person who um made this um you can't see the stats super clearly but this is ludicrous um he's like a yon anathema and he had like seven classes or six classes and 755 hit points and uh so yeah you could go you could go really really really crazy in Third Edition d and d and so people really like to min max but we also know that when the potential for Min maxing the optimal

character when that character gets too far away from the viable character it's really hard to balance content in any meaningful way because the gulf is so huge so in Pillars of Eternity when we set out to make our ability scores we had a few goals um these things were not necessarily things that we started out with but that we move toward them over time six attributes which is very is D and DS uh class abilities that are not explicitly associ with attributes so for example if you look at Third Edition Dungeons and Dragons a Paladin's Smite is in part based off of their Charisma score we didn't do things like that where classes have very specific abilities keyed to specific ability scores the statistics um the the attributes would influence statistics of importance to all classes and obviously there's variability in there as to how significant that is but our goal was to say try to influence things that all classes use rather than things that classes can safely some classes can SA ignore or they don't have any meaning to

them because then you get into situations where you can just dump stuff or certain classes really need stats more stats than other classes a primary goal was no bad builds uh what that really means in practice is that if you have an idea for a wacky character like the Super Genius Barbarian that is a viable character doesn't mean it's necessarily an optimal character but it is a viable character so we wanted people to be able to make the crazy characters like their charismatic fighter or their clumsy Rogue or whatever sort of weird thing they wanted to come up with and instead of that being a character that is just fundamentally you're not going to get anywhere with it it was viable not optimal but viable and we had a secondary goal which was to ensure that dumping stats stung a little bit uh this was kind of a reality check on our systems just to make sure that you know like if we thought a benefit from a stat was really important if people dumped it then really they didn't think the bonus that was being granted was that important so we needed to go back and look at it

again so why did we do these things why do we have six attributes tradition I mean that's that that's the main thing uh we asked our fans on our forums and they'd like H you know somewhere between 5 and seven I don't know maybe six I guess um but it was it was pretty common I mean some people genuinely did say like 05 or 07 but a lot of people said you know six they they wanted six it just it felt comfortable it it fit this type of game so that's what we did uh also there were some nice things though that pairs worked out well for our three secondary defenses which are fortitude reflex and will and defenses are not opt out um I saw someone who made a more elegant sort of way of saying this but defenses are things that you use when you're attacked and as long as the content actually attacks these different defenses with some regularity the player can't they can't not participate in that if that makes sense whereas certain things in D andd like if you like I don't don't care how long the duration of anything is so

I'm going to just dump this stat defenses as long as they're being attacked you don't want them to be bad because you will suffer for it so six attributes worked into three pairs and that worked out pretty well class ability is not associated with attributes the reason we did this is um if you make explicit ties between certain stats and fundamental class abilities then there's an emphasis there on viability so it's sort of saying if you don't Max the stat or if you don't have the stat at at least a pretty decent value this character sucks so for comparison if you think about the example of the the uh wizard with a 14 intelligence and d and d versus intellect and pillars of Eternity intellect affects durations and it affects area of effect for all characters and pillars of Eternity so you can make a wizard that has small low durations and small area of effect that's actually fine it's not going to prevent the character from learning spells or casting spells or anything like that so there's a big difference there similarly if you look at Charisma in Third Edition D and D it's extremely

important for paladins it affects their defenses it affects their Smite it affects all sorts of things resolve is not anywhere near the same for paladins it's a nice defensive stat it boosts their deflection increases their concentration so if they're on the front lines it's important but it's no it's not vital and it's not associated with any other class abilities directly and the sort of question that I ask is if something is required for viable Play Why is it an option like we want players to make meaningful choices and not put in gotas and traps that say like well yeah you can kind of put these stats wherever you want but not really like like you can dump this no you can't um so if something's required for viability that shouldn't be an option so rather we would say no none of these are required for viable play so it is really an option so influencing statistics of importance to all classes this took a long time and as I'll say later we uh we had to actually revise this during beta and post launch and into a couple of

patches uh so that process was fun but this is uh uh someone put this together on the pillars of Eternity game pedia site which is very nice um but it shows how things are affected so you can see that each pair of stats contributes to one of the secondary defenses uh two of the stats or actually one of the stats contributes to two defenses which is resolve uh those defenses are important to everyone uh even if you have a character for example with that's not on the front lines deflection is still attacked by arrows and guns and things like that so you don't want to tank any of those endurance and health are important to everyone obviously for Frontline characters they're a little more important but if you're playing a class like a wizard that already does not have much endurance or health uh dumping that will be not good and giving it a boost will be nice if you want to be a character that weads into the front lines damage and healing is important to everyone uh action speed is important to everyone uh interrupt and basically the ability to interrupt someone with an attack uh concentration which is the ability to resist that uh accuracy which

is used for resolving every attack you get the point so the idea is that this is stuff that all classes use because they all have abilities that rely on them broadly uh and that was to make sure that if you wanted to build a character of a certain type you could play towards those strengths so why do we not want to have bad builds well role playing is about building and expressing a character both through social interactions and through mechanics and viability limitations as I suggested before can result in mid or late game roadblocks you know as the challenge sort of ramps up you may find that how you built your character doesn't work and you know I've seen so many people playing a lot of role playing games that said like I got eight hours in and I realized I did this all wrong and now I have to start all over some people think that's fun I don't think most people do and if we have fewer viability limitations then we have more viable character Concepts so you can make the perceptive genius Barbarian which is actually a pretty darn good character in Pillars of Eternity because they affect

a very large area with their Carnage and they're very accurate you can make the idiot muscle wizard which is a favorite of mine so that's a character a wizard that has a low intellect but has a very high might so they don't have big areas of effect they don't have long durations but they do a huge amount of damage because might in Pillars of Eternity affects all damage not just weapon damage but spell damage all sorts of damage or you can make the clumsy determined Rogue who has a low dexterity and a high resolve they go into the front lines they have good defense good concentration they don't attack that quickly but they can still do a lot of damage and they have good defenses for Frontline attacks we wanted to avoid the cool concept bad character problem so I have been a dungeon master for most of the time that I've been playing d and d and I really hate the feeling when someone has a really cool idea for a character that is a sucky character because they have a dream of playing this wacky dude or lady and uh you know I have to kind of say like there's nothing like the game doesn't support that like you're not going to be able to make anything like that that character is going to die

or you're going to be frustrated playing them because they're going to fail all their checks it's not going to be good so how does this happen how it happens is that ability scores in classes can conflict um if you say like Class A is good at X Y and Z for example like causing damage Rogues are good at causing damage but ability B drastically impacts x and z then those are in conflict because it means that if you set ability B to some low value you're basically torp torpedoing the thing that your class is supposed to be good at the mechanics and the content balanced around uh one another run into problems when these things are sort of you know converging so in D and terms A Warrior's two hit Armor class zero or base attack bonus uh goes up and their strength is kind of supposed to be high and probably will get higher um and thus armor classes for Monsters increase assuming that those things are going up together if they're not then that character is going to fall behind and they're not going to be able to actually hit those characters as much as they should so the

example of the sort of the reasonably strong fighter I gave before the other thing is that strength also affects your uh chance to hit so that character not only is doing less damage per hit but is heading less frequently so they're even farther behind or further behind sorry uh another example is Rogues and reflex saves uh they're already high they have skills that are very good and based off of dexterity they assume that your dexterity is high so reflex saving throws and the difficulty classes of the skills that they use assume that those are both high if you're not Min maxing them then it's very difficult for the DM or in our case as people designing a computer role playing game that doesn't adapt to the player's stats uh it's very difficult to find a difficulty class that actually works so you get this thing the cool concept the smart fighter that is bad in practice the failure to M Max your key attributes means that your traits are low or your totals are lower overall and the content that scals assuming that the player is min maxing uh will just leave suboptimal characters behind and that's where you get that midgame nonviability

or late game mid viability both are very frustrating we wanted to ensure that dumping would sting so dumping a stat is again like if you you know it starts at 10 and you're like I don't need this and you take it down to three and then you put the point somewhere else so why did we want to ensure that dumping stings it wasn't to be punitive we weren't trying to say like we'll get you player but um it was supposed to be a reality check so again if we said hey uh we put this Stat or this you know thing on this attribute for example area of effect and dur so we think area of effect and duration are important and we think the value we give to it is important but if we see players continually dumping them then clearly it is not important to them that might be a perception problem which I should also say that a huge part of balance is actually just the player's perception whatever the sort of there's no objective sort of understanding of it it's all what they perceive of it so if they're dumping it you have to find a way to either communicate it better or you have to look at the values of what it's affecting if people don't care about dumping it it's not that important go back and choice and consequence is important to obsidian games uh we make

role playing games and in our role playing games we think that if you make choices there should be consequences for those good consequences and also bad consequences so that is another part of it is how you build your character there are trade-offs to how you do that here are some good things about how this all turned out I'll start with good things and then I'll go to the things people didn't like um it is hard to make non-viable characters uh unless you're playing on the path of the dam difficulty which is the high level of difficulty it's pretty hard to make a character that can't get through the game um and as long as you play toward the strengths of whatever the the build of the character is like again if you make a barbarian that high has a high uh intellect you want to emphasize your Carnage which is your big AOE melee attack um and other things that have an area of effect and duration based effects because that's what int affects uh it's easy to build a lot of unusual and viable character Concepts I've seen lots of people that propose a lot of the weird weird characters that if if you tried to make them in D and D would just be bad characters but they're able to

play them in Pillars of Eternity and they're just fine and they love playing them because they had that fantasy and that's part of why we allow people to do these things it is not hard to make different optimal builds that emphasize different things again I'm not going to say that everything can be made into an optimal build because then they wouldn't be optimal optimal sort of the cream of the crop so I've seen people make an interrupt Focus Barbarian so a barbarian that really focuses on just causing hit reactions on a lot of enemies at once they ramp up their in they ramp up their perception and they just smack those guys all day long uh a lot of people will make Gish muscle Wizards gishes are sort of the fighter wizard combos um they emphasize personal Buffs and summoning spell weapons super strong and powerful uh versus a crowd control wizard this is someone who focuses very heavily on big AOE stun locking enemies keeping them locked down lots of debuffs and minmaxers can still go wild like you can't make Lawler skate in in this um no one can make Lawler skate um by the way did I mention that that guy

lost a semester of his university making that character I personally admire that a hell of a lot because uh I was a terrible student um but yeah Min maxers can still go crazy on this stuff uh yeah it does not get as wild as as 3E or 3.5 but there's a lot of room for optimization in it and despite the relative Simplicity the mechanics still allow for a lot of interesting analysis um so for example accuracy and damage it's a common trade-off each point of perception that a character has in Pillars of Eternity adds one on 100 point scale to accuracy and that shifts your Miss Grays hit and crit ranges for all attacks now one point is not very much but it's shifting that whole all those attack windows and so people do perceive that as being valuable um it's not you know crazy powerful but it does make a difference each point of Might adds 3% to damage and healing from any source so again casting spells shooting a gun stabbing someone with a knife whatever

those are all affected by your might score if a character casts a healing spell cast lay on hands or you know cure stamina then that is affected by your might score so certain builds will rely more heavily on accuracy than damage output so for example the crowd control wizard that I mentioned they're not really focusing on doing damage but they really want to hit they want their accuracy to be as high as possible so with that you can say like well my might doesn't really need to be that high because I'm not really focusing on doing damage that's other characters are doing that I'm debuffing people and I would just want to make sure that they're more likely to hit or crit and so people do breakdowns and analyses of these and a lot of it comes down to personal preference what sort of character do you want to make and how do you want to make that or how do you want to play that character so that's always good is that we do want that room for interpretation in there um if it's really open and shut well this is superior all the time then we kind of screwed up so here are some common criticisms things people either didn't like or continue to not like about pillars of Eternity attributes they aren't realistic I'm not going to argue that they are uh they don't have enough of an

impact certainly compared to D andd I think that's a fair criticism perceived lack of value when again this goes back to the difference between like what the numerical value is and what people really understand of that value and the companions are awful oh my God the worst stats you've ever seen we'll we'll see about that so they aren't realistic well uh ad andd is not super realistic either which is kind of our starting point uh ad and D kind of has some interesting things there are ad and and d and d there are certain things that it seems like it is really trying to simulate like it is really trying to say like no this is like things work in the real world also you can jump off of a cloud castle fall and take 20 D6 damage and get up and dust yourself off and that's fine so it's not super concerned necessarily with ver similitude being really like reality um pillars already abstracts a lot there's a lot of stuff we already knew we were going to abstract so for us realism was

not a priority so you know for players who you know say that it is not realistic and and we go even we I would say we go even uh further in some cases so for example might affects gun damage it affects spell damage it affects you know melee weapon damage that's not realistic that's done so that all that stuff is on one stat rather than being spread out in a way that certain classes can kind of ignore certain stats or it doubles up for certain people which is what happens in D andd like if you get things that align on one stat then you're always supposed to Max that and it doesn't make any sense to you know focus on the thing that affects the damage types that you don't use so our Focus was to try to make good gameplay out of these stats and realism was not really a priority for us not enough impact so because we wanted these characters to be ensured to be viable that means that that limited their relative impact if you have something that goes up this high at the highest end and it drops down this low there's a big chance because that Gulf is so huge

that something about the viability of characters dropping into The Nether world so whereas if you keep it in a more constrained range it's much more likely to avoid viability problems if you dump that stat the character will suffer but they're not an unplayable character if you give a wizard in D and D a three Constitution they get one hit point per level and will die linear scaling means that we have a regular progression so if you look at the strength table that I'm going to show you in a bit you'll notice that there's a big dead zone and then it ramps up pretty quickly at the ends and so there is a much more significant sense of progression once you start really ramping up things like strength you like wow like the difference between these scores is really really huge but it kind of ignores The Dead Zone in the middle so when you have a linear scaling that does mean that it just kind of goes up at a very regular Pace Third Edition skips the uh the odd numbers so they actually get a sort it's a sort of skewed thing where every two points it goes up so it's a different feeling than having a little teeny increment every single

point and we were trying to avoid what we already were known and when I say known I mean by the designers of d and d uh imbalances in the scores so if you look at some of the ways that things were designed in Third Edition or 3.5 uh they understand that strength is an insane stat it affects two hit it affects damage or rather it affects base attack bonus um or being added to attack bonus and damage um and so for example half Works get twice as many penalties as they get bonuses because they get a plus two to strength because they knew that if you give a half orc a 20 strength and give them a great Axe they going to chop through the entire world um so we knew based on looking at how they designed those mechanics by reverse engineering how they tried to solve those problems that they knew that they were too powerful so we said like well you know we know that these things are crazy so let's you know let's scale them back so that they they're more balanced across the Spectrum so as an example you look at strength in D and D if you have a

two-handed sword with an 1800 strength which is sort of like you know good times top human strength you're doing 1D 10 plus 6 1 to 10 plus 6 is a lot better in terms of percentage than 14 to 20 plus 24% so yes uh much bigger impact in D andd so if you want your stats to have a huge impact we decided to ear on the side of viability not making characters uh really swing really wildly in terms of how their stats were built up so this table shows ad andd strength damage adjustment this is for comparison uh you can see that big dead zone in the Middle where it's just flat that means there's no uh damage bonus this is only looking at damage bonuses there are other things that strength effects but it's primarily damage assuming a two-handed sword because it's 1 to 10 that means I can do percentage modification and it's easy um if you use d8 then suddenly it's even bigger percentage modification um but you can see how it it kind of has some spots where it'll go to points um and flatten out and then it you know

keeps going up and then at 25 strength they just go ah get an extra you get an extra plus one for fun so there's not really a lot of Rhyme or Reason to it I should also note that uh before Third Edition strength had this weird percentile system so if you were a warrior and you had an 18 strength you'd get you'd roll percentile and that's how they incremented above 18 until you got to 19 that's why this chart is weird d andd is weird you got it now if you look at pillars of Eternity uh strength damage adjustment you can see that this is extremely linear um it's going up by 3% just so you can go back you can see 140 is the top end uh at 25 strength for us 30 is the top end and that's 60 strength so it is quite a bit less of an impact but again that is by Design um some people don't like that and that is okay but we're not changing it and then there's problems with perceived lack of value and again balance and perception of value is not just about what you set up but about what the player actually perceives about

it and lots of stuff goes into perception of strength I mean I talked to someone uh one of our system designers who had worked on a game where uh a gun was in a test and everyone's like God this gun sucks like it's just so horrible everything about it it's just it's junk I never want to use it and then they went through a tuning phase and they changed the audio effects on the gun and they did not change any of the stats and they put it you know they did another test and the players like oh man the sounds so good now it's you you guys got it and it was just purely it was purely audio so lots of things go into perception of these things uh so why did people have problems perceiving uh the impact of stuff or or or continue to have trouble perceiving them one is that percentages are hard if you ask people to do percentile modifiers like yeah this is 23% better like they're not immediately like yeah I got it um so you know they kind of round off but sometimes it gets difficult to do those versus oh 1 to 10 plus 6 that is a little easier to conceive um larger numerical values post

problems so for various reasons we used bigger numerical ranges I think you noticed before our uh our great swords do 14 to 20 points of damage so once you start saying like hey you have two or three digit numbers that you're supposed to do percentile operators on players kind of start going like I don't I kind of get that but I don't really get it and we didn't have great feedback in the UI initially thankfully Brian MacIntosh and C aruga um have improved this a great deal so that players can actually see comparisons like this is the bass this is the modified version these are all the things that are contributing to this value um transparency within the UI is extremely important um and that was a reason why though is that early on people were just like I don't I mean it says this damage value I don't know what the base is I don't know what's modifying it I don't know how much my strength is actually contributing to this and the companions are awful um so people complain about the stats of our companions a

bunch when we designed the companions we did them this way for a few reasons one is that the companions attributes are reflecting who they are as characters so go back to the khed example ked's not a super tough guy he's got a 15 strength that's okay man um our attributes are designed to be middle of the road so you know if you make your let's say you have a paladin and that Paladin is super Min maxed for damage uh our companion Paladin palagina is set up to be kind of like well you can use her for healing you can use her for damage you can use her for defensive stuff her stats are kind of more evenly balanced across what paladins can do and despite the fact that players you know plenty of players have played through on path of the dam the hardest difficulty using just the stock vanilla companions without modifying them and they're just fine a lot of players are still like no I want to change these stats so I'll give an example of how people kind of change these stats first example palagina she is our Paladin companion her character is described as

being very willful uh she's tough but she describes failing her physical examinations in the Brotherhood more than once so she's not especially strong she is an Aven Godlike meaning that she sort of has these birdlike features she has heightened senses and her eyes are frequently described in reaction text like she's very perceptive of things so the way that I sted this character because this is the character that I made um it's kind of hard to see there but she has a 12 M 13 Constitution you know nothing too crazy and then uh 14 perception because she's AV and Godlike and what is that a 13 in intellect and a 15 resolve because she's very very strong willed so this is from a mod called better companions she has a 21 might she's incredibly strong uh she has a 15 con okay that's fair she has a four dexterity extremely extremely clumsy a three perception which is about as bad as you could possibly get uh she's a genius she's a 20 intellect and an 18

resolve so clearly this person wanted to do like one thing or actually two things they wanted to cast lay on hands and they wanted to use their auras because paladins have nice oras if you want to hit something good luck because you have a three perception so you're tanking your main attack stat you can play you can play a paladin this way but this is like the way that you have to play this Paladin so our stance on this is we don't really care about people adjusting the companion stats uh maybe a little it does seem weird to see that character who's described in that way be so different from how she's described and uh but the attribute system is designed to handle it again like it's robust enough that it doesn't it's not that that character can't be played you just have to play them a certain way uh so sure like knock yourself out here's an example of how you could go crazy with another companion double of Carrick is a rogue that you get in the white March a lot of people like to complain about her stats and her armor she has a

relatively modest might score it's 13 uh might bonuses to damage are not multiplicative but they add up so every bonus to damage they just kind of add up as a percentage modifier so if you set her M to 18 you'd get an additional 15% damage but a rogue sneak attack which is their common like they just get it all the time they do it to anyone who's vulnerable to it is plus 50% damage so that alone is a huge bonus most Rogue active abilities add 25% damage and you get a bunch of those so just netting 15% more damage is not really I mean it's making a difference but it's not that big of a difference uh if you dumped her might down to three that would be actually wouldn't be 30% I think it did it's about 30% though I think it's actually 27 but you it's not that big of a deal because you're already doing plus 50 from your sneak attack and you're doing plus 25 on any active ability so Rogues are good at doing damage even if you T tank their might they're still pretty good at doing damage and you probably put those stats

somewhere else so they're getting a bonus somewhere else significantly lower damage but it's certainly not non-viable so I'd like to talk a little bit now about the changes that we made during the development cycle and after launch so we did have a pretty extensive beta which is great we got a lot of feedback a lot of feedback on attributes uh initially all attributes granted bonuses from one so if you looked at the scale before like D and D andd kind of is balanced around nine as being neutral and then things above that give a bonus things below it give a penalty so we said no no no or actually I should say I it's my responsibility I said yeah at one that gives you a bonus and it just keeps getting better well people didn't really dig that um it was different from D&D which in itself some people don't respond well to the other thing is that we have this concept of perceived loss like our perception of loss is way stronger than our perception of value so with that sense of we want people to feel like dumping a stat is not really a

great thing as soon as we set those like we basically just normalized all the math and then said oh yeah a 10 is zero and if you go down one you're at minus 3% damage and immediately people are like like like oh my God I can't put anything to nine and if you look at people's builds a lot of people are re like they'll they'll take everything down to 10 but that perception of Badness from going negative is extremely strong and powerful and I knew that at the beginning I should have probably just had them at 10 to begin with but well I didn't uh we also had perception and resolve both contributed to uh deflection so though we had two things feeding into that stat perception initially granted plus two to accuracy per Point everyone maxed this out as much as possible it was really dumb uh like everyone in QA made Max perception characters every little teeny bonus they could get to perception because it's shifting your Miss graze hit and crit chances it's like the whole thing up so that was

dumb Constitution's bonus to endurance and health was only 3% this is a minor thing that we adjusted later perception modified range that felt terrible um intellect modified radius not area those are different things um and it did it in 3% increments so throughout the course of the beta and into launch and I don't remember when each of these things happened like I said we were I think we continued patching attributes through patch 2.0 Maybe maybe even later than that we averaged attributes around 10 with lesser values inflicting penalties uh perception now contributes to accuracy at one point per and people seem to find that worthwhile uh and resolve contributes to deflection just by itself we increased the Constitution bonus to 5% and that actually did make a significant difference in how people perceived it 3% to 5% doesn't seem like it would be that big of a deal but people perceive it as being a bigger deal and so they actually kind of said okay I'm not going to dump this so much anymore and in intellect modifies area in 5%

increments so you can't make everybody happy um obviously not everyone was super thrilled with how this worked out um it's true of any design but we did not we did not set out to try to make everyone happy we knew that we had an audience with a variety of tastes our implementation was within uh what we believed our audience's tolerance threshold were so we know that some people that play our games um they don't they would rather have attributes at all there are role playing games um like Mass Effect that has classes but it doesn't have attributes at all you don't have to do that um it's easier when you're not making a fantasy game it's easier when you're not making a fantasy game based on D and D very clearly um so we know that some people just sort of ideologically or because of their preferences for system design they'd rather have no attributes at all that's fine we understand that and we know that a lot of people wish that there were stronger or different attributes and during the during our beta phase there were people that got real real upset about how we initially specked out uh

attributes how we changed attributes there are some people that sort of forswore the game they're like I'm over this uh so people have really strong feelings about this stuff um but we knew that there was a big group within our core audience that we could probably make something for them where most of those people would be happy so usually you know you still I still hear people criticize that which is sort of like we wish this were different but it's not a deal breaker and so that that applies to class design that applies to armor design weapon like all the stuff that we have in our game there's a lot of stff that people will say like well we wish this were not quite this way but we're okay with it as it is no data to support this this is just anecdotes we do have very active uh forums lots of people that give feedback a lot also there are lots of offsites um you know neoaf RPG codec something awful lots of very active communities there that have very robust discussions about this stuff and ultimately ideological Purity doesn't necessarily mean the players are going to be happy um you know I think it's important to have ideology for what

you're doing as a designer but at a certain point if what you're trying to accomplish sort of hits the players and the players go I don't like this this isn't fun um you know whatever your ideology is you kind of have to step back and say like am I approaching this wrong is this a bad goal is this not really accomplishing what the players needed to accomplish uh so start with start with an ideological point but then you know move toward the players uh in a sense of like you want them to have fun with it so yeah that should always be a goal would we do it this way again probably uh the benefits seem to outweigh the drawbacks uh I will say that I spent an obscene amount of time talking with people about how to set up these attributes how to tune these attributes uh going back and forth with programmers on it um because the math gets really complicated seemingly it wouldn't because we have a simpler system in many ways but it does get get very complicated uh listening to Forum feedback and all that stuff it really

did take a lot of time and effort to make something that actually accomplished all the goals that we set out to do and even there like you know we didn't necessarily do a perfect job on it obviously but uh the benefits of it do seem to outweigh the drawbacks uh we were happy with the gameplay so internally people seem to like overall how stats work um they like you know adjusting and getting little stat boosting bonuses and using spells and Buffs to raise those up that's fun gameplay and most of the players seem happy too like I said there are some players that really don't like it that is okay would I do this personally no I don't like that stuff like I don't I I think attributes conflict with classes they have a lot of problems I don't even like class-based systems so I've been playing RS Magicka for the past two and a half years that's a goofy goofy game so yeah this isn't really about like my personal preferences though like as a designer I'm looking at the audience I'm looking at their expectations obviously we kickstarted a game based on nostalgia and that you know that's based in a lot of memories and feelings and emotions that people

have um you know the point of Designing this stuff is not to make something exactly as I want it but to make something that I think will make a good game that sort of fulfills the expectations of the people that backed it and the people that want to come into this genre for the first time I do want to thank uh a bunch of people that helped out with this stuff I want to uh thank all of our backers obviously we wouldn't have been able to make this game if it were not for them our beta testers we had a subgroup of beta testers we still have beta testers that beta test our patches they give a lot of great feedback and players we have a lot of players we have a great YouTubers a lot of us at work you know we watch these YouTubers and we see their breakdowns of their analyses of the classes sometimes it's very interesting to see how they perceive things um and we understand like oh crap like they think it works like this uh what are we doing wrong here uh yeah but players give a a lot of really fantastic feedback our QA testers um I know that I think that especially in the in the

non-development world people don't really give QA enough credit uh I think that you know in development also I think there are a lot of places where QA is underappreciated I spent a lot of time talking with our testers and I still continue to spend a lot of time talking with our testers they play the game all the time they're trying to M Max and break everything uh they have really good insight into how the game is played um you know I don't always agree with their analyses of things but again that's the point that was like you want to listen to their View point and hear where they're coming from on that stuff but they were they were extremely valuable in how we built this stuff and lastly to Brenda Romero so I had given up on trying to pitch uh talks that were this narrowly focused because every time I did they were like this is really Niche like no one's going to no one cares about this stuff um so I had gone broader and broader and kind of stopped submitting things and Brenda Romero actually contacted me and said hey I want you to do um or consider doing a really narrowly focused Niche RPG talk so thanks to her I was able to do this we have some time for questions if

anyone has anything oh thanks hello hey uh thanks for great presentation thank you um in the old like Ballers skate games you usually had Dum stats and I remember specifically that you had uh if you you could have a low Charisma you can dump that down to three and like in the first tent outside of re dunion you can get a ring that set Charisma to 18 yeah uh so lots of people did that was like um goto strategy for many classes how did you work with items uh to yeah affect attributes and and balance so the question was uh sort of compared to Boulders gate where there were lots of items that influenced stats and in fact would set stats to certain values regardless of what starting uh value was how do we approach that in Pillars of Eternity so in Pillars we did

something that's more like Third Edition D and D in the sense that we don't have items that give uh that set stats to certain values um instead they give bonuses and there are stacking rules for those so you can minmax them a little bit but at a certain point you know it's only going to take the highest value from that part of that is also because um if you think about especially in the context of a a single player crpg if a person has a strat guide or if a person read a forum and they already know oh there's a girdle of Hill giant strength right here well then they can start dump the points that they would have put in strength they can put them in another stat go straight to that uh girdle of Hill giant strength and effectively they're playing with many more points in their attributes than other players are so that kind of stinks um like obviously it's cool to have game knowledge and like know knowledge the world and be able to be line for things and get that cool stuff but that Advantage I think should be a marginal advantage shouldn't be so gross like cuz the difference a girl of Hill giant strength is 19

strength if you set your strength to three you just got 16 free attribute points so yeah we tried to keep those bonuses a little more constrained um I know that again some people would like those to be more significant in the white March expansion we added plus four items which you oh my god um you know people really like when those step up and they can you know keep ramping them up so thank you thank you hello uh first of all thanks for the game it's great thank you and um I just had a question actually about the um early process of design and the prototyping process since you were so steep in tabletop design uh what it was like to test such a complex system if you went straight to digital or if you did paper prototyping or even a tabl toop session and uh how that sort of looked like and how you kind of got to where you are before the code happened um we did not do any so the question is if we how we prototype prior to uh implementation yep um and since we had a lot of basis in tabletop did we do

any tabletop gaming uh we did not uh part of that is because these stats would be a nightmare it's funny too because some people say like Hey when's the tabletop version of this coming I'm like you want to do percentile operations on damage ranges that are like double digit I don't think you actually do um so we didn't do that we we sort of charted things out in Excel to see like what are the boundaries of this how is this really mapping out with how we're planning out abilities and damage ranges and things like that uh so it was mostly sort of it was theoretical for a very long time but we also we looked outside of D and we looked to other games and said you know how do we feel about balance in in these games and what are the um the sort of margins of influence like how how impactful are these attributes on that and how does that feel uh for the type of game that we're trying to make and the nice thing was that um you know Tim Kane did a lot of the implementation of the attribute things and it was uh it was usually pretty easy if I went to him and I said H I was wrong like these values are bad we need to adjust this up or down or whatever it was usually

pretty quick turnaround so we could test out those changes relatively rapidly but um yeah I I would have done tabletop uh sort of testing of that stuff but the mechanics were really designed and we we wanted to design them for a crpg we wanted to say like we got a computer here you can do lots of calculations really easily for you we don't have to base things on dice um you know and so that's why our damage ranges are more things like you know 14 to 20 uh so yeah great thank you thank you hi um so obviously the tributes are pretty big exception but was there anything else that came up kind of organically while you were designing the game that you wanted to put in but you thought the um the Nostalgia Focus wouldn't allow were there things that we wanted to put in that we thought the Nostalgia Focus would allow um I not really there were certain things that we put in that people resisted very strongly because they weren't kind of part of the their vision of what the experience was so for example we have firearms in

Pillars of attorney but they're all they're all like 16th century era firearms and while Forgotten Realms did have guns as early as second edition they weren't part of the infinity engine games at all and so a lot of people thought that that was just like they were like that's no that doesn't work that that feels completely wrong um and there were certain things that early on we decided we should probably have these because we think a people a lot of people will miss them if they're not there so for example we have humans we have elves we have dwarves I think a lot of people like humans elves and dwarves uh we don't have halflings we don't have half Orcs so we're like ah we don't need to have those we can put in AA they're our big race and then we have orans as our small race so we kind of figured that people like that idea of like the sort of big race and small race we didn't think people were that attached to the idea of like I must play my halfling or gnome in this game so thank you thanks hi um I was wondering if you could elaborate on the different difficult ulty options in your game and kind of how that influenced your attribute design I find that you know

the higher difficulty goes usually that leans more towards Min maxing less towards viable characters and since uh some of your most dedicated players are going to be like interacting with those systems I feel like an urge to always look there and not look at like viable characters I was wondering what your experience with that was so the question is about our difficulty design and how that influenced our attribute design and so we have uh easy normal hard and then we have have a few other Crazy Ones the most significant of which is path of the Damned and we except for path of the Damned we did not do stap modifications on creatures instead we changed the composition of encounters so if you're on normal you might see you know like a wizard and two paladins but on hard it's like an arch Mage it's a completely different like more powerful wizard and two paladins and you know like a rogue gets added into the group and and it's in a weird position where it makes it more tactically difficult for you to deal with it but path of the Damned was specifically designed for the most

aggressive players and so we said like this mode not only does it have all the creatures from all levels of difficulty but they all have increased stats so they have better defenses and better accuracy and we're kind of we kind of said with path of the Damned like we know that there is a gulf between viable and optimal and this is where the players who want to push that envelope like they need to do that um if you just kind of stumble and like hey path the dam what's this and like you just kind of try to stumble and Bumble through you're probably going to die so um yeah we wanted to make sure that uh viability for characters didn't require crazy optimization even on hard but on path of the D there are some really good path of the damn players who don't even feel like you need to optimize that much on path of the Damned but they're really good players um most people have to get a little more ambitious with how they're looking at their stat spreads and things like that their gearing becomes much more important on path of the Damned so that's how it tied together thank you you're welcome uh yeah so talking about

dump stats um I get the feeling that a lot of RPG players kind of naturally assume that you know if you're playing a fighter you want to dump Charisma if you're playing you know a wizard you want to dump like dexterity or something like that how did you guys communicate to new players of your game that you don't necessarily want to you know just dump all your perception um yeah so that that is a tricky that's a tricky thing and we did have to deal with that a lot thanks for bringing that up um you know we did have a thing where you know people would look at somebody and they're like wizard strength I don't think so and then they're like casting their you know damage spells and like what the heck is going on because you know in D and D there usually stats don't affect your spell damage at all so they had no expectation of that so one is again like trying to make the UI reflect uh like you know Clearly say like this affects damage like all damage but the other thing we did is uh one of our programmers Robbie put in a thing where we could add Little Stars next to character attributes in certain classes to say like this is kind of important

like to sort of say like you might think this is not important look again um so I think Wizards had like a either a silver or gold star next to might to sort of say like this is more significant than you might think it is you can still dump might but if you dump might you're going to make a a crowd control character or something like that you're not going to be focusing on damage but yeah it it is a difficult problem because again we're dealing with a lot of expectations we also we called it might hopefully to make a player go like wait a minute isn't that normally strength is this something different it is something different friend it's something that affects damage and healing for everything so sometimes it's the nomenclature like Charisma we got rid of and replaced it with resolve it's sort of similar but it affects like deflection it affects concentration very different from how Charisma Works in mechanically in D and D although resolve is also used in conversations for you know being like a real intense kind of a person so it's difficult um but we tried to use naming uh and UI sort of

signaling to help guide people in that direction thank you thanks any last questions all right thank you everyone for coming [Applause]