Jump to content
  • Sign Up

Recommended Posts

Time and culture may have shaped Eora’s borders, piety may have strengthened them, and conflict may have shattered them, but all things sprout from a shared root in the vast unknown.

Pillars of Eternity Guidebook, Volume One

 


A few days ago I bought the game and launched it. In return, the game launched a volley of numbers at me. Numbers I did not understand. But I like numbers. You can depend on numbers. They are always there for you. So I asked myself: how do I make those numbers sing?
 
After some time spent nudging them here and there, I realized that my past experience with Infinity Engine games and their AD&D mojo wouldn't get me far on this particular journey. And, frankly, that's great – it means that PoE will possibly be an exciting and refreshing adventure in terms of gameplay mechanics.
 
But still, what do I do with those unwieldy numbers?
 
So I started reading [1, 2] and it seems the news (OK, it's pretty old news, but new to me) is fantastic: the devs at Obsidian are experimenting with the ancient RPG formulas in an attempt to make something fresh, approachable and enjoyable for newcomers and veterans alike. So far so good.
 
But how do I navigate this brave new world of Might, Constitution, Dexterity, Perception, Intellect and Rhesuses? (Or was that Resolve?)
 
In other words, how do I build a character? How do I build a party?
 
A lot of talk about character/party builds gravitates around the so-called min/maxing. For me, this is really just another flavor of good ol' optimization problems. And actually it's pretty natural. Every time you level up a character or equip some shiny new gear, you're optimizing something. What? On the meta-level, I'd say it's your enjoyment. You build a better character, so that playing with it gives you more pleasure.
 
But what makes one character better than other?
 
This whole “enjoyment” thing seems pretty subjective and ethereal. Most people talking about min/maxing use hard facts, objective truth and gargantuan Excel spreadsheets. In fact, there are many things you would want to optimize, which makes min/maxing a multi-objective optimization problem [3]. For example, two of these objectives could be damage output and survivability. For brevity, let's assume these are the only two – I find two dimensions easier to visualize than seventeen, or one hundred twenty-three, or lady-knows-how-many-there-actually-are.
 
So, you could have multiple builds, each with different scores in damage output and survivability. Cf. Figure 1., where build B has the highest damage output, while build C is the most survivable one.
 
wOuxRv0.png
Figure 1.: Character build configuration space
 
Based on these ideas, we may try to divide all possible builds into classes, e.g. offense-oriented and defense-oriented, cf. Figure 2.
 
V7R6C2s.png
Figure 2.: Different classes of builds in the configuration space
 
We can also try to establish a notion of how good (powerful) a particular build is. This could be analogous to Pareto frontiers [4]. Builds lying on the same “frontier” would be similarly powerful, however they may achieve their power in different ways, e.g. one build would be more offensive, another more defensive. This is pretty intuitive – usually you have a limited number of attribute points to spend, and so if you want to increase one parameter (e.g. damage output) you have do decrease another (e.g. survivability). See Figure 3.
 
ZTAZXny.png
Figure 3.: Frontiers of build power
 
If I understand this correctly, the problem with D&D is that there are many “trash builds”, i.e. legal character configurations which are not powerful enough to be comfortably playable. Similarly, there are possibly some uber-builds, which make the game too easy. We may call such a situation a Variably Balanced System (VBS), cf. Figure 4.
 
XsH2ott.png
Figure 4.: Variably Balanced System (VBS)
 
The opposite of a VBS would be a Uniformly Balanced System (UBS), i.e. a system in which every character build is equally powerful, cf. Figure 5. 
 
1bfejFF.png
Figure 5.: Uniformly Balanced System (UBS)
 
I presume that Pillars of Eternity is not strictly a UBS, for the following reasons:

  • It would be practically impossible to ideally balance such a complex system.
  • When creating an RPG system, you probably would want to give advanced players some space to optimize their builds.

So, as far as I understand, Obsidian were aiming at a situation depicted in Figure 6, i.e. eliminating the “trash builds”, while still leaving some space for build min/maxing. All builds are equal, but some builds are more equal than others wink.png 
 
xAeWhlU.png
Figure 6.: PoE-type system?
 
As far as I know, many people are trying to analyze PoE character and party builds in various contexts, e.g.

Yet, a number of open questions persist:

  • Is the above reasoning about the game system correct? It seems pretty basic and straightforward, but maybe I'm making some wrong assumptions here.
  • Did anybody analyze the equations the game uses for computing various variables (damage, defense, etc.)? Which parameters are the most important? What should I pay attention to?
  • Did anybody do any number crunching in the context of optimizing PoE builds? Do you have any idea how big the configuration space actually is?
  • What criteria do you use when min/maxing your builds? A fellow researcher suggested that a good criterion could be to maximize DPS while maintaining a certain level of survivability. What are your opinions?

I'm thinking about playing a Cipher.
 
References
[1] Sawyer, J.: How To Balance An RPG, 2014, URL: http://kotaku.com/how-to-balance-an-rpg-1625516832, accessed on: 2015-04-02
[2] Hall, C.: How Pillars Of Eternity Rewrites The Rules For Role-Playing, 2015, URL: http://www.polygon.com/2015/3/25/8284763/how-pillars-of-eternity-rewrites-the-rules-for-role-playing, accessed on: 2015-04-02
[3] Wikipedia, Multi-objective optimization, URL: http://en.wikipedia.org/wiki/Multi-objective_optimization, accessed on: 2015-04-02
[4] Wikipedia, Pareto efficiency, URL: http://en.wikipedia.org/wiki/Pareto_efficiency, accessed on: 2015-04-02

  • Like 5

Share this post


Link to post
Share on other sites

Oh snap, this just got real.

 

...

 

Unfortunately, Pillars is presently much much much more a VBS than a UBS. There are myriad reasons for this, but the big ones are that (a) the armor and attribute systems tend to reward extreme offensive and defensive builds, and that (b) the options for many build types are not abundant, resulting in pigeonholing.


If I'm typing in red, it means I'm being sarcastic. But not this time.

Dark green, on the other hand, is for jokes and irony in general.

Share this post


Link to post
Share on other sites

This game favors extremes and anything in between is trash. You want two tanks in plate armor with max deflection possible and  for other characters you go with clothing and max damage possible.

  • Like 1

Share this post


Link to post
Share on other sites

Support-focused classes muddy the waters. Wizards and Priests specced for Intellect, Dex, and Accuracy, for example. These characters aren't focused on the damage or survivability of themselves, but are key to amplifying the damage and survivability of the other party members, through buffs, debuffs, and crowd control. There will be many builds developed in the future that will only work because of the support abilities provided by other builds, builds which would, considered on their own, rank on the low end of the DPS/tank scale.

 

Furthermore the balance state that exists right now will not last long. JSaw needs to be given time for a couple of patches at least before we will begin to see where the power level of the classes, races, and abilities will ultimately rest.


Exoduss, on 14 Apr 2015 - 11:11 AM, said: 

 

also secret about hardmode with 6 man party is :  its a faceroll most of the fights you will Auto Attack mobs while lighting your spliff

 

Share this post


Link to post
Share on other sites

Furthermore the balance state that exists right now will not last long. JSaw needs to be given time for a couple of patches at least before we will begin to see where the power level of the classes, races, and abilities will ultimately rest.

 

Has Obsidian indicated a desire to do serious balancing? Because the game needs serious balancing.

 

The talent list needs to be condensed, big time. So many trash-tier picks in there it's not even funny. Still, I'd rather not draw up a list of suggestions if there are no plans to fix anything.


Curious about the subraces in Pillars of Eternity? Check out 

Share this post


Link to post
Share on other sites

Support-focused classes muddy the waters. Wizards and Priests specced for Intellect, Dex, and Accuracy, for example. These characters aren't focused on the damage or survivability of themselves, but are key to amplifying the damage and survivability of the other party members, through buffs, debuffs, and crowd control. There will be many builds developed in the future that will only work because of the support abilities provided by other builds, builds which would, considered on their own, rank on the low end of the DPS/tank scale.

 

The best way to run a support character is with a setup virtually identical to that of a DPS character.

 

(Also you can't really spec for accuracy beyond Wood Elf.)

 

The talent list needs to be condensed, big time. So many trash-tier picks in there it's not even funny. Still, I'd rather not draw up a list of suggestions if there are no plans to fix anything.

 

Condensed and then expanded. There are a lot of bad picks, but even worse, there are hardly any good ones.


If I'm typing in red, it means I'm being sarcastic. But not this time.

Dark green, on the other hand, is for jokes and irony in general.

Share this post


Link to post
Share on other sites

Unfortunately, Pillars is presently much much much more a VBS than a UBS. There are myriad reasons for this, but the big ones are that (a) the armor and attribute systems tend to reward extreme offensive and defensive builds, and that (b) the options for many build types are not abundant, resulting in pigeonholing.

This game favors extremes and anything in between is trash. You want two tanks in plate armor with max deflection possible and  for other characters you go with clothing and max damage possible.

 

OK, so it seems that right now a narrow specialization in terms of ofense vs. defense is preferred, thanks for the tips wink.png

 

 

Support-focused classes muddy the waters. Wizards and Priests specced for Intellect, Dex, and Accuracy, for example. These characters aren't focused on the damage or survivability of themselves, but are key to amplifying the damage and survivability of the other party members, through buffs, debuffs, and crowd control. There will be many builds developed in the future that will only work because of the support abilities provided by other builds, builds which would, considered on their own, rank on the low end of the DPS/tank scale.

Hmm, so optimizing single characters would be pointless then. Optimization would have to consider the whole party.

 

Furthermore the balance state that exists right now will not last long. JSaw needs to be given time for a couple of patches at least before we will begin to see where the power level of the classes, races, and abilities will ultimately rest.

Yeah, I know there may possibly (likely?) be some balancing patches in the future. Nevertheless, I'm pretty interested in establishing some methodology. Once you have a successful method for finding the right builds, you can easily apply the method after every new patch, with updated input parameters (at least, that's the idea).

 

 

 

Furthermore the balance state that exists right now will not last long. JSaw needs to be given time for a couple of patches at least before we will begin to see where the power level of the classes, races, and abilities will ultimately rest.

 

Has Obsidian indicated a desire to do serious balancing? Because the game needs serious balancing.

 

The talent list needs to be condensed, big time. So many trash-tier picks in there it's not even funny. Still, I'd rather not draw up a list of suggestions if there are no plans to fix anything.

 

I don't know, but a community list of suggested adjustments might come in handy one day. Even if Obsidian won't use it, then maybe the modders?

 

@Everyone, thanks for opinions and suggestions  :)

Edited by Moorglade

Share this post


Link to post
Share on other sites

Condensed and then expanded. There are a lot of bad picks, but even worse, there are hardly any good ones.

Well if you're condensing it you'd be turning multiple options into single ones. Like I'd take the talent that gives you an extra quick weapon slot and the talent that makes you switch weapons more quickly and put them together. I'd give Field Triage the power of the one that can only impact the user but still allow it to be used on anyone. Etc...


Curious about the subraces in Pillars of Eternity? Check out 

Share this post


Link to post
Share on other sites

This was an interesting read. I'll try and answer your concerns and questions with less mathematics because I think that the math people use to 'figure out' this game is dubious at best.

 

I think figure 6 is fairly close to how PoE currently functions, with the following exception: builds that favor focusing on damage or survivability in lieu of the other should be closer to the "overpowered" spectrum, while those that dabble more towards creating a hybrid of the two should ease more towards the "underpowered" spectrum.

 

This game definitely favors extremes over hybrids for a variety of reasons. The way shields and armor work is the crux of this tendency: shields reduce accuracy and armor reduces action speed, meaning that survivability and damage increase inversely. Add to this the fact that the player is fighting against the limited tactical capability of an AI; typically, you can force the AI to become distracted by your 'tanks' while your damage dealers are able to flank for additional damage while avoiding the gaze of enemies that might chunk them. Additionally, a collect of four hybrid characters would never have the capability of two damage dealers and two tanks. You are always going to sacrifice power in either category for versatility and because of the way the AI seems to function, that versatility is wasted. A dwarf of aumaua with maxed might has a much higher damage ceiling than a hybrid (21 might vs. 19). A class that doesn't specialize in defense will have their abilities and talents split between two purposes that rarely intersect. And so on and so forth.

 

But this is mostly moot. I would say that the system is varied enough that hybrids aren't necessarily terrible. The difference between a poor build and a good one is fairly large, but that's not to say that a poor build is altogether bad so long as you don't try something completely goofy. Might applies damage, dexterity supplements that damage with speed, and intelligent supports that damage with increased duration and area. Constitution provides hit points which are then supplemented by resolve and perception's deflection. The system is fairly intuitive and clear and there is room for gimicky builds because of it.

 

I do think you're overcomplicating things, though.

  • Like 1

Share this post


Link to post
Share on other sites

Oh, and about playing a Cipher -- they're great because they consistently contribute to each encounter because they don't have any per encounter or per rest abilities, but they struggle in large, difficult, extended 'boss' fights because they have to recharge their Focus pool to cast more spells. They're ultimately a ranged class. You should max might and equal parts dexterity and intelligence. Start every fight with a blunderbuss shot for extra Focus. Fire off as many spells as you want or can -- Mind Blades is excellent for softening up large groups of enemies. It's the most accurate AoE because it typically has the capacity to hit every enemy and it will never hit your allies. Afterwards, grab a bow and fire off shots to build focus.

  • Like 1

Share this post


Link to post
Share on other sites

This was an interesting read. I'll try and answer your concerns and questions with less mathematics because I think that the math people use to 'figure out' this game is dubious at best.

 

I think figure 6 is fairly close to how PoE currently functions, with the following exception: builds that favor focusing on damage or survivability in lieu of the other should be closer to the "overpowered" spectrum, while those that dabble more towards creating a hybrid of the two should ease more towards the "underpowered" spectrum.

 

This game definitely favors extremes over hybrids for a variety of reasons. The way shields and armor work is the crux of this tendency: shields reduce accuracy and armor reduces action speed, meaning that survivability and damage increase inversely. Add to this the fact that the player is fighting against the limited tactical capability of an AI; typically, you can force the AI to become distracted by your 'tanks' while your damage dealers are able to flank for additional damage while avoiding the gaze of enemies that might chunk them. Additionally, a collect of four hybrid characters would never have the capability of two damage dealers and two tanks. You are always going to sacrifice power in either category for versatility and because of the way the AI seems to function, that versatility is wasted. A dwarf of aumaua with maxed might has a much higher damage ceiling than a hybrid (21 might vs. 19). A class that doesn't specialize in defense will have their abilities and talents split between two purposes that rarely intersect. And so on and so forth.

 

But this is mostly moot. I would say that the system is varied enough that hybrids aren't necessarily terrible. The difference between a poor build and a good one is fairly large, but that's not to say that a poor build is altogether bad so long as you don't try something completely goofy. Might applies damage, dexterity supplements that damage with speed, and intelligent supports that damage with increased duration and area. Constitution provides hit points which are then supplemented by resolve and perception's deflection. The system is fairly intuitive and clear and there is room for gimicky builds because of it.

Oh, and about playing a Cipher -- they're great because they consistently contribute to each encounter because they don't have any per encounter or per rest abilities, but they struggle in large, difficult, extended 'boss' fights because they have to recharge their Focus pool to cast more spells. They're ultimately a ranged class. You should max might and equal parts dexterity and intelligence. Start every fight with a blunderbuss shot for extra Focus. Fire off as many spells as you want or can -- Mind Blades is excellent for softening up large groups of enemies. It's the most accurate AoE because it typically has the capacity to hit every enemy and it will never hit your allies. Afterwards, grab a bow and fire off shots to build focus.

Thanks for the tips! They certainly clarify some things.

 

I do think you're overcomplicating things, though.

 

Yeah, well, that was the point, sort of  ;) In the beginning, I just wanted to avoid creating a "trash build" for my first playthrough, and generally get some grip on the system. But then some more general questions started popping up in my mind, such as:

  • What are the ideas behind PoE & RPG balancing?
  • How do you find "the best" builds? What approaches do people use?
  • Is the problem computionally tractable? Could you devise an algorithm for finding the best build? Would searching the configuration space work? Or maybe machine learning?
  • Why can't I have a dual-shield-wielding fighter tank? (Can I?)
  • What is the meaning of life?

...and so on.

 

So don't take all this too seriously. I'm just curious about how those things work. 

 

Share this post


Link to post
Share on other sites

A friend's just pointed me to an interesting paper on the subject:

Solving the balance problem of massively multiplayer online role-playing games using coevolutionary programming
http://www.sciencedirect.com/science/article/pii/S1568494614000209
 

So I'm pretty sure this is/could be an area of intensive research  ;) I wonder how those things are done in practice...

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

×
×
  • Create New...