Design questions about the affix / suffix system.

Hello folks :)

I've posted the same message on reddit a while ago but got no answer there, so I'm tying the official forums to pick your brains !

I'm trying to design a small RPG game and I'm wondering how to organize the stats on my items. If you take PoE, there a few main mechanisms that shape-up the item dropping/crafting systems:
- stats are divided into two mutually exclusive categories: prefix and suffix.
- some stats are slot-restricted (ex: critical multiplier cannot be rolled on boots)
- unique items can bypass these rules and introduce new stats to items

I'm wondering what do you think are the reasons / pro / cons of that system? I'm working on a game where the player can decide what stats he wants on his items so we can set aside the influence of having prefixes/suffixes on the item drop/availability. I want to empower to player to theorycraft his builds with no right/wrong answer regarding the stats on his items.

The last part about unique items bringing new stats, often build-enabling but with some form of drawback in term of raw power seems perfectly fine and understandable.

What I can't really figure out is why the prefix/suffix system is beneficial to the player's experience. It does restrict what stats combination you can get on your items and so it seems to be detrimental to the character's building depth that we all love in PoE. It seems to mainly be there for balance purposes and to avoid some stats to coexist but since we can already have 3 prefixes and 3 suffixes, it does not limit the craziness of what we can do by that much.

For example, on chests, we cannot have %ar, flat ar, hybrid %ar/stun and life on the same chest. Do you really think it is OP? Is having life compared to a high resistance roll that much better? In that particular case maybe it is, but then wouldn't a simple balance pass fixes that? After all, these are all defensive stats and in a world where they are all equally powerful and perfectly balanced (utopia!), the pre/suff system feels more like a burden than something interesting to build upon.

I’m not against restrictions regarding item stats since sometimes it can force the player to make tough choices and that is good in the long run but in the current iteration of the itemization in the game I don’t feel PoE achieves that very well.

Regarding the stats restriction per slot, I don’t mind it that much but I also don’t see the benefit of it. It does allow you to control how much of stats Y the player can get on items since you know it can spawn only on certain item slots. But if you could have the same stats on every item slot, you could still control the stats growth by lowering its amount par slot. A small drawback of doing that is that the item-power overall will decrease since we have to care for extreme cases now.

I’m neutral to the fact that having 3 items where you have a stats X or 9 items where you can have that stats leads to different progression patterns if you want to have the maximum of that stat. In the first case, you only have to care about 3 items and you are done. In the other case, it is a longer process that may be more rewarding in the long run but may feel a bit underwhelming when you are starting to gear-up and only got the stat on one or two items. What do you think?

I’m a bit lost in my thoughts and I hope all of this make sense. I’m very interested in you guys opinions on the PoE prefix/suffix/stats-per-slot restriction system. Keep in mind that I’m working on a game where the player can choose easily all the stats on his items, so RNG has no place there. And if a GGG dev is hanging around, I’d love to have your insights on this! Cheers guys.

"

Mydi wrote:

What I can't really figure out is why the prefix/suffix system is beneficial to the player's experience.

The benefit it that it's comprehensible - especially for naming magic items with on prefix and one suffix. The system is at least as old as Diablo I and somehow stuck.

Personally, I would do implement this differently:

Create a list of all stats that any item can have.
Group the stats in categories.
Give any base item a number of slots (possibly in relation to rarity) and assign categories which can fill each slot. Also define which categories are supposed to be mutually exclusive or have a mutually bias towards each other or a bias to other factors (for example class or character that it is dropping for, monster it drops from, location where it drops).

For each stat create a number of tiers which define the different ranges a value stat can roll and assign conditions when this tier can be rolled (most common would be item level, but others like base type or previously rolled stats are also possible).

This would give you a much greater flexibility when you polish your item design, allowing you to create a more unique gaming experience.

"

Mydi wrote:

What I can't really figure out is why the prefix/suffix system is beneficial to the player's experience.

It's about systematization of data to achieve the ability to operate with it in your favor, as a developer. Then it becomes a part of the player's experience, through the game's internal rules and mechanics, set, once again, by the developers.

Such an approach is by no means restrictive, it's what the developer manage to make out of it.

As long as you keep working, you'll find out that it's impossible to achieve any kind of ingame balance and predictive (controlled) behavior, without systematized rules and values locked in brackets.

Making every stat available on every item and giving players full control over crafting would make it way too easy to min/max builds. Experimenting with builds might seem fun at first, but it would ultimately be detrimental for the game's longevity. Of course, it may be that your target audience is more casual than PoE's, in which case that choice could be fine.

Diablo III changed the traditional prefix/suffix division into primary and secondary stats, with the most important stats being primary. They also reduced the number of different stats. Set item modifiers do not take up normal stat slots but are immensely powerful. The end result is that for most item slots the choice of stats is very clear. It's also the same choice for most slots. You farm for the correct set or legendary item and check if it has at least three of the four stats you want. If it doesn't, you discard it. If it has three correct stats, you reroll the wrong one into a correct one. If it has four correct stats, you reroll one of them for a better numeric value. The whole gearing process is over in less than a week.

"

The benefit it that it's comprehensible - especially for naming magic items with on prefix and one suffix. The system is at least as old as Diablo I and somehow stuck.

Personally, I would do implement this differently:

Create a list of all stats that any item can have.
Group the stats in categories.
Give any base item a number of slots (possibly in relation to rarity) and assign categories which can fill each slot. Also define which categories are supposed to be mutually exclusive or have a mutually bias towards each other or a bias to other factors (for example class or character that it is dropping for, monster it drops from, location where it drops).

For each stat create a number of tiers which define the different ranges a value stat can roll and assign conditions when this tier can be rolled (most common would be item level, but others like base type or previously rolled stats are also possible).

This would give you a much greater flexibility when you polish your item design, allowing you to create a more unique gaming experience.

You mention grouping the stats into categories. I've got nothing against it but could you give us more details about what kind of categories would you make ? why ? how would you make them interact with item slots? What wold be the benefit for the game's itemisation system ?

In my game, the stat's "power" or tiers as you mention it is tied to something else. I'm, for now, only concerned about the stat in itself, not its strength.

"

It's about systematization of data to achieve the ability to operate with it in your favor, as a developer. Then it becomes a part of the player's experience, through the game's internal rules and mechanics, set, once again, by the developers.

Such an approach is by no means restrictive, it's what the developer manage to make out of it.

As long as you keep working, you'll find out that it's impossible to achieve any kind of ingame balance and predictive (controlled) behavior, without systematized rules and values locked in brackets.

You almost lost me...or maybe you did, I'm not sure :) What i've understodd your your answer is that a game need rules. So far, I agree 100%. I initally was not saying that I want a system overly permissve. I want a system that emphasize choice. I know that sometimes having restrictions is good to promote different choices / bahvior / builds.

That beeing said, I'm still wondering about the pro & cons of stats system in modern hack n slash such as D3 or PoE. I'm trying to find what kind of effect these systems have on the player's exeprience.

"

Making every stat available on every item and giving players full control over crafting would make it way too easy to min/max builds. Experimenting with builds might seem fun at first, but it would ultimately be detrimental for the game's longevity. Of course, it may be that your target audience is more casual than PoE's, in which case that choice could be fine.

Diablo III changed the traditional prefix/suffix division into primary and secondary stats, with the most important stats being primary. They also reduced the number of different stats. Set item modifiers do not take up normal stat slots but are immensely powerful. The end result is that for most item slots the choice of stats is very clear. It's also the same choice for most slots. You farm for the correct set or legendary item and check if it has at least three of the four stats you want. If it doesn't, you discard it. If it has three correct stats, you reroll the wrong one into a correct one. If it has four correct stats, you reroll one of them for a better numeric value. The whole gearing process is over in less than a week.

In the context of a game like PoE or D3, I agree that allowing players a full customisation of their items is a bad idea. In my game, it will not be about farming items to drop the one you want but spend time figuring out what kind of item would better suits your needs. In PoE or D3, you basically know what stats you want in each item slot. I'm trying to figure out a stat/item system were the player has to think about what stats configuration is better for his build. There should not be an easy min/max answer to what stats he want in each item slot.

Thanks everyone for your answer :)

D2 did it right. It reserved the top tier of affices for blue items and made rares select from lower tiers. This gave blue items a limited role in the game. In PoE here, the only thing blues are good for is turning into rares... "crafting". D2 did much better simply because they put that limit on rare items.

"

Mydi wrote:

I want a system that emphasize choice.

Which has nothing to do with your concerns about the popular affix system. Give your players the choices. But mind that they also can't have unlimited of these.

Are PoE and Diablo series the only aRPGs you've played, as you constantly give them as an example? Familiar with TQ? There's affix system as well, but affixes are hybrid and coupled in one way or another. Then properly mixed together. It encourages diversity, itemization in TQ/GD is quite good because of it. But there are limitations as well. Intentional.

I'm trying to get what do you exactly want when saying "choice", if you are aware there will inevitably be limitations and there's no unlimited choices if you intend to balance the game.
The popular affix system doesn't limit choices.

"

torturo wrote:

"

Mydi wrote:

I want a system that emphasize choice.

Which has nothing to do with your concerns about the popular affix system. Give your players the choices. But mind that they also can't have unlimited of these.

Are PoE and Diablo series the only aRPGs you've played, as you constantly give them as an example? Familiar with TQ? There's affix system as well, but affixes are hybrid and coupled in one way or another. Then properly mixed together. It encourages diversity, itemization in TQ/GD is quite good because of it. But there are limitations as well. Intentional.

I'm trying to get what do you exactly want when saying "choice", if you are aware there will inevitably be limitations and there's no unlimited choices if you intend to balance the game.
The affix system doesn't limit choices.

I haven't played titan quest but that is definetely something I should do !

My concern is that I want limitations only where they bring something to the game. I want the restrictions of my item/stat/slot system to make the theorycrafting of a build and of what stats you want to be more interesting that without these restrictions.

As I mentionned, I'm working on a game with no drop system so the interest of having random affixes (whether there are pre/suff) is null for me. This is one of the biggest benefit I see in games like PoE of such an item/stat system. And since it does not bring anything for my game (it only would bring the restriction but no extra reflection would needed on the player's side to make his builds), I'm questionning whether or not to keep something like that.

I'm pretty sure I'm missing a few key points other than this, that makes the current system a good one even when you set aside the rng / item drop factor. That is what I am looking for.

So, by choice I mean that I want a system where you have X item slot with Y stats slot and that chosing them all is something interesting / challenging / rewarding where there is no min/max cookie-cutter right-or-wrong answer that can be easily math'ed out. An example of that would be like having a stat slot limited to two stats where neither of them rbing you something taht you can say "hey, this is 100% better for me than the other stat". I want both options to be on par with each other. Maybe one is better in situations X and maybe the other one is better in situations Y and as long as both situations are encountered frequently this is fine.

There is also the issue that I do not want the player to think purely in EHP (effective hit points) and DPS but some stats will inevitably provide a EHP/DPS boost and these needs to be accounted for in my stats/item system. [not the easiest task ^^]

I hope I clarified a bit what I am looking for :)

"

Mydi wrote:

You mention grouping the stats into categories. I've got nothing against it but could you give us more details about what kind of categories would you make ? why ? how would you make them interact with item slots? What wold be the benefit for the game's itemisation system ?

Let's take a knife for example:

It was dropped by a lvl 35 goblin bartender in the 'goblin party district'.
It is of base type 'goblin party knife'.
And somehow you determined it is your equivalent of a rare item.

A kitchen knife has a maximum of 4 stat slots and because it is rare, three to four will be filled with stats.

Category A (damage): flat damage, percentage damage, both in smaller amounts
Category B (attack): better hit chance, faster attack,
Category C (leech): mana leech, life leech
Category D (on hit special): poison on hit, bleeding on hit
Category E (special vs): plus damage against trolls, plus damage against dwarves
Category F: nothing
You have decided that such a knife can use the following categories for each slot:

(Slot 1 and 2 would be those that are also used for a standard magic item)
1: A or B (always good to stab things)
2: A, B, C or D

(Slot 3 and 4 are the rare item slots)
3: the same as 2 (rare knife stab even better)
4: C, D, E or F (E is a rare item only category, F because we said it has either three or four stats)

The result would be something like:
goblin party "Stabby stab thing" with +damage, bleeding on hit and +percentage damage and room for a fourth stat. Simplest solution four crafting the fourth stat would be to just allow any stat that was not previously rolled.

So far it's not so different from the prefix and suffix system.

Now imagine you add another set of categories with similar stats for spell casting, some more for elemental damage and a set of skill modifiers (+1 to stabbing things) and adjust the rng-calculations while you are determining stats:

All daggers can roll for slot 1 any stat from basic attack (category A and B from above), basic spell or basic elemtal attack category. Goblin party knives have a higher chance to roll an attack mod than the other two.
All daggers can roll for slot 2 any stat from the extended categories with the chance for the stat being fitting to the one from slot one being the highest (slot 1 is attack: second attack > elemental > spell).
Slot 3 can be filled with any stat from the basic category. Chances are changed based on previous slot rolls.
Slot 4 gets a stat from categories of the dominant aspect or a fitting skill bonus (e.g. no fireball on a melee dagger).

This would create items that make somehow sense. For crafting you could just allow stats that could roll with a certain percentage.

Pure crafting a white item could look similar:
Slot 1: Choose any stat from one of the basic categories
Slot 2: Choose any stat from the advanced categories - fitting stats of the same aspect are either cheaper than non fitting stats or have a higher chance to roll better values.
and so on....

I have to admit, my examples for the stats are a little boring, because we've all seen those a hundred times before.

Benefits:
- lots of room for custimization of the base items allowing you to make them really distinct
- no arbitrary limitations of the prefix/suffix system (may three of those and three of the others)
- slots and categories (if you name them properly) should allow you easy to write item property tables
- and (hopefully I showed you that) you can make lots of fun stuff with this idea by making categories roll more likely together, make them mutually exclusive or tie them to any other condition you seem fit (only rolls on Mondays, can only be rolled by bald guys with a grey beard...)

"

Mydi wrote:

I'm trying to figure out a stat/item system were the player has to think about what stats configuration is better for his build. There should not be an easy min/max answer to what stats he want in each item slot.

When putting gear on a character, it may look like this:
- Enough resists for max resists,
- as much life as possible,
- combined with as much damage as possible,
- with at least moderate primary defensive stats
- and some crit and leech as bonus.

We all know what item slots can give you the stats I just mentioned: many/most/all.

If you want decisions, one way would be to limit the possibilities, examples (a little exaggerated):
Life can only be on chest armor or belts.
- And you get either +life or damage reduction.
- Rings give you either resistances or leech.
- Gloves give you either attack speed or +damage.

Main problem would be to keep the item variety high enough and interesting.

A different approach would be to limit the quality of stat rolls on an item:
- A sword can either have godly +damage and moderate attack speed or the opposite.
- A chest armor can either have high +life or a high armor stat.
- A rare helmet can have one very high, two moderate and two low stats of Life, two individual resistances, armor and movement speed.

Of course, every stats have to be more or less equally important. A choice between +all resists and light redius isn't really one in most cases.

I'd suggest reading "Design Patterns of Successful Role-Playing Games" * and thinking about how the conflict mechanics in your RPG are going to work.

Other than that, go build the game with a small, basic set of modifiers first. You can still balance the stats and add more complex mods later.


* see http://rpg-design-patterns.speedykitty.com/doku.php/start


Edit: Been mulling about this some more - you'll need to look at your game balance as a multidimensional optimization problem. You want build diversity, so you need to make sure that there is not a single optimal build, but a set of multiple optimal builds (see https://en.wikipedia.org/wiki/Pareto_efficiency). Designing a stat system and content so that "find the optimal build" leads to a large variety of solutions is highly non-trivial.

At the most basic level, you could look at damage dealt and cast speed:
Skill A - 1 dmg per seconds
Skill B - 0.5 dmg per 0.5 seconds

Both skills have a dps of 1, so this trivial example would have 2 viable builds.

Adding different damage types and resistances, damage over time and defensive skills - it becomes much more complex ofc.