How to Fix the Attribute Design in Pillars of Eternity

[Sensuki]

Perception and Dexterity would probably be a fun build - would give you more chance to get interrupts, sure.

 

Remember that interrupts reduce enemy DPS by 0.5s per Interrupt, I don't know if that's actually as good as simply just pumping Might instead, I'm not sure. We (well, Matt) haven't done that math as it's more complicated.

 

I'm glad everyone's receptive to our proposal

 

It would just be a shame if by release, the system didn't achieve the design goals properly as the ideal behind the system is really cool.

Edited September 12, 2014 by Sensuki

[Mr. Magniloquent]

You two have certainly outlined why attacking more is always good, but not necessarily better than alternatives--no doubt. I made sure to read the paper thoroughly before commenting.

 

Chants are still an ability, so if Dex were to improve ability/spell speed, then Chants would be no exception. This would cause their chants to progress X% faster, compounding to output far more invocations than otherwise. The effect would be similar to your spellcasting example on page 16. For Ciphers, this is all the more true. Faster attack speeds would mean more resources generated over a shorter period of time. They spend less time generating and less time casting--allowing them to do far more of both. Because of their resource type, doing things faster is always the best thing to do--particularly in the case of the Cipher.

[Matt516]

Perception and Dexterity would probably be a fun build - would give you more chance to get interrupts, sure.

 

Remember that interrupts reduce enemy DPS by 0.5s per Interrupt, I don't know if that's actually as good as simply just pumping Might instead, I'm not sure. We (well, Matt) haven't done that math as it's more complicated.

 

Yeah, an interrupt build is basically a support build. Which makes sense - an extremely perceptive and dextrous character would naturally be able to get in enemies' way and interrupt their flow. Kind of a trickster build - I can see the potential for a PER/DEX Rogue working pretty well.

 

And now you see the increased character build capability and diversity this system offers.

 

Re: magnitude of DPS reduction from interrupt - what it boils down to is that it reduces the enemy's effective attack speed. By how much? Depends on the ratio of your attack speed to the enemy's attack speed, who attacked first, what weapon you're using, their concentration, and a whole host of other factors. It'd be very very difficult to boil Interrupt down to something that could be displayed on a single plot (which is why I haven't tried). I may, in the future, see how far I can get with Interrupt DPS calculation - but for now, we can simply assume that Interrupt can be easily tuned to be as strong or as weak as OE wants it to be (they've got a few global variables in place).

[Matt516]

You two have certainly outlined why attacking more is always good, but not necessarily better than alternatives--no doubt. I made sure to read the paper thoroughly before commenting.

 

Chants are still an ability, so if Dex were to improve ability/spell speed, then Chants would be no exception. This would cause their chants to progress X% faster, compounding to output far more invocations than otherwise. The effect would be similar to your spellcasting example on page 16. For Ciphers, this is all the more true. Faster attack speeds would mean more resources generated over a shorter period of time. They spend less time generating and less time casting--allowing them to do far more of both. Because of their resource type, doing things faster is always the best thing to do--particularly in the case of the Cipher.

 

Well yes, doing things faster is always better than doing them slower. Even for characters with limited spells, getting those spells off faster rather than slower is a definite advantage. The point we try to make is that doing things faster is rarely if ever better than simply doing things better - whether "things" is casting buffs, hitting dudes, etc. IAS, if implemented, would improve everything - but not beat any attribute at its own game. 

 

Hence the "Jack of all Trades, Master of None" nature of the attribute.

[Sensuki]

Chants are still an ability, so if Dex were to improve ability/spell speed, then Chants would be no exception. This would cause their chants to progress X% faster, compounding to output far more invocations than otherwise.

I actually don't think that would be the case, I think you have to activate the chant but it's like an aura, or a passive. The speed would not be affected by IAS I don't think. And if it is, they can always change it so it isn't.

Edited September 12, 2014 by Sensuki

[illathid]

Looks good. Definitely makes the mechanics of building a character a bit clearer.

[Lasweetlife]

Josh himself put out the call for ideas to help solve the issue with perception and resolve and Sensuki and Matt516 have answered. From reading the doc it sounds like it would definitely allow for some interesting choices in character creation and gameplay. I would definitely love to get a chance to test of these proposed mechanics in PoE.

 

@Sensuki and @Matt516, given the respect the two of you have earned in these forums, I think you should put your names in the thread name, just to give it that extra bit of gravitas.

Edited September 12, 2014 by Lasweetlife

[Sensuki]

That's why we went to all the effort. I know he listens when people make strong cases, so we thought we'd make a strong case. That's also why I addressed some of his personal reservations in the paper as well.

Edited September 12, 2014 by Sensuki

[Mr. Magniloquent]

You two have certainly outlined why attacking more is always good, but not necessarily better than alternatives--no doubt. I made sure to read the paper thoroughly before commenting.

 

Chants are still an ability, so if Dex were to improve ability/spell speed, then Chants would be no exception. This would cause their chants to progress X% faster, compounding to output far more invocations than otherwise. The effect would be similar to your spellcasting example on page 16. For Ciphers, this is all the more true. Faster attack speeds would mean more resources generated over a shorter period of time. They spend less time generating and less time casting--allowing them to do far more of both. Because of their resource type, doing things faster is always the best thing to do--particularly in the case of the Cipher.

 

Well yes, doing things faster is always better than doing them slower. Even for characters with limited spells, getting those spells off faster rather than slower is a definite advantage. The point we try to make is that doing things faster is rarely if ever better than simply doing things better - whether "things" is casting buffs, hitting dudes, etc. IAS, if implemented, would improve everything - but not beat any attribute at its own game. 

 

Hence the "Jack of all Trades, Master of None" nature of the attribute.

 

I understand what you're saying. That's why I opened my comment with stating exactly that. My whole point is that there are two notable exceptions of IAS being a worthy compromise. These are in the cases of limited resources (Druid, Priest, Wizard) and infinite (Chanters, Ciphers). For the former, they do not spend most of their time using their name-sake abilities. The utility of being able to cast the infrequent spell 30% faster is marginal. Yes, it can enhance tactics, but not quite the same way as for a character which is expressed primarily through attacking.

 

Chanters and Ciphers make the opposite end of the spectrum. Chanters chant continually, and must chant to project invocations. Doing this comes at no cost, and is done continuously. Chanting and invocations are the prime actions of the class; therefore, chanting faster is always better than alternatives. The same holds true for a Cipher, but on a greater scale. They generate resources through standard attacks. They will always need to attack and always need resources. Time spent casting spells is time not spent generating focus. Faster casting times = more focus. Gathering focus more quickly = more spells. This is the primary feature of the class, so it will always be best to do these things faster.

 

I'm not trying to knock the idea. I'm generally in agreement with it. I've said this multiple times. I makes sense that while all attributes will be good, some classes will still favor some over others because of the inherit class design. It just immediately became poignant to me about the ramifications IAS changes would have for resource (in)dependent classes.

 

Chants are still an ability, so if Dex were to improve ability/spell speed, then Chants would be no exception. This would cause their chants to progress X% faster, compounding to output far more invocations than otherwise.

I actually don't think that would be the case, I think you have to activate the chant but it's like an aura, or a passive. The speed would not be affected by IAS I don't think. And if it is, they can always change it so it isn't.

 

That is an unknown. So much in that Chants have a duration, it is safer to assume that they function more like regular spells than a static aura.

[PK htiw klaw eriF]

 

 

I have to agree with the proposed changes.

 

Would the bomus in recovery be applied before or after the armor penalty?

It would be applied to everything. So previously if you were 1s attack, 1s recovery, 1s armor penalty, that's 3s total attack time. Then with 30% IAS (15 DEX under our system) that becomes 2.3s total attack time.

 

And just to reiterate - I know the paper's long (boy do I know that xD), but we'd very much appreciate if anyone stumbling on to this thread takes the time to read the paper before replying - I've seen 1 or 2 posts so far that are answered in the paper, albiet maybe in the middle of a long paragraph or something.

I'll take some time to thoroughly read it when I get off work.

 

With the values you've given, it does seem things could get chaotic with a max Dex, naked, dual fast weapons character under the effects of a haste spell. Admittedly a min Dex character in plate mail casting long spells is too. I'm just trying to play Devil's Advocate.

 

Edited for clarity, forgot DA was a game abbreviation.

Edited September 12, 2014 by KaineParker

[Sensuki]

That is an unknown. So much in that Chants have a duration, it is safer to assume that they function more like regular spells than a static aura.

I will ask about them then.

 

The same holds true for a Cipher, but on a greater scale. They generate resources through standard attacks. They will always need to attack and always need resources. Time spent casting spells is time not spent generating focus. Faster casting times = more focus. Gathering focus more quickly = more spells. This is the primary feature of the class, so it will always be best to do these things faster.

All good points, and your theory may be correct - perhaps you have found the class that the attribute would be one of the best for

 

I don't think it would be OP though, as that speed comes at the cost of other attributes. You could probably make a pretty nice glass cannon build with that actually. That would be cool.

 

But this is the type of stuff that I think our proposed system encourages. Whereas currently you just pump might, int and dex, maybe con if youre a tank, and go to town. It's really a non-choice every time.

Edited September 12, 2014 by Sensuki

[Razsius]

Bloody superb guys. Looking through all that it sort of makes me understand that the attributes system is a very closed but very balanced system in and of itself.  What I mean is that going from 17 to 18 Might is "only" a 2% increase in damage.  Seeing as how damage ranges in PoE seems to be so much broader than what i'm used to it makes the stats in general feel less impactful despite the fact that I know a swing of like 10 accuracy or 10 deflection is *huge*.  One accuracy tends to be hard for me to quantify or "feel" as a gamer.  Ten, however, is rather noticeable.

 

Is there any credence to the statement that the attributes themselves don't matter at all (or much)?

Edited September 12, 2014 by Razsius

[Matt516]

True, true - all good points. Some classes will invariably favor some attributes above others simply because of differences in their design and role.

[Matt516]

Bloody superb guys. Looking through all that it sort of makes me understand that the attributes system is a very closed but very balanced system in and of itself.  What I mean is that going from 17 to 18 Might is "only" a 2% increase in damage.  Seeing as how damage ranges in PoE seems to be so much broader than what i'm used to it makes the stats in general feel less impactful despite the fact that I know a swing of like 10 accuracy or 10 deflection is *huge*.  One accuracy tends to be hard for me to quantify or "feel" as a gamer.  Ten, however, is rather noticeable.

 

Is there any credence to the statement that  themselves stats don't matter at all (or much)?

 

They're smaller relative boni than the D&D attributes. IMO it's more about broad stat ranges though - sure, one point in any won't make a huge difference, but every little bit helps. In D&D, a melee character had to have between 12 and 18 STR to be even remotely viable. So you effectively only have a 6 point range to play around with when trying to build "viable" characters. PoE's system is designed such that the entire stat range is viable. That doesn't mean the stats don't matter though - just that the game isn't punishing you for not speccing into any given stat. The difference between a MIG build and a CON build and a DEX build will still be quite noticeable though.

[Sensuki]

Is there any credence to the statement that the attributes themselves don't matter at all (or much)?

Well one of the goals of Josh's system is that it's kinda supposed to protect people from making bad characters. It's not that they don't matter, it's just that a 13 in every attribute will be a good build most likely, you just have to play to it.

 

I haven't tried a build of 13 in all stats in the current system because Perception and Resolve are terrible, but in our proposed system I wouldn't be surprised if you could get away with it.

[Mico Selva]

Nice work, Sensuki and Matt. Hopefully the devs will at least take these suggestions into consideration.

Edited September 12, 2014 by Mico Selva

[Matt516]

 

Is there any credence to the statement that the attributes themselves don't matter at all (or much)?

Well one of the goals of Josh's system is that it's kinda supposed to protect people from making bad characters. It's not that they don't matter, it's just that a 13 in every attribute will be a good build most likely, you just have to play to it.

 

I haven't tried a build of 13 in all stats in the current system because Perception and Resolve are terrible, but in our proposed system I wouldn't be surprised if you could get away with it.

 

 

Which would be fine. All around decent characters should be an option if someone doesn't want to specialize - they'll just never be quite as good at any specific thing as the specialists are.

[Seari]

I would still like to see penalties for going bellow 10(or somesuch) in an attribute, and an overall buff to the numbers to make them more impactful. Chargen is extremely important to me, and I would hate to have it be meaningless.

[morhilane]

 

That is an unknown. So much in that Chants have a duration, it is safer to assume that they function more like regular spells than a static aura.

I will ask about them then.

 

The same holds true for a Cipher, but on a greater scale. They generate resources through standard attacks. They will always need to attack and always need resources. Time spent casting spells is time not spent generating focus. Faster casting times = more focus. Gathering focus more quickly = more spells. This is the primary feature of the class, so it will always be best to do these things faster.

All good points, and your theory may be correct - perhaps you have found the class that the attribute would be one of the best for

 

I don't think it would be OP though, as that speed comes at the cost of other attributes. You could probably make a pretty nice glass cannon build with that actually. That would be cool.

 

But this is the type of stuff that I think our proposed system encourages. Whereas currently you just pump might, int and dex, maybe con if youre a tank, and go to town. It's really a non-choice every time.

 

 

The only thing it really affect for the Cipher is focus generation making one of the talent a bit useless or OP depending how you look at it (it increase focus generation). In term of spell usage, they are relying a lot on positioning, so IAS might not affect their spell slinging speed that much even if it could.

[Razsius]

I suppose what i'm getting at is: is a build of like 3 in every attribute "viable" as well?  Personally, I think 15 less accuracy would really hurt but maybe others can get it to work.  I think it should at least be somewhat important that the attributes themselves feel like they have some signficance.  While, I understand that the system is going to promote not punishing you for making the "wrong" choice.  I do think 3 of every attribute + race + culture bonuses should be considered a "bad build."

 

As for 13 in every attribute I believe it would very much work and i'm perfectly okay with that.

 

Edit: I'd consider being able to click past assigning any points at all to be a design error is all.

Edited September 12, 2014 by Razsius

[Lasweetlife]

I would still like to see penalties for going bellow 10(or somesuch) in an attribute, and an overall buff to the numbers to make them more impactful. Chargen is extremely important to me, and I would hate to have it be meaningless.

Even if any such "penalties" were illusionary?

[Seari]

 

I would still like to see penalties for going bellow 10(or somesuch) in an attribute, and an overall buff to the numbers to make them more impactful. Chargen is extremely important to me, and I would hate to have it be meaningless.

Even if any such "penalties" were illusionary?

 

Explain.

[PrimeJunta]

I suppose what i'm getting at is: is a build of like 3 in every attribute "viable" as well? 

 

It shouldn't be IMO. I don't think the "no bad builds" philosophy need extend to cases where the player voluntarily decides not to distribute character points.

 

(Right now it kind of is because the attributes lack oomph, but I understand that's about to change.)

[J.E. Sawyer]

Hey, guys.  First of all, thank you for putting this together.  It's great work.  Your research is (obviously) extremely in-depth and well thought-out.  This will sound like BS, but just before lunch, I wrote this chart on my board:

 

 

And while waiting for my food, started reading your paper.  I think we've reached largely similar conclusions, though honestly mine were based more on *~ gut feelin's ~* and less on deep statistics.  

 

The main conclusions we reached internally were:

 

1) Interrupt chance should be primarily attack/weapon-based with Accuracy (or rather, attack resolution) being the modifying factor.  This doesn't entirely align with your conclusions, but it essentially decouples Interrupt from an Attribute independent of what's affecting Accuracy.

 

2) Accuracy makes as much, if not more, sense on Perception as it would on Dexterity.

 

3) Dexterity should modify Action Speed by 2% per point.

 

4) We should establish 10 as the baseline for any stat, with values below inflicting penalties.  It feels more traditional and it's extremely easy to make the math work either way (i.e., nothing "bad" really happens because of it, gameplay-wise).

 

With Resolve, we were still torn on a few issues.  We also considered putting Deflection onto one of the stats, but having a stat be purely defensive didn't feel great.  Keeping Concentration on Resolve seems good/solid/sensible.  We had discussed what I believe was an idea originally from the forums, which was having Might not affect healing output, but having Resolve affect healing received.  I think that could work well, as could simply making Endurance (FAK Stamina) be Resolve-based, with Health being Constitution-based.

 

Anyway, those are my quick thoughts, but again I want to let you know that I appreciate all of the effort you put into researching these problems.

[aeonsim]

Interesting though I'd argue that thematically Dexterity for Deflection and Resolve for speed would make more sense, as while Dex works for both, Resolve for deflection seems odd, mechanically it may be better but thematically odd.

 

Dexterity for deflection implies high coordination for dodging or deflecting an attack.

 

While Resolve for speed works along the lines of with great resolve/no hesitation they acted instantly, also while resolve is already linked to concentration and concentration makes sense when linked to speed if your concentrating solely on your actions you are likely to move an act faster than if distracted.

 

Any way I'll be interested to see Obs thoughts on this and what alternatives they had been thinking about for improving Per & Res.

Right I see that there thoughts were running along similar if not identical lines.

 

I also agree with some of the others that this may require a rebalance of the Cipher class at least with regards to speed. As they get Focus from attacks attacking faster increases the rate of focus gain while the increased action speed also increases the number of abilities they can cast in a set time period. Thus your boosting both the rate of acquisition of abilities along with the rate of use of abilities.

Edited September 12, 2014 by aeonsim