Classes as attributes

[Greywander]

I was playing with the idea of a skill-based system, but borrowing some ideas from D&D 5e. Namely, skills having binary proficiency, and instead you mostly spend XP on feats, but also a very costly "power level" thing that increases you proficiency bonus (and probably other things) as you raise it. The idea being that your power level is a broad representation of how strong you are, but your actual performance would vary depending on which skills and feats you had.

I don't think I'll end up pursuing that particular idea, but it did develop into a sort of spin off idea. What if, instead of a singular power level that governed everything, you had several different power levels, each one linked to a specific subsystem or mode of play? We could express these as classes. For example, the warrior class might be linked to the combat subsystem, the mage class to the magic subsystem, the thief class to the stealth and security subsystem, the politician class to the political intrigue subsystem, etc. Let's ditch binary skill proficiencies and for now assume a typical skill-based system where you buy ranks in skills, so for example let's assume that that the resolution mechanic is something like "roll a die and add class ranks and skill ranks to the result".

In this hypothetical system, your choice in classes would represent the subsystems or modes of play you're most interested in engaging with, likely with some overlap. For example, being a better warrior makes you better at all kinds of combat. But a thief might be able to apply thief ranks to weapons that rely more on hand-eye coordination and less on athletic ability, like guns or crossbows. A mage could apply their mage ranks to combat spells. But warrior ranks could be applied to all of those as well. It might be strange to think of someone being better at combat magic just because they're a strong warrior, but if someone was a competent fighter and also could use combat magic, they would definitely be good with the combat magic, too. Conversely, a thief might be a crack shot with a crossbow, but they'd have little ability using a sword, or dodging or blocking attacks. Class ranks would be quite expensive, but not investing in the warrior class would mean that you would never be that great at combat, even if you could still contribute.

At some point, it occurred to me that I'd basically reinvented the concept of attributes. But I think framing them as classes might work a little better. You no longer have to be a body builder to be able to use a sword effectively, and also it was always kind of weird that wizards would somehow get smarter as they went along (though this can be explained as INT representing knowledge and education as much as it does raw intellectual ability; while your IQ might not increase, your knowledge certainly does). Also, attributes tend to be intertwined through the various subsystems, whereas this concept of classes-as-attributes links each class to a specific subsystem. The advantage of traditional attributes is that it makes it easier for different characters to contribute to different pillars of play, though they're limited to contributing in particular ways based on their attribute choices. The advantage of this classes-as-attributes is that it makes it easier for players to focus in on the one specific pillar they're most interested in. I'm not sure one is better than the other, but either one should be functional.

I think I've seen some other games do this, actually. Sort of. I think that's why a lot of the stealth skills in Vampire: the Masquerade are filed until Mental skills, as the Mental stat is linked to the exploration pillar.

Another interesting consequence of this is that these classes-as-attributes fulfill some of the same functionality as classes do in a class based system, namely, class identity and niche protection. A warrior is a warrior; it's not just that they get stuff that makes them good at combat, being good at combat is the very definition of the class. Anyone can get access to magic or pick up skill ranks in things like stealth or lockpicking, but only a mage can truly excel with all their magic, and only a thief has the skill to pull off an epic heist. Classes can also act as a point of reference in their respective subsystems, e.g. you need enough ranks in mage to unlock higher level spells or something. There might be resource mechanics or action economy that depends on your class rank, something that might be more difficult to achieve under a more pure and open skill based system.

One more thing I just thought of is that it's common in skill based systems for everyone to always grab the "best" skills, making every player equally competent at that skill. It's kind of like how almost every D&D 5e player tries to get Perception proficiency, and often Stealth as well. But with this class system, even if you put ranks into the "best" skills, you'll still fall behind the person who also put ranks into the class that uses that skill the most. The reason you can't also pick up those class ranks is that they would be much more expensive than a regular skill, and you'd want to focus more on your primary class so you don't fall behind there. Though this does raise the question of which class would apply to things not linked to a subsystem, such as basic perception checks.

TL;DR, classes act more like attributes, giving you bonuses to a very broad set of things, while skills are more focused. As in most RPGs, to be the best of the best, you need both attributes (class ranks) and skills. Class ranks would be a lot more expensive, encouraging you to stay more focused instead of branching out and grabbing a little of everything. Classes could also tie in to the mechanics of whichever subsystem they're linked to.

Thoughts on this? Does this sound like a good idea? Or is it a trap?

[Herbert_W]

Does this sound like a good idea? Or is it a trap?

Yes.
To both.

It sounds like you've independently reinvented the basic concept behind 3e/5e-style multiclassing. You're implementing it in a different way and you've approached the idea from a different direction, but the core seems to be the same.

This could be the basis for a great game. There's also a lot of traps that this sort of system can lead into, which can be avoided if you know what to look for.

A game only has one DM and will run only one encounter at a time. A player who is only interested in a few subsystems might find themselves with nothing to do in an encounter which uses a non-intersecting set of subsystems and, worse, potentially nothing to do over the course of a whole campaign if this is repeated. You can see a lesser version of this problem in DnD and classes that were designed for the exploration pillar of the game: they have features that just never get used and "waste" power budget in a campaign that neglects that pillar in favor of combat. (You don't see the full impact of this problem becasue classes are bundles of features that always include some combat-oriented ones, and this problem is partially hidden behind the related problem of the mechanics of the exploration pillar not being fleshed out, but it's still very much there.)


The obvious solution here is to ensure that every class has a way to contribute to every type of encounter. Imagine a grid showing every class/encounter combination; every square in that grid needs to be filled. That's not hard, although some squares may take a bit of thinking - it's just a matter of being aware of the issue and not overlooking any squares.


In a game where the DCs (and monster AC and saves etc.) scale upwards to match the expected capabilities of a character at a given level, skills which receive only moderate investment quickly fall off and become completely useless. You can see this problem in 3e's skill system. Theoretically it allows players to spread skill points across a variety of skills however they wish; in practice players choose a limited set of skills to maximize and then completely neglect (or just raise to a set point which is usually low, for e.g. a synergy bonus) everything else.


Every class should have options that are still be useful at low investment. For example, a poison that adds only a small amount of poison damage to an attack is still slightly useful at high levels, especially if there's no opportunity cost to apply it to weapons outside of combat. (Let's say that the rogue class can grant characters the ability to harvest poison from foes, so there's no opportunity cost in cost either.) This stands in contrast to a poison with a low save DC which becomes completely useless against tough foes. That's not to say that the latter poison shouldn't exist; it just shouldn't be the only kind of option for that poison-use class ability.


If abilities can't be used in combination, then it becomes severely suboptimal to split resources between them. You can see this problem with several multiclass combinations in 3e, but I'll pick on a wizard/fighter as that's a combination that the PHB says is common (for elves). An evenly split wizard/fighter can make attacks a bit better than a fighter of half their level and cast spells as well as a wizard of half their level - but, crucially, they only have enough actions to make full use of some of their abilities. Every turn they spend casting a spell in combat is a turn that they don't get to spend making attacks, thus wasting some of their potential. Action economy isn't the only antisynergy in this multiclass combination, but it is a big one.


This means that classes should be designed in a way that respects the action economy of a character who may also have other classes. For example, you could fix the wizard/fighter's problems by making spells benefit from a fighter's abilities just as much as weapon attacks do, or letting fighters make weapon attacks while casting a spell (or do both for different spells, e.g. making all spells that produce a bolt/ray/etc. have somatic components and benefit from a fighter's abilities, while spells that don't have somatic components can be cast while fighting.)


If an increasing number and magnitude of stacking modifiers can apply to a d20 roll at higher levels, then the gap between an optimized character and an unomptimzed one will get wider and wider. Eventually, you'll reach a point where characters who fall above or below of a certain band of optimization are practically guaranteed to succeed or fail, both of which are kinda boring.


5e fixes this with bounded accuracy, at least in theory. Proficiency ranges from +0 (not proficient) to +6 (or, for skills only, +12). Ability modifier cap at +5 with the exceptions being high-level capstones. Spells etc. might add another +1d4 or +5 or so. Getting outside of the range where a d20 roll matters is much much harder than in 3e.
Alternatively, I like the idea of not having anything stack. You could replace all bonuses with dice; players roll applicable bonus dice and add the single highest result to a d20. A character who has bonuses from many sources will more reliably obtain good results, but their ceiling and floor won't change. A character who has a few high bonuses will have a higher ceiling but not necessarily a higher average.


This is probably not be a complete list of pitfalls, or a complete list of ways to avoid each one. It's just what I can think of now.


Though this does raise the question of which class would apply to things not linked to a subsystem, such as basic perception checks.

This gives me an idea - why not have an "adventurer" class? Adventuring wizards have a broader skill set than their ivory-tower peers, and likewise for adventuring druids vs. those who spend their lives tending to their home forest, and so on and so forth. The skills which don't fall under the purview of some other class are general adventuring skills - why not make a class for them?

Obviously this class could end up being very powerful, since it specializes in the thing which the PCs spend all of their time doing. I'd recommend putting a cap on the level in this class that a PC can have, based on total character level, or maybe making PCs level up this class automatically based on character level.

If you do make this class level up automatically for PCs, then it makes sense to also put skills into it that you want every character to be competent at - say for example, attacks with simple weapons (3e terminology), thus ensuring that everyone has a chance to be at least somewhat useful in combat.

[Goobahfish]

I've basically done this before (currently playing a game in this kind of system). I think it is a good idea.

So, some things to look out for:

#1: Level is important. Level makes the game functional for DMs so having a 'max skill ranks' per level is essential. This also means you don't have to have some strange exponential skill point cost.

#2: The split is important. How many martial skills are there? How many magic skills are there? These are important questions. How fine you make these skills will make or break the game.

#3: Flexibility vs Specialisation. So, the natural reaction is: I will take Fireball, Stealth, Longsword, Healing... What is the advantage of taking three kinds of magic? What is the advantage of stacking martial skills? Unless there is some inherent synergy between overlapping skills they won't be chosen and all characters will be mage/fighter/thieves.

#4: Generic checks vs interesting abilities. So I have twenty ranks in magic. Does that let me cast fireball? How many spells do I get? Are they any spells or are they more limited (i.e. Illusion). So I have Illusion magic... what else does that make me good at? What saves/defences/attacks does that improve?


If abilities can't be used in combination, then it becomes severely suboptimal to split resources between them.

#5: As above... what is the answer to this question.

------------

Some potential answers:
I used a fixed level. I also tried to keep the range of inherent modifiers between minimum and maximum level low (I only have 10 levels).

Fire Magic is a skill. Illusion is 3 skills in my system. You get 10 skill points/level. So you can be an Illusionist/Warrior easily enough or just a Shadow/Rogue. I have found this is about the right amount of flexibility.

I have inherent synergy for skills from the same 'theme'. You get mana depending on your total skill ranks, Divine Favour depending on total skill ranks, Stamina based on total Martial skill ranks etc. This kind of means that a Wizard has less HP, a Fighter more and a Fighter-Mage somewhere in between.

Characters gain talents and techniques (read passive and active abilities). These have level requirements, but a Fighter-Mage gets access at the same time as a pure Fighter or pure Mage. Fighters of course have more stamina, mages more mana so a Fighter-Mage spreads their abilities across their skills and have less 'charges' of each.

If you are interested, I can post a link to my wiki for inspiration ^_^

[Maat Mons]

I'll agree that a level-based cap on abilities is important. But I don't think it eliminates the need for non-linear costs. To see why, just look at the 3e skill system. It had linear costing, and a level-based cap. It wound up where all such systems wind up. Everyone noted how many skills they could max, picked that many skills, and kept them maxed. Maybe they'd keep one fewer skill maxed, and spend the remaining points on one rank each in various Knowledge skills, or something. It didn't do anything to make in-between values a viable choice. It was still pretty much "all or nothing" in any given skill. Well, more so "maximum allowed or minimum required," than "all or nothing." But you get the point.

[Goobahfish]

I'll agree that a level-based cap on abilities is important. But I don't think it eliminates the need for non-linear costs. To see why, just look at the 3e skill system. It had linear costing, and a level-based cap. It wound up where all such systems wind up. Everyone noted how many skills they could max, picked that many skills, and kept them maxed. Maybe they'd keep one fewer skill maxed, and spend the remaining points on one rank each in various Knowledge skills, or something. It didn't do anything to make in-between values a viable choice. It was still pretty much "all or nothing" in any given skill. Well, more so "maximum allowed or minimum required," than "all or nothing." But you get the point.

I broadly agree with this. I think the 'natural' state of the game is that players should have 'maxed' skills and that splitting your skills shouldn't be an 'obvious' choice. That said, my players have definitely chosen a few 'extra skills'.

I think it works in my system for a few reasons.

#1: The default DC is 8 (rather than 10 or 11) so most things usually succeed and an extra +1 isn't the be all and end all.
#2: Bonuses are calculated as (Class Level + Skill Ranks)/2. So sacrificing 2 skill points for an extra skill proficiency can be worth it at the margins.
#3: The game caps out at 10. My players level much slower than regular D&D (mostly because they don't really need to level up that much due to the flexibility of the game. Their characters are kind of already their characters at level 1 and aren't waiting until level 7-10 before they mature into butterflies).

So the difference between 12 skill points (the max) and 2 skill points is about... a +5. Most of the 3.5e issues was that the lack of bounded accuracy of any kind led to a 20 pt disparity by level 20.

I think my broad point is that while your criticisms are pretty valid, there are ways to make them less 'terribad'.

Addendum:

Actually, given that my system gives 120 skill points at max level, it is the difference between 10 skills with 12 ranks or 12 skills with 10 ranks which changes your rolls by 1 in exchange for 2 extra 'proficiencies'... As your mana etc are governed by total ranks invested you can spread out a bit without really hampering your character (that said you'll miss out on some of the highest level techniques/talents) but some characters don't need that stuff.

[Herbert_W]

#4: Generic checks vs interesting abilities. So I have twenty ranks in magic. Does that let me cast fireball? How many spells do I get?

This raises an important point. Class features (and feats, etc.) fall into two broad categories:


Do a thing with bigger numbers. This increases a character's performance ceiling, floor, and/or average when attempting something that they can already attempt. The usual effect of this is to increase a character's probability of success. Increasing the ceiling (and floor) can expand the range of things that the character has a chance to do (and can reliably do) - but there's always a bit of system soup in between.
Do a thing. This gives a character a new ability, and whether or not it has the desired effect, it is nonetheless a thing that that character can just do.


Attributes traditionally provide bigger numbers. Reframing classes as being like attributes means that you'll be getting bigger numbers out of them. This may make a game seem simpler from a "number of moving parts" perspective, but it actually makes the game more complicated from a player's perspective because they need to wade through more system soup in order to understand their character's basic abilities.

For example: will my character be OK if they fall off a boat? In order to answer that question, I need to look up the rules for swimming, typical DCs for that check, consequences for failure, and then compare that to my character's swim check modifier. There's work that needs to be done to understand a character's capabilities and that work requires at a minimum some system mastery and at most also knowledge of how your DM likes to set DCs. For a contrasting example: can my character cast web? I could answer that just by looking at my character sheet - and if I can, I'll also know exactly how many times per day I can do it, too.

These examples also show that "bigger numbers" abilities also lack the resource management which "do thing" abilities often have. (You can add resource management back in, such as what the 3e truenamer does with increasing DCs for each "spell" each time it is used, but that means even more system soup.)

This system soup may or may not be a problem depending on your design goals. Personally, I would see it as a problem - it makes onboarding new players more difficult, and onboarding new players is the very first hurdle that a game needs to overcome. If you are designing a game for players who enjoy learning new systems and/or already trust you enough to view it as worthwhile to put in the effort to learn one of your games, then onboarding is less of a problem.

Likewise, the lack of resource management may or may not be a problem. You might already have enough resource management in your game with items, wounds, etc.

So . . . what if system soup and lack of resource management are problems? Well, the solution is to add "do things" back into the game. Goobahfish's game already does this by adding resources such a mana, stamina, and divine favor - so classes aren't just like attributes. They provide both attribute-like benefits and resource-increase ones. You could also let players obtain resources in ways that aren't tied to classes, or make some classes provide resources and attribute-like benefits and others just provide one or the other.

I can easily imagine a game where there are a small number of attribute-like classes, such as warrior/mage/theif/cleric, plus a number of optional templates such as barbarian/knight/sorcerer/warlock/wizard/priest-of-specific-deity/paladin/etc. Templates would provide "do things" abilities, would be largely optional, and would cost XP to acquire. These templates could have prerequisites, either in terms of characters having certain levels in certain classes or requiring characters to spend XP on certain classes going forwards. Players who want a simple character could take one or two templates and then put all of their XP into classes, and players who want a complex character could pile on templates.


If you are interested, I can post a link to my wiki for inspiration ^_^

Please do.


I'll agree that a level-based cap on abilities is important. But I don't think it eliminates the need for non-linear costs.

I'll second what Goobahfish said. Non-linear costs are a good solution to a real problem, but they're not the only solution, so I wouldn't go so far as to say that they're needed.

[Greywander]

Sorry it's taken a while for me to get back to this thread. 10 hour workdays can do that, and this thread got some pretty lengthy responses that would in turn take a bit of time to read through and respond to. In order to try to keep this post from getting too long, I'll respond to some of the general points made instead of to individual people. I'll bold some important words to increase readability as well.

First, it seems there might be some confusion here, so I'll try to make this as clear as possible: this would be for a skill-based system, not a class-based system. There are no classes in a functional sense. Instead, there are attributes which have been given names typically associated with classes. So instead of a Strength attribute, you have a Warrior attribute. Instead of an Intelligence attribute, you have a Mage attribute. But these aren't real classes, and they don't give class features. They might interact with specific subsystems in certain ways, e.g. your Warrior rank might give you more HP or additional attacks or something, which I suppose would be similar to class features, but it wouldn't give you new abilities to use. Instead, you'd spend XP to purchase those abilities separate of any class. For example, Mage ranks wouldn't give you any magic; you'd either purchase a skill for using magic, or purchase an ability that grants access to magic, depending on how magic ends up working.


In a game where the DCs (and monster AC and saves etc.) scale upwards to match the expected capabilities of a character at a given level,
One problem that has plagued RPGs since their inception was the idea that when you level up, all your numbers must go up. This created a problem, however, because low level content would quickly become trivialized. Sure, it was supposed to be easier, but not that much easier. So someone came up with the idea to artificially scale difficulty according to your level. So the same lock now has a higher DC to pick when you're a higher level, and this offsets the massive bonuses you've gotten to picking locks. A consequence of a system like this is that if you aren't actively raising a skill, you actually get worse at that skill as you level up. If the DC to pick a lock has increased, but your skill in lockpicking has not, then it's actually harder to pick the lock now.

This is the problem that Bounded Accuracy was created to solve. A lot of people seem to misunderstand what Bounded Accuracy actually is, but Bounded Accuracy is simply that the difficulty doesn't artificially scale to your level. The same lock has the same DC to pick, no matter what level you're at. As a consequence of this, we can't let player abilities scale too much, or else they'll quickly trivialize that content. But Bounded Accuracy isn't a new idea, and it seems that most skill-based systems already adhere to Bounded Accuracy, and for good reason. If a skill-based system had difficulty scaling, then all it would do is force players to max out a few skills and dump all other skills, as they'd be terrible at any skill they can't max out. Instead, skill-based systems seem to want you to be really good at a few things, and moderately good at a lot of other things, as a hyper-specialized character will quickly fall victim to a situation they're not specialized to handle.

Anyway, all that said, how I envision it working is similar to D&D 5e's ability score + proficiency system. If you have ranks in a class, but not in a skill, it's like making an ability check with a maxed out ability score but no proficiency. If you have ranks in a skill, but not in a class, it's like making an ability check with proficiency but you've dumped that ability score. If you have ranks in both the class and the skill, then it's like having a maxed ability score and proficiency. Hmm, I think I might want to explore different ways of integrating class and skill ranks into a resolution mechanic, as I'd like it to be beneficial to have both, but maybe not quite so overwhelmingly beneficial. The thing is, in D&D at least, the value of a +1 because greater the more of them you have, which is another reason why hyper-specialization is considered optimal, and that's not really how skill-based systems are meant to work.

Classes as skill caps isn't an idea I'd considered, but I may give it some thought. My plan was for classes to simply work like an overly broad skill that stacks with skills, but is also a lot more expensive. If classes instead act as caps, it might be worth considering having classes not grant a bonus to skill checks (granting access to higher rank skills achieves the same purpose, assuming you purchase those skills). I think this could work, but it wasn't really what I had in mind. It's worth noting that skill-based systems don't typically have "levels" per se; your strength as a character is wholly dependent on which skills and special powers/traits you've purchased. That said, classes-as-attributes could act as a stand-in for your level if we did want to include something like level-based caps.

As for escalating costs, this was what I was planning for classes, but I was considering a different idea for skills. Namely, skill trees. You'd start with a set of very broad, basic skills. Instead of ranking up a skill, you either have the skill, or you do not. Once you purchase a skill, it would unlock the next layer of the skill tree, giving you access to more advanced, and more specific, skills. When you make a skill check, you would then chain together skills on the skill tree, getting a +1 bonus for each skill in the chain. For example, you might have one skill chain that looks like Combat -> Melee -> Polearms -> Halberds. This gives a +4 bonus to using a halberd, a +3 bonus to using other polearms, a +2 bonus to other forms of melee combat, and a +1 bonus to combat in general. The deeper you get into a skill tree, the narrower the skills become. This allows us to make every skill have the same cost, while making more advanced skills "more expensive" because of the narrower application.

Skill trees would likely need to be somewhat complex, with potentially hundreds of skills, depending on how many layers we had, but I think it could also serve to simplify things for new players. When you're creating a new character, all those advanced skills are unavailable to you, so you've no need to concern yourself with them. Instead, all you need to worry about are the most basic 20-some odd skills. Once you've settled on which of those skills you want, you can then explore the next layer in those particular skill trees to see which of those more advanced skills are of interest to you. This is essentially breaking the complexity down and hiding it from a player, only showing them the immediately relevant choices they can make.

I can probably solve the "generic skill check" issue by linking each skill to a class. The link is only loose, so which class is used would depend on which subsystem you're interacting with, but most of the time you'd be using the linked class, and if there was ever a doubt as to which class would apply, you'd use the linked class. Perception, for example, might be a Thief skill most of the time, unless you're specifically using it in combat (where it becomes a Warrior skill) or to identify magic (where it becomes a Mage skill), and so on. Some skills might link to more than one class, in which case you could use either one if you weren't sure. For example, Perception might also be a Ranger skill (with the Ranger being associated with the travel and wilderness survival subsystem). If one person has Thief ranks and another person has Ranger ranks, and they both need to make a generic Perception check, then they could use the class they're better with.

Re; how do I cast a fireball? This would depend on how magic is implemented. There might be a Pyromancy skill, which you'd use for any kind of fire manipulation, including throwing fireballs. Fireball might be its own skill (maybe Pyromancy is the root of the skill tree that also includes Fireball). Or, you might have an ability you can purchase with XP that gives access to fireballs, and then you'd use some kind of Spellcasting skill to cast it. Fudge makes a distinction between ranked traits, such as skills, and unranked traits, which it calls "gifts". I prefer the rhyming scheme of "perks, quirks, and irks", denoting traits that are positive, negative, or mixed. Some magic might be a perk, giving you access to the spell which you'd then cast using a spellcasting skill. Some magic might be an irk, giving you access to something neat but with a drawback. For example, you might get super hearing, but also become vulnerable to thunder damage.

I do have some plans for specific magic systems, but I'm not entirely sure how I'd implement them mechanically. I know that geomagy involves drawing magic circles, and sometimes carrying around huge scrolls with pre-drawn circles on them. Is it a skill? Do you roll every time you draw a circle? Can you freestyle effects, or are you restricted to a set list of effects? You might purchase a perk for each individual spell that represents your character learning how to draw that circle with no need for a skill check, and you could also freestyle by making a skill check using the associated skill. I know witchcraft will involve writing up contracts with spirits to get access to their power. Again, is that a skill? Is it a perk (or irk)? If you make a deal with a spirit as a result of roleplay, do you still need to spend XP to purchase it? Can you spend XP to purchase a contract without encountering a spirit during roleplay and making a deal with it? What if a contract is terminated, do you get the XP refunded? There's a lot of questions that will need to be answered, and I don't think I'm quite there yet.

Well, looks like I failed at keeping this post short. But hopefully this explained some of my thoughts a bit better.

[Goobahfish]

Well, looks like I failed at keeping this post short. But hopefully this explained some of my thoughts a bit better.

Ha ha ha, don't worry, this is an interesting enough discussion to make we want to read every word carefully.

So, if I get this right the mechanics go as follows (perhaps mechanics is too strong a word... musings?):

There are 'classes' as attributes. You can invest in one or more classes in variable amounts. There is also a kind of specialisation within these classes so they are kind of skill like. The Combat->Melee->Polearm->Halberd structure is pretty informative of this general idea.

So the main concern I have here with such a system is essentially harkening back to the trained/untrained divide where untrained skills go from being 'sometimes' to 'never ever' from level 1->20.

This works as long as the external DC's don't escalate at the same rate as the player skills OR that the untrained skills escalate at some other rate.

When designing Patera, I spent a long time thinking about this particular mechanic. I had settled on a D20 system but basically wanted the following to be true:

* A trained level 1 vs a trained level 1 should feel similar to a trained level 10 vs a trained level 10.
* A trained level 10 vs a trained level 1 should feel 'unfair'
* A trained level 1 vs a untrained level 10 should be more scary than a trained level 1 vs a untrained level 1.
D&D sort of has this hit-point bloat vibe which makes higher levels sometimes seem like tedious 'extra maths'. I do twice as much damage, but they have twice as many hit points. In Pathfinder, it goes from +5 vs 20 to +35 vs 50... (i.e. why bounded accuracy and slowly growing bonuses is a good idea).

So as long as 'just being a warrior' or even 'being a nothing' isn't that much worse than being a 'halberdier' you should be fine I guess.

The other concern is that you will probably find is that Warrior/Mage/Thief etc aren't mutually exclusive concepts and so there is going to be a huge amount of hand-wavy overlap exceptions involved. Perception is X unless it is Y, unless it is Z. I think this is where attributes/characteristics/stats really shine.

Mike Mearls had an interesting talk where he posited a hypothetical D&D-lite which was 6 characteristics and that was it. Can I do X? Make a Strength check etc etc. If your class system was clear enough that answering 'can I warrior this task' was a clear choice, then I think it could be viable.


This system soup may or may not be a problem depending on your design goals.
Yeah, this was one of my major considerations in designing my game. I've always found games like 5E and Pathfinder great until you want to veer off the beaten track, even a little bit. Moreover, as you pointed out the fact that classes are this mixed bag of abilities and passive bonuses is really problematic from a design perspective. Whenever I start looking at making a character via multiclassing, my 'good intentions' of creating an interesting character are almost always subverted by trying to 'collect passive bonuses and interesting class features' while avoiding 'boring features'. (i.e. the horrible class-building Tetris where none of the pieces ever fit quite right).

The main goals was that once the character is created, playing at the table should be easy so trading prep complexity for game simplicity is a high priority.


Please do.
https://patera.fandom.com/wiki/Patera_Wiki
I'm in the process of writing this up as a coherent PDF so some of the things on the WIKI aren't precisely canon (~97%). I'm running a campaign at the moment and the wiki is the campaign-specific canon (I don't want to pull the rug out from beneath the players where I can avoid it).

If you are interested in running a game using the system please let me know and I can 'help out' with some suggestions etc. I'm trying to move towards a more user-friendly version.

[Greywander]

There are 'classes' as attributes. You can invest in one or more classes in variable amounts.
Yeah, this is generally how skill-based systems work; any feature is available for purchase so long as you have the XP. If you're mostly used to class-based systems like D&D, then this seems like a radical idea. Even 5e's multiclassing seems much more forgiving that past editions. Class based systems prefer to focus your character onto a specific archetype and not deviate from that. Skill based systems are more open, allowing you to assemble your own character archetype from the available skills and traits. With that in mind, classes-as-attributes would absolutely allow you to raise your rank in more than one class at a time, so long as you had the XP for it.

Now, that said, some skill-based systems do lock you out of purchasing certain features unless you meet certain conditions. Vampire: the Masquerade has mutually exclusive disciplines, e.g. you can learn Celerity (super speed) or Temporis (time manipulation), but not both. Your vampire clan also dictates which disciplines you can learn without a teacher. TBH, while I yearn for a skill-based system, there is part of me that likes figuring out interesting builds within the limitations of a class-based system like D&D, so I might try to figure out a way to include that within a skill-based system somehow. This might manifest in something like the magic systems, where each system of magic has inherent limitations and you can mix them together to get more interesting combination. I might try to apply this to other subsystems as well, for example having martial schools that grant specialized fighting techniques or something.


There is also a kind of specialisation within these classes so they are kind of skill like. The Combat->Melee->Polearm->Halberd structure is pretty informative of this general idea.
To clarify, that example was for skill trees. Classes would probably just raise your rank in that one class, while skills would split into specialization trees. So while your skills might go Combat -> Melee -> Polearms -> Halberds, your classes would instead just go Warrior 1 -> Warrior 2 -> Warrior 3, etc.

Now, one downside I can see for this is that the XP cost to raise a class would get pretty prohibitive vs. the cost to buy another skill (even a very narrowly focused skill). Let's say for example that each skill costs 10 XP. Classes are broader than even the most basic skill, so those have to cost more, say 20 XP for rank 1. Let's say the cost doubles every rank, so rank 2 is 40 XP, rank 3 is 80 XP, rank 4 is 160 XP... and this whole time it's always just 10 XP for another skill. Would you want to save up XP to rank up a class instead of picking up a bunch of skills you use sometimes? So that might be something I need to figure out as well.


This works as long as the external DC's don't escalate at the same rate as the player skills OR that the untrained skills escalate at some other rate.
[...]
So as long as 'just being a warrior' or even 'being a nothing' isn't that much worse than being a 'halberdier' you should be fine I guess.
I haven't settled on a core resolution mechanic yet (mostly, I was inquiring about the idea of classes-as-attributes, and less about a specific implementation), but for now I'm imagining something that isn't too different from D&D 5e in terms of the gap between strong and weak characters. Forget about HP for a moment and just think about ability checks. In 5e, a PC without proficiency who dumps a stat will have a -1 to the roll, while a PC with proficiency who maxes a stat will have +11. That's a decent difference, but considering that the core resolution uses a d20 roll, it's actually not impossible for the weak character to beat the strong character (it's merely unlikely, particularly over the course of many rolls).

Generally, someone who's maxed everything out should almost always beat someone with 0 investment in that area, but even a little investment should still pose a credible threat to the maxed out character. One goblin isn't really a danger, but a lot of goblins might be, and a single goblin might be enough to distract you from a more dangerous enemy long enough to make a difference (e.g. allowing an enemy mage to cast a spell when you could have stopped them if not for the goblin). Players want to feel powerful when they've maxed out their stats, but I think instead of letting them trivialize everything, it works better to still have those things pose some challenge, even if it's an easy one. There can be just as much tension in a series of easy rolls as there is in one difficult roll.

As for "just warrior" vs. "just halberdier", I would probably make these equal (e.g. like a D&D character with a maxed ability score but no proficiency vs. one with proficiency who dumped that ability score). Maxing out a class would likely give an identical benefit to maxing out one branch of a skill tree. For example, classes might max out at 5 ranks, with each rank providing a +1 bonus, while skill trees would then have 5 layers, with each skill in a chain providing a +1 bonus, for a total of +10 if you max both. I might decide to have classes affect the core resolution mechanic differently from skills, in which case the benefit might not be 100% equal, but even then I'd probably aim to make them at least comparable.


The other concern is that you will probably find is that Warrior/Mage/Thief etc aren't mutually exclusive concepts and so there is going to be a huge amount of hand-wavy overlap exceptions involved. Perception is X unless it is Y, unless it is Z. I think this is where attributes/characteristics/stats really shine.
True, so this will require a bit of work to get right. Linking classes to subsystems should take care of the bulk of determining which class to use. Then I can go through and add reasonable exceptions, e.g. allowing either Warrior or Thief class to be used for crossbow attacks. For situations not directly related to subsystems, that can be mostly resolved by linking skills to classes, so you can default to that class if you're not sure. (And only the most basic skills need to be linked, as that link would probably propagate down the skill tree.) I could even make sure to link each skill to at least two classes, allowing players to use whichever was higher.

I do wonder how this would end up feeling in play. It's hard to say for sure until you've had a chance to playtest it. I might try it and decide I hate it, or I might end up really loving it. Hmm, I should probably whip up a really loose prototype before I start nailing down the nitty gritty of the system. One of the nice thing about tabletop RPGs is that they can be really simple; three stats (physical, mental, and social) ranging from 1 to 5 using a roll-under d6 system is simple and straightforward. Maybe too simple for what I want to test, but my point is that I don't need to write up a full complex system just to test a mechanic.

[Goobahfish]

As for "just warrior" vs. "just halberdier", I would probably make these equal

Probably a good way to go (to avoid forced specialisation).

On the same note, I should point out that in my game, the way I resolved 'saves' and 'characteristics' checks was that if you had ranks in any Strength skill you could use that for your corresponding untrained checks. The idea is that if you are strong you can probably hit someone with a hammer without too much trouble, you just don't know how to do anything fancy with it like knock them out.

Using classes to this end will likely achieve a similar objective (handling generic stuff). For example, you could have it that when you choose 'halberd' as a skill you just are a warrior by default (i.e. your Warrior ranks are equal to your highest weapon rank). Then you could have 'extra stuff' on top (read spells/manoeuvres) which encourage you to stay within your speciality but don't punish too hard for going outside it. If you just divided it into skill types (rather than classes per se) then you'd probably end up with less confusion too.

That way a thief is really a archer/infiltrator or a soldier/infiltrator depending on the build. The archer/infiltrator might be good at throwing objects to knock open a trap mechanism (archery) whereas the soldier would be more suited to forcing a jammed door open quietly.

It definitely has some potential.

[Thane of Fife]

Are you familiar with Warrior, Rogue & Mage (http://www.stargazergames.eu/warrior-rogue-mage/)? It's a (free), rules-light game that's fairly similar to what you're describing, using the eponymous warrior, rogue, and mage as the three base attributes.

It's almost certainly lighter than what you're thinking about, but it is a fairly well-liked mini-RPG, so I don't think you need to worry about whether the idea of using classes as attributes can work - it clearly can.

[Greywander]

Having given this some more thought, I don't think this would be the best choice for the kind of system I'd want to make. While this would technically be classless, classes-as-attributes is pretty close to what I'd expect a class system to look like in a skill-based system. As such, I think it actually does have similar properties to a class-based system, and some of the same design principles will apply. Skill-based systems are often universal, able to be applicable to a wide variety of modes of play, which is what I'd want for the system I would design. But where class-based systems excel is with a more narrowly focused game, where each class represents a specific archetype within that niche.

For example, consider an RPG designed around doing heists. In such a game, you'd expect to see classes for the face, hacker, demolitions, safe cracker, driver... yeah, the getaway driver would have his own class. And in a narrowly focused system like this, such classes would work very well. But if this system was actually a generic and universal system that could just as easily be used for dungeon crawling, political intrigue, or as a merchant sim, these classes would be way too specific to be useful in most of these other modes of play. In essence, a class-based system is a master of one trade, while a skill-based system is a jack of all trades. You can try to find a middle ground, but you can't play both sides, which is exactly what I was trying to do with classes-as-attributes.

The problem with the proposal in the OP is that I can see adding lots and lots of subsystems. That's something that skill-based systems are actually quite good at; because the basic rules can be used for almost anything, you can create a more complex set of rules to layer on top, not unlike how D&D's combat system is layered on top of the ability check system (attack rolls and saving throws are functionally identical to ability checks). But, as I've said, I wanted to have one class for each subsystem, so as I add more and more subsystems, I'd need to add more and more classes. And those classes start to get pretty niche. A fighter class that relates to combat seems generic enough to be broadly applicable; a warlord class that relates to tactical mass combat and strategic warfare is... well, is that even an RPG anymore? And that's not even that niche. I think what did it was thinking about a sailor class and wondering if it would apply to things like space ships or tank crews.

So in the end, I think it would be a better choice to do almost the same thing, but using true attributes. I do have a set of four that I think would work pretty well.

Vigor/Physique would cover physical fitness, including "macro Dexterity" (e.g. acrobatics). This covers most of what a warrior class would cover, but can be logically applied to non-combat situations.

Precision would cover hand-eye coordination, attention to detail, and "micro Dexterity" (e.g. manual dexterity). This covers most of what a thief class would cover, and excludes social deception, which not all thieves might be good at.

Wisdom/Insight would cover education, intelligence, and practical experience. So it's a perfect fit for a mage, but also applies broadly to a whole host of other situations.

Lastly, Heart/Social covers all your social skills.

Four is a nice number that keeps it simple, and each one has its own clear niche, so it shouldn't come up often that you're not sure which one to use. I think this will work much better, as I can come up with as many subsystems as I like without needing to add new classes to the game. I can also more easily insure that each of these attributes is integrated into each subsystem somehow, though not all to the same degree. Physical fitness is less useful for political intrigue, while a silver tongue won't help as much on the battlefield.


Are you familiar with Warrior, Rogue & Mage (http://www.stargazergames.eu/warrior-rogue-mage/)? It's a (free), rules-light game that's fairly similar to what you're describing, using the eponymous warrior, rogue, and mage as the three base attributes.

It's almost certainly lighter than what you're thinking about, but it is a fairly well-liked mini-RPG, so I don't think you need to worry about whether the idea of using classes as attributes can work - it clearly can.
I took a brief look at this, and it looks interesting. It does seem like it gets a bit fiddly with some parts, though I suspect that's out of necessity, because sometimes the game just plays a lot better if you e.g. add a random +2 to a stat to get some derivative value instead of just using the value as-is. Still, it seems pretty straightforward. I noticed it also uses yards for distance, which I was planning on doing as well.

[Goobahfish]

Yeah, I figured this might be where this thought experiment ended up.

The analogy to the 'Spycraft' game (this is very close to your example btw and was a pretty fun 3e skin) is pretty good. The classes in that game made no sense to multiclass, as in there was sooo little incentive. And it worked because everyone was so niche. So if there was a Fantasy niche equivalent I think this could work pretty well.

[Maat Mons]

Is each ability score intended to be useful in a wide variety of situations? The magic stat, presumably, can do anything. Enchantment for social interactions, invisibility for stealth segments, blowing things up for combat, et cetera. The social stat seems like it would only be useful for, well, social interaction. Physique and precision would have combat applications and some utility, but no use in social interactions. Except maybe using your muscles to intimidate people. Using the concept of pillars you mentioned earlier, what do you intend for the pillars of gameplay to be, and how many of them should any given ability score be applicable to?

Personally, I don't like doubling the magic stat up with non-magic things. I don't like how D&D magic users have to be well-read, insightful, or persuasive, depending on which sort of magic they use. It leads to weird things like having to give Magical Beasts good Charisma scores, so their spell-like abilities have a chance of landing. But Magical Beasts shouldn't necessarily be good at social interactions. So the fact that the system lumps spell-like-abilities and social skills onto the same ability score creates an issue. In a system where intelligence governs all magic, you have a similar issue where anyone born with innate magical ability must be smart, even if their method of harnessing magic doesn't have anything to do with studying.

Do all of the ability scores have both active and passive aspects? Strength can pretty clearly be used both to swing a weapon, and to resist certain effects. Dexterity can clearly be used to shoot an arrow, or dodge an attack. I assume the winsight stat is used to resist mental assault and notice sneaking enemies. Is the social stat used to see through lies, or is that also insight? If the social stat doesn't defend against social manipulation, does it defend against anything?

[Greywander]

So if there was a Fantasy niche equivalent I think this could work pretty well.
Yeah, it's not that the idea itself is bad, it's just bad for what I wanted to do with it. Like I said, it's how I'd expect a class system to look in a skill-based system. If there was something about skill-based systems that you wanted to have, but you also wanted to retain the feel of a class system, this would be a way to achieve that. The problem is that class systems tend to not do well in a universal system, since classes by their nature are defined by a niche. The system needs to be focused around making that niche useful, and a universal system is not focused on anything in particular.


Using the concept of pillars you mentioned earlier, what do you intend for the pillars of gameplay to be, and how many of them should any given ability score be applicable to?
This is a bit of a tough one. What makes sense is to have four main pillars, each with an associated attribute. That attribute isn't the only one that can contribute to that pillar, but it will do most of the heavy lifting. Some of them are self-evident: Physique is associated with combat, and Heart with roleplay. It's the remaining two I'm not so sure about. I think Precision could be filled under "puzzle-solving" (and anything that delves into more minutia, like crafting), but I'm not sure about the Wisdom stat. If we're talking about subsystems, then "magic" is the obvious subsystem, but magic isn't really a mode of play, but rather one of many tools that can be employed during a mode of play.


Personally, I don't like doubling the magic stat up with non-magic things. I don't like how D&D magic users have to be well-read, insightful, or persuasive, depending on which sort of magic they use. It leads to weird things like having to give Magical Beasts good Charisma scores, so their spell-like abilities have a chance of landing. But Magical Beasts shouldn't necessarily be good at social interactions. So the fact that the system lumps spell-like-abilities and social skills onto the same ability score creates an issue. In a system where intelligence governs all magic, you have a similar issue where anyone born with innate magical ability must be smart, even if their method of harnessing magic doesn't have anything to do with studying.
This is an issue in any system that simplifies its attribute system. In the proposed four attribute system above, Physique makes no distinction between a dead lifter and a gymnast. However, attributes are only half the story. If you're want to create a character who is more specialized in one area than another, you can deliberately ignore skills that would make you better at the thing you're "dumping". Remember that "mage" (or "magus/magi", actually) originally just meant "wise man", so it makes sense that they would be well read. If you don't want that, just don't take any skills that relate to that. You'll still get a boost to your base competency, but it will lag behind the things you do pick up skills for.

This will make certain archetypes a bit tougher to execute, such as the absent-minded scholar (wizard, inventor, etc.), but the alternative is to have a longer list of attributes that interact in complex ways. Either way is a trade-off. I suppose I could do something like what I suggested for skills, where each attribute branches off into more specific attributes. That might be able to achieve the best of both worlds, and it would make it consistent with how skills would be handled.


Do all of the ability scores have both active and passive aspects? Strength can pretty clearly be used both to swing a weapon, and to resist certain effects. Dexterity can clearly be used to shoot an arrow, or dodge an attack. I assume the winsight stat is used to resist mental assault and notice sneaking enemies. Is the social stat used to see through lies, or is that also insight? If the social stat doesn't defend against social manipulation, does it defend against anything?
I'm still in the early brainstorming stage, so nothing definite yet. Most likely you would actually use Physique, not Precision, to shoot a bow or dodge attacks*. Precision would be used for shooting guns or crossbows, but regular bows require a lot of strength. I'm unsure if the Perception stat should be Precision or Insight; perhaps similar to D&D 5e I'd make an Investigation skill that uses Precision and a Perception skill that uses Insight. Social stat likely defends against emotional manipulation; even if you know someone is trying to get a rise out of you, it can still work if you can't stop yourself from reacting to them. Mental stat is more about mental manipulation, e.g. logical deception ("Huh, I guess that makes sense"), which isn't the same as emotional manipulation ("Every minute, another child goes hungry, donate now"). I suppose "Will" could be another potential name for the social stat.

A while back I was playing with some ideas for a video game using those four stats. My plan was to have a unified resource pool that acts as health, mana, and stamina, so things like swinging a weapon, casting a spell, or running all take away "health". Physique gives you more of that resource, Precision reduces the resource cost for various actions, Insight lowers the resource floor before you get tired (e.g. you can push yourself closer to exhaustion, which is dangerous, but it's better getting stuck mid-combat unable to swing a sword or run away), and Heart increases resource regeneration speed. Obviously this won't work as well in a tabletop game, but I might be able to find similar ways to incorporate all four attributes into a subsystem.

*Re; dodging:
Here's the thing: dodging is a very physical activity. You'd expect only someone who is very physically fit to be able to dodge attacks very effectively. But it's not just any kind of fitness that will work; a buff weight lifter isn't going to be dodging much, rather, it is the agile gymnast that will be evading attacks. That person needs both agility, to be able to move with speed and contort their body out of the way, as well as endurance, to be able to do this continually over an entire combat. Basically, what I'm saying is that dodging is very athletic. It doesn't make sense that the same stat that makes us better at painting Warhammer minis also makes us better at evading attacks.

But here's the other thing: let's think about character archetypes in fiction. You've got the Big Guy, the Smart Guy, and the Fast Guy. The Big Guy is built like a JoJo character; they're fit, but it's the wrong physique to be dodging. Instead, they block. Or they just tank the hit, because they can. The Smart Guy has the constitution of wet toilet paper, and so specifically uses techniques that allow them to fight from long range, or even remotely. Finding/reaching the Smart Guy is how you beat him. The Fast Guy sometimes also fights from range, e.g. a gunslinger, or sometimes uses lighter melee weapons. He's somewhere between the other two in terms of defense. But how exactly does he defend himself? He isn't tough enough to just tank hits, nor is he strong enough to just block an attack, so all that left is dodging.

My point is, it makes zero sense for Dexterity to help both with fine motor skills (e.g. lockpicking, crossbows) and acrobatics and evasion, as these two are completely unrelated to one another. That said, fictional archetypes just so happen to pair these two ideas together quite frequently. It isn't so much that one helps with the other, but rather that most characters with one also independently acquire the other. "Dexterity" isn't so much a base characteristic as it is an archetype.

The question then is, should I preserve that archetype by pairing those two concepts together in my own system? Or should I relegate those abilities to the attributes that are more "proper" for them? And that, I'm not sure. I suppose one thing I'll concede is that Precision probably would influence reflexes, which are a key component of evasion.