The Logic Behind Attribute Design

[Batbro]

For those of you who don't follow Josh Sawyer on social media, he has graciously shared his slides from his GDC talk, "Gods and Dumps: Attribute Tuning in Pillars of Eternity."  They're fairly self-explanatory, and it's a fascinating read for those interested in systems design.

 

Read here.

 

I love a lot of the points he makes, especially the reasoning behind de-marrying attributes and class abilities, and the quote "If something is required for viable play, why is it an option?"  Definitely a lot of notes I'll be using when designing my own game.

[MaxQuest]

Thanks for posting this.

 

Many interesting points made there. And yeah, it's indeed harder to come up with a completely non-viable build (e.g. maxDex/minInt chanter, minMight/minInt priest)

 

Although I wouldn't compare Might from PoE with Strength from DnD. Because there is DR in this game. 

Also when you try to figure out the final dps, you do stumble upon wonderful damage coefficients and attack speed stacking formulas

Edited March 22, 2016 by MaxQuest

[Braven]

I really like how Pillars of Eternity handled stats.  It is a lot more interesting than DnD and it actually allows for a lot more build options and play styles.  

 

I have done low INT casters (who favor single target and non-duration spells) and HIGH INT casters (favor the opposite) and they both work.  Similar can be done with other classes.  You could have a character that focuses on buffs/debuffs with low might or ones that already have a lot of additive damage bonuses so they relatively benefit less.  Even with all stats being equal, you can still optimize by choosing talents and abilities that benefit the most from your stat distribution since different items and abilities benefit more from some stats than others.  

 

Low might?  Don't choose healing abilities.  (opposite for high might)

Low Int?  Favor abilities without durations or AOE.  (opposite for high int)

Low Resolve?  Favor items that benefit from being crit. (example: shod-in-faith), use ranged weapons or seek out concentration boosts. 

High Resolve?  Benefits more from deflection stacking (shield) and better suited for light armor since DR is needed less.

High Perception?  Favor items with on-crit or hit effects, or weapons with a superior interrupt rating.  Or abilties with great debuffs (increases duration)

Low Perception?  Use graze-to-hit or accuracy increasing items/abilities.  Favor abilities without an accuracy roll (like buffs) and avoid debuff abilities which suffer the most from grazes.

Low Con?  Favor endurance regeneration and lower incoming damage, particularly burst damage (attack from range, increase DR (heavy armor), increase deflection). 

High Con?  Best for characters likely to receive high burst damage (low deflection, low DR).

High Dex?  Increase DR-Bypass since you are hitting more often.  Best for those with many activated abilities to dole out (high level casters), frequently use actions that other stats don't benefit (example: summoning figurines), or use weapons with on-hit effects (spell-striking % procs (like soul bounds often have), crit-to-stun, high interrupt, etc).

Low Dex?  Best for characters that don't need to attack often to be effective and have few activated abilities (chanter, pure tanks, retaliation).  Can mitigate with zero attack recovery if stats needed elsewhere.

 

In conclusion, all stats are viable, but to be optimal, you should build your character around your chosen stats to minimize the drawbacks and maximize the benefits.

Edited March 22, 2016 by Braven

[jsaving]

The presentation nicely summarizes Josh's belief that a good rules system should prevent players from creating gimped characters.  Not everyone agrees it is a desirable goal, but I do think PoE largely achieved it -- as evidenced in part by the number of people who've completed PotD with the suboptimal in-game NPCs. 

 

But there is one part of the presentation I find slightly problematic.  Josh first says the game's attributes are abstract/unrealistic by design and reflect neither the way characters look/think nor the traditional D&D definitions -- might isn't strength, resolve isn't charisma, etc.  Yet elsewhere in the presentation we're told that the devs don't want NPC attributes to change because the current values reflect how those characters look/think, specifically calling out Pallegina's weak physical strength as a reason to give her a suboptimal might score.  This kind of thing is also seen in the many in-game scripted interactions that equate PoE's attributes with D&D's -- resolve is invariably used for charisma checks, perception is used for wisdom checks, might is used for strength checks, etc.  But if PoE's attributes really are thinly disguised synonyms for D&D's instead of being abstract, then we're once again faced with the questions the PoE team was trying to avoid by delinking them, like why charisma would physically deflect arrows from hitting you or why being muscular would power-up your fireballs. 

 

But these are nitpicks, really.  The PoE team did something really innovative with how they handled attributes and they should be applauded for it.  And if there are imperfections to be worked out, what better place to do that than PoE2?

[Gromnir]

The presentation nicely summarizes Josh's belief that a good rules system should prevent players from creating gimped characters.  Not everyone agrees it is a desirable goal, but I do think PoE largely achieved it -- as evidenced in part by the number of people who've completed PotD with the suboptimal in-game NPCs. 

 

But there is one part of the presentation I find slightly problematic.  Josh first says the game's attributes are abstract/unrealistic by design and reflect neither the way characters look/think nor the traditional D&D definitions -- might isn't strength, resolve isn't charisma, etc.  Yet elsewhere in the presentation we're told that the devs don't want NPC attributes to change because the current values reflect how those characters look/think, specifically calling out Pallegina's weak physical strength as a reason to give her a suboptimal might score.  This kind of thing is also seen in the many in-game scripted interactions that equate PoE's attributes with D&D's -- resolve is invariably used for charisma checks, perception is used for wisdom checks, might is used for strength checks, etc.  But if PoE's attributes really are thinly disguised synonyms for D&D's instead of being abstract, then we're once again faced with the questions the PoE team was trying to avoid by delinking them, like why charisma would physically deflect arrows from hitting you or why being muscular would power-up your fireballs. 

 

But these are nitpicks, really.  The PoE team did something really innovative with how they handled attributes and they should be applauded for it.  And if there are imperfections to be worked out, what better place to do that than PoE2?

speaking o' nitpicks, why is there a need to make analogous to d&d attributes?  poe resolve is not d&d charisma d&d, nor is it 'posed to be as far as we can tell.  poe resolve represents will and purpose, which is why durance has high resolve.  is arguable that durance has crummy charisma, but am understanding why the developers and writers o' poe would see durance as being resolute.  resolve boosting deflection is the abstract quality o' the poe attribute.  somehow in the poe universe, strength o' will adds to deflection, but resolve insofar as ordinary dictionary definitions is not an abstract concept. is not charisma.  is not 'posed to be charisma-by-another-name.

 

poe might?  is more abstract than even resolve and while it largely follows the mechanical role o' d&d strength, it has not a specific d&d analogue insofar far as representing the qualities o' a personages character.  IF might represents some kinda quality o' the poe character, it is a largely unseen and enigmatic descriptor o' o' the being's soul.  is not d&d strength that represents a character's physical brawn.

 

poe perception is not the same as d&d wisdom. the actual oddity o' perception is not how poe uses perception but rather how d&d has, at least since 3.0, used wisdom to represent both good judgement, but also as some kinda modifier o' situational awareness. johnny manziel could arguable have fantastic perception, but am doubting anybody would  think to call him wise, yes?  play football at a high level, particularly at the qb position, requires amazing awareness.  conscious (or unconscious) recognize dozens (hundreds) o' small details simultaneously while being chased by multiple men weighing 'tween 250lb-350lb, men o' who is intent on slamming you to the ground with force similar to a car crash, requires near preternatural perception.  be a qb at college or pro level, as has been shown many times, does not require any kinda wisdom.

 

etc.

 

discussion o' poe attribute strengths and fails will be more productive if the d&d baggage is abandoned at the door.

 

HA! Good Fun!

 

ps  one quibble we got with josh presentation and comments regarding the poe attribute system is the insistence on referencing the intelligent poe barbarian as some kinda representation o' the diversity possible with poe characters.  *snort*  one o' our main gripes regarding the poe classes (not attributes) is that the barbarian is integral linked to carnage.  we can make a low intellect poe mage who focuses only on single-target spells that have instant durations, but try and make a decent poe barbarian with crappy intellect.  one fail o' poe classes is that for the barbarian, intellect is not much o' an option.  carnage is a class defining ability o' the barbarian (which is 'posed antithetical to poe philosophy)  and intellect is necessary to fully realize carnage.  for josh to continue using the example o' the smarty poe barbarian to illustrate points 'bout freedom o' the poe attribute system strikes us as intellectual dishonest.

Edited March 22, 2016 by Gromnir

[Ymarsakar]

I saw how they were designing stats in the backer beta forums, sort of a project in motion view.

[Ymarsakar]

The presentation nicely summarizes Josh's belief that a good rules system should prevent players from creating gimped characters.  Not everyone agrees it is a desirable goal, but I do think PoE largely achieved it -- as evidenced in part by the number of people who've completed PotD with the suboptimal in-game NPCs. 

 

But there is one part of the presentation I find slightly problematic.  Josh first says the game's attributes are abstract/unrealistic by design and reflect neither the way characters look/think nor the traditional D&D definitions -- might isn't strength, resolve isn't charisma, etc.  Yet elsewhere in the presentation we're told that the devs don't want NPC attributes to change because the current values reflect how those characters look/think, specifically calling out Pallegina's weak physical strength as a reason to give her a suboptimal might score.  This kind of thing is also seen in the many in-game scripted interactions that equate PoE's attributes with D&D's -- resolve is invariably used for charisma checks, perception is used for wisdom checks, might is used for strength checks, etc.  But if PoE's attributes really are thinly disguised synonyms for D&D's instead of being abstract, then we're once again faced with the questions the PoE team was trying to avoid by delinking them, like why charisma would physically deflect arrows from hitting you or why being muscular would power-up your fireballs. 

 

But these are nitpicks, really.  The PoE team did something really innovative with how they handled attributes and they should be applauded for it.  And if there are imperfections to be worked out, what better place to do that than PoE2?

 

Their problem is that the writers are used to writing for D and D, and dungeon mastering for such games.

 

So the combat system wasn't polished in the backer beta, so the writing used archetypes that the writers were familiar with as the writing began way before the backer beta. Josh wasn't responsible for the writing and creative narrative. In the context of the historical lore Josh Sawyer created for Pillars, the stats make sense.

 

I remember they specifically did D and D games when writing for Pillars, as a way to prototype some ideas.

Edited March 23, 2016 by Ymarsakar

[Stasis_Sword]

I don't think the system is perfect, but it significantly improved the relevance of stats. Think of how many low int low wis barbarians littered the D&D landscape. In Pillars those barbarians wouldn't be able to hit anything and would have a criminally small carnage. I also like that you can tailor attributes based on your build priorities. Want to pump con and res to 18 to make a riposte rogue, go for it.

[Baron Pampa]

The presentation nicely summarizes Josh's belief that a good rules system should prevent players from creating gimped characters.  Not everyone agrees it is a desirable goal, but I do think PoE largely achieved it -- as evidenced in part by the number of people who've completed PotD with the suboptimal in-game NPCs. 

 

But there is one part of the presentation I find slightly problematic.  Josh first says the game's attributes are abstract/unrealistic by design and reflect neither the way characters look/think nor the traditional D&D definitions -- might isn't strength, resolve isn't charisma, etc.  Yet elsewhere in the presentation we're told that the devs don't want NPC attributes to change because the current values reflect how those characters look/think, specifically calling out Pallegina's weak physical strength as a reason to give her a suboptimal might score.  This kind of thing is also seen in the many in-game scripted interactions that equate PoE's attributes with D&D's -- resolve is invariably used for charisma checks, perception is used for wisdom checks, might is used for strength checks, etc.  But if PoE's attributes really are thinly disguised synonyms for D&D's instead of being abstract, then we're once again faced with the questions the PoE team was trying to avoid by delinking them, like why charisma would physically deflect arrows from hitting you or why being muscular would power-up your fireballs. 

 

But these are nitpicks, really.  The PoE team did something really innovative with how they handled attributes and they should be applauded for it.  And if there are imperfections to be worked out, what better place to do that than PoE2?

 

I think that the attributes are abstract in ways they contribute to combat, but usually have clear, down to earth meaning in non-combat interactions. So you can create any kind of character and he'll be viable in combat, while retaining the characteristics you wanted him to possess.

[Madscientist]

Very nice

I think that the combat system and character design is much better then the IE games or NWN1+2. (sorry, I know only the DnD rules for version 2 and 3 bacuse I have played those computer games).

 

DnD games in general:

- high level mages are immortal gods of destruction, who can only be stopped by another mage

- priests could buff themselves to be better fighters than anyone with the fighter class could ever be.

 

In the IE games:

- only extremly high or low values had an effect

- Some stats were completely useless for some classes.

- some classes were much more powerful than others, especially if you look at some dual or multi classes (Kensai -> mage for maximum mellee AND magic damage, Mage/cleric for spell triggers/contingencies with priest spells, ranger/cleric for all cleric and druid spells while dual wielding hammers and flails, . . .)

- You must have a thief in your party for traps, chests and scouting (unless you do lots of meta gaming)

- You could reroll until you had maxed out all stats

- The DnD 2 system is hard to understand. Your stats should be as high as possible but your AC, thaco and saves should be as low as possible. A sword+1 reduces your thaco by 1. Obscure rules for stacking magic items. I am still not sure if it is good or bad if a spell is done against a saving throw  vs. magic -2.

- After character creation, your only choice was weapon profiencies.

 

DnD3 was an improvement, but now you had the option between tons of classes, feats, skills and each one had different requirements.

In the beginning you are completely overwhelmed with options and it is easy to make a very useless char.

In NWN2:

- Some class combinations are extremely powerfull (kaze no kama, obsidian sorcerer, . . .)

- you can do lots of cheese (1 level of shadowdancer, one level of cleric for domain powers, . . .)

- tons of bugs and exploits (you can get more than 2 cleric domain powers, some prestige classes give progress to both arcane and divine spells, . . .)

 

PoE is an improvement in almost every way:

- all classes are useful and you have several viable build options for most classes

- all attacks are calculated accurancy + 1d100 - defense

- bigger numbers are always better (except damage you get ;-)

- each attribute has at least some use for most builds.

 

The only problem may be the barbarien, because he is all about carnage and so he needs int.

 

I really liked the IE games a lot, but after playing PoE I say:

good riddance DnD system in computer games.

[hilfazer]

HA! Good Fun!

 

ps  one quibble we got with josh presentation and comments regarding the poe attribute system is the insistence on referencing the intelligent poe barbarian as some kinda representation o' the diversity possible with poe characters.  *snort*  one o' our main gripes regarding the poe classes (not attributes) is that the barbarian is integral linked to carnage.  we can make a low intellect poe mage who focuses only on single-target spells that have instant durations, but try and make a decent poe barbarian with crappy intellect.  one fail o' poe classes is that for the barbarian, intellect is not much o' an option.  carnage is a class defining ability o' the barbarian (which is 'posed antithetical to poe philosophy)  and intellect is necessary to fully realize carnage.  for josh to continue using the example o' the smarty poe barbarian to illustrate points 'bout freedom o' the poe attribute system strikes us as intellectual dishonest.

 

 

True words of intellect wisdom.

 

 

 I am still not sure if it is good or bad if a spell is done against a saving throw  vs. magic -2.

 

 

Good. Unless you're the target (:

[jsaving]

 

True words of intellect wisdom perception.

Fixed that for you...

[Odd Hermit]

I don't think they really succeeded at making dumping sting or making all attributes unnecessary. Intellect for every caster(especially Priest and Wizard) feels close to mandatory because duration is such a dramatic benefit for all buff/debuffs/control spells and the way AoEs work strongly encourages going for larger ones just for more comfortable no-friendly fire edges. I would never make the "muscle wizard" they discuss because the buffs to make a melee wizard shine still need int to last long enough in combat.

 

There are also some no-brainer dump stats - fighter-ish classes have several builds that gain so little from Int you can drop it to 3 without a second thought.

 

Basically, you can still almost completely ruin some classes, they'd be "viable" if only because with meta-knowledge anything can use the right items and tactics(pulling, kiting, stealth, speed, figurines, traps, scrolls, etc.) to pull through regardless of how poorly built the character is.

 

The Might damage adjustment graph is also overly simplistic and doesn't reflect the actual benefit of might in practice. Different abilities and classes gain substantially more from might, and enemy DR also means higher might can lead to up to and beyond double your actual damage on some targets particularly in the early game. Which is why early game it's best to just wear plate armor on everybody regardless of recovery penalty(especially with the new enemy AI, I put my more fragile character in slightly more DR than my tankier ones).

 

It's not as bad as in DnD perhaps, but it's not as balanced and smooth a progression as the chart would imply.

 

Also I really wished personality hadn't been tied up with combat statistics - especially if realism isn't a priority. Trying to work my PC around a high Per/Int/Res build is a no brainer for dialogue options, but also they've been very opposed to allowing us to customize NPCs presumably due to something like "resolve might not be desirable for your preferred priest build but Durance's character requires it" while in game resolve doesn't actually allow Durance to do anything for your party in conversations.

 

I do think however, that they're better than DnD and have allowed for a nice variety of builds for some classes, and more importantly that the different builds play substantially differently. There are still some classes where I'll almost automatically min-max certain attributes.

 

The way deflection works means that low amounts don't do a whole lot for an already low deflection class, same for constitution for lower endurance classes. So maximizing their offensive capabilities becomes a no brainer. Especially because dexterity almost doubles as extra interrupt resist because attacking/acting/casting faster = less chance to get interrupted. The buff to con made dumping it more undesirable at least, but I'm still comfortable knocking it down a bit even on PotD.

 

Resolve at the moment I'd say is the weakest/hardest to justify attribute overall. There are some classes you might not dump it on, but I could build a full "power gamed" party without going above 10 in it on any character - same can't be said for anything else. I have a reason to stack Might, Dex, Per, Int, and even Con for some melee. I only take high resolve on my PC character(chanter tank) for dialogue options.

 

 

To be fair though, the more progression a game has the harder it is to balance things for both bigger and smaller numbers. Scaling causes tons of issues and some things end up nigh obsolete earlier on or later on. For example I don't care for high dex on a caster with only 5 spells or whatever to cast in combat, might and perception are way more important for making those few spells hit and hit harder. But later on when I want to cast 5-10 buffs just at the start of any difficult combat dex becomes dramatically more beneficial.

 

I'd really like to see what an IE style game with much less dramatic progression throughout would be like.

Edited March 24, 2016 by Odd Hermit

[DreamWayfarer]

There are also some no-brainer dump stats - fighter-ish classes have several builds that gain so little from Int you can drop it to 3 without a second thought.

They also have several builds that benefit enough from INT for you to seriously consider raising it instead of other stats, so I wouldn't say it qualifies as a no-brainer dump stat, as much as a build-dependent one. Like attributes were supposed to be anyway.

[hilfazer]

 

Resolve at the moment I'd say is the weakest/hardest to justify attribute overall. There are some classes you might not dump it on, but I could build a full "power gamed" party without going above 10 in it on any character - same can't be said for anything else. I have a reason to stack Might, Dex, Per, Int, and even Con for some melee. I only take high resolve on my PC character(chanter tank) for dialogue options.

 

 

That's because Resolve is very obviously imbalanced. I'm surprised it's still the case in WM2.

Resolve and Perception want to mirror each other. But they fail. +2 reflex is as good as +2 will, +3 interrupt is as good as +3 concentration...

 

... but +1 deflection is not as good as +1 accuracy!

 

It's half as good. Let's double this and see what happens.

[Ineth]

But Resolve has the best dialog choices...

[hilfazer]

I knew i forgot something!

[limaxophobiacq]

 

 

Resolve at the moment I'd say is the weakest/hardest to justify attribute overall. There are some classes you might not dump it on, but I could build a full "power gamed" party without going above 10 in it on any character - same can't be said for anything else. I have a reason to stack Might, Dex, Per, Int, and even Con for some melee. I only take high resolve on my PC character(chanter tank) for dialogue options.

 

 

That's because Resolve is very obviously imbalanced. I'm surprised it's still the case in WM2.

Resolve and Perception want to mirror each other. But they fail. +2 reflex is as good as +2 will, +3 interrupt is as good as +3 concentration...

 

... but +1 deflection is not as good as +1 accuracy!

 

It's half as good. Let's double this and see what happens.

 

 

I to feel that resolve is the weakest attribute in the game since con was changed to +5% health/stamina.

 

+2 deflection would be to much though I think (it would let you make crazy tanks too easy) but 1.5 wouldn't really look very good and neither would adding a third minor effect to Resolve. It's not an easy problem to solve.

 

If anyone thinks resolve isn't the weakest, I'd say consider the spell Crowns for the Faithful. This spell gives +25 resolve for everyone in the AoE and while very strong it is not really a gamebreaking spell.  Which other attribute could you give a +25 bonus to with the same AoE and Duration whithout it being ridiculous? None. +75% damage to entire party? +75% action speed? More than double health and stamina? +150% AoE and +125% duration? +25 accuracy and +75 interrupt? All those would be broken good.

[Odd Hermit]

Way back in the beta I suggested Resolve be the +duration stat, for both gameplay and RP reasons(will to keep an effect going under pressure or whatever).

 

I'm not sure what you'd do with Intelligence at that point, but Resolve would then have a purpose for far more builds.

 

 

Probably too late for that though. 

[Trashos]

The proof is in the pudding. I will like Josh's system better when:

 

1) Character progression is not broken. Right now, it is. That's what I call it when all my derived statistics rise automatically by a huge amount on every level up.

 

2) Encounter design gets interesting. I haven't played the expansions (so I guess I may be a bit harsh here), but the base game's combat encounter design is far inferior to the one in Shadows of Amn.

 

3) Difficulty gets tuned up. It is weird to claim that the system makes builds viable, when the reality is that the game is too easy, so basically anything will work. Next time give me a challenging game in order to judge.

 

Having said all that, I do realize the following:

a) Innovation is hard work and takes time to get it just right. I respect what Josh is trying to do and I am willing to give it more time. I generally like people who try to innovate.

b) It is hard to get everything right on the first version of a game. BG1 to me is just "that game that made BG2 possible". Maybe PoE2 will also be better than PoE1 in the above aspects.

 

So. The proof is in the pudding. Give me a game that is better than BG2, and then I will like this system more.

[Batbro]

Update: You can now watch a video of the whole talk.

 

It's pretty much the same, more good stuff, but if you wanted to hear Josh Sawyer expanding on the bullet-points from the slides, here you go.  Also includes a brief Q&A session afterwards.

 

They'll probably include this in their next blog update anyway, rendering sharing this moot, like they did with the slides, but whatever.

[Valorian]

Nice presentation.

 

I'm going to mention two examples that he made:

"The clumsy, determined rogue (low Dex, high Res). The idiot muscle wizard (low Int, very high Might)."

 

In the first example you can notice a discrepancy. A high resolve person is in no way clumsy in PoE's setting. Resolve improves the likelihood to deflect attacks and being good at deflecting attacks is not a clumsy person's virtue. Words, despite souls, still have meaning.

 

Voice of opposition: "But D&D also has game-isms in its ability scores."

Sure, but to a lesser degree. On a scale of 1 (absolute gameism) to 10 (no gameism), I'd say PoE's attributes would be a 3 and D&D's a 7. This is clearly not exact mathematics, but subjective observation as 4 out of 6 attributes in PoE are quite gamey.

 

 

2nd example; the famous muscle wizard.

Here you see a missed opportunity to sell a new idea. Instead of fomenting the "muscle wizard" expression, they could've told a story about magical energy affecting your ability to move heavy objects and cause more damage with weapons, because Might is also magical/spiritual power. Physical strength enables spiritual strength to flow more easily through your body, but if your spiritual/magical strength is very high, your body can be weak and still accomplish extraordinary feats (no practical difference in PoE's setting, sadly). Basically leaving to the player to imagine for their character which strength prevails.

In short, a high might wizard doesn't have to be a muscle wizard.

[PutCashIn]

I do like this thread, having just brought POE today (at full price on Steam, more the fool for me for not waiting for a special deal, I guess) I would be very interested to hear what the obviously competent posters on this self necromancied thread have to say about the attributes in Tyranny - especially Resolve(or its equivalent), as 'as' suggested in this thread, I find a reduction in affliction time almost meaningless, it is far better to aim to making the enemy focus on standing back up than worrying if I ever need to, which I don't as I just smashed you in the face etc etc .. .

[Elric Galad]

Resolve (in Tyranny) and Intellect (in PoE) both increase the affliction time.

And it is VERY powerful, at least in the right build. Affliction includes Damage over time (so basically, damages), relevant debuff (such as Terrified or Blind) and hard Crowd Control that completely prevents ennemy from attacking. 
Furthermore, Intellect also increases your own buff duration. Some buff are not that long, so it's quite relevant too. 

It should be underlined that high Accuracy (which could be increased by buff) + Hard Crowd Control is the easier way to win at the higher difficulty levels.

 

However, if your character does not rely too much on such abilities, it is true than dumping Intellect does not hurt that much compared to other classes.

 

But be warned one needs several characters with high INT for an optimal party.

Edited March 29, 2018 by Elric Galad