Iteration: Can You Perfect It?

Sucking is part of getting better! Since you're never done getting better, you're never done sucking. This week, I look into how I sucked it up last week, and how the process of iteration makes things a LOT better!

---

So, last week I presented an example of something I think would be a solid model for what a "class" is, in a class-based game. In spending the last few weeks looking through how "class" and related concepts have been conceived of in D&D history, I identified a struggle between a class as a primarily broad, mechanical concept (ie: "I use spells, which are daily abilities") and a class as a primarily narrow, narrative concept (ie: "I study arcane lore in dusty old tomes"). These concepts aren't necessarily related to each other at all, and the struggle to unite them has been a tension in D&D since the Greyhawk supplement. I made a stab at combining the strengths of both into a system where every class was a narrow, narrative concept, embedded in a broader, mechanical structure that was flexible and accommodating.

There were some pretty good reactions to it! Iteration -- taking what you've done and identifying issues, refining it, and putting it out again -- is an essential part of any creative process, and a uniquely vital part of the design process. So lets walk through that process here.

Ultimately, there were a few broad categories of reaction, and each of them tells me a little something about the design and the goals.

Looks Good!

I don't see why the GM couldn't use the entire spectrum of all class abilities to design classes

Three (don't judge me) thumbs WAY up ! Like, WAY WAY up. We're talking exospheric levels of agreement.

Tiny classes and a toolbox to build your own sounds just about perfect.
...
I'd like to see something like this as a ruleset, then setting books including the collection of classes you want for that setting.

These responses are worthwhile, and definitely show me that there's something worth exploring here. They're the foundation that tells me I'm not wasting my time on this little rabbit-hole, that there's at least a few people out there who are at least curious about an idea something like this. That's more than enough rope to hang myself with! Iteration is fundamentally a process of breaking something apart, so it's good to remember what people like about what you're doing, too, so that you don't chuck out something valuable as you refine your idea. It's also useful to show you if you're off on the wrong path -- if people didn't "get it" or didn't see the point, you might want to at least explain it in a better way.

One specific reply showed me that this was something it was possible "get":

The example class doesn't have any of the general mechanics listed (Hit Die, saves, etc.) Was this intentional? Do those things exist outside class? That might be okay, come to think of it.

Funfact: last week's article came out of my thinking presented in my article on dissociating class from equipment. What also emerged from that was the idea that mechanical ideas like durability, attack power, and resistance are independent of a class structure. Presumably, this poster never read that older article, but still identified the proximal source from what I didn't say. While the poster's overall response was a little more lukewarm, their brain followed my thinking, which is a pretty good sign.

Isn't This Just A Classless System?

You could take a point buy modell or free traits.

you don't have a class system at all, but a point buy system that offers packages

Of course, in the end, that starts to look a lot like a point-buy classless system, albeit with a bit more structure in it. That's not a bad thing, necessarily, but it might be seen as undercutting the whole emphasis on class that you have.

If classes are so specific that anyone can build a 'class of one' that only they have and has any combination of approved abilities, then you don't really have a class anymore.

The first real "actionable" item seems to lie here. The idea of this system as a system to "roll your own classes" is compared to a system in which there is no class. It's a fair cop, right? If a DM can essentially build their own class, it's also possible for a player to do that, using the same rules.

"Am I a class, or just a collection of points? Or is there much of a difference?"

On the one hand, this is me accomplishing one of the design goals: I wanted to have a system where people COULD invent their own classes, and that is essentially the same idea as a classless system: the classes are not hard-coded into the mechanics of the game. No specific class can be essential to play the game if you want to give DMs the power to make their own classes without re-designing the whole thing. So, yeah, comparing it to a classless system makes sense.

On the other hand, the intent was to keep many of the virtues of having classes in the game. Like the man says:

Classes have advantages and disadvantages.

I wanted, here, to appropriate as many advantages as I could. Generally, my own review of the material suggested that some of the main virtues of classes were:

1. Grokability: Classes are easy to understand and relate to, because they're fundamentally archetypal. You don't need to be told to try and steal things when you play a thief: You know what you have to do, because it is also what you are called.
2. Streamlining Decision-Making: A class includes all the things you want to do, and none of the things you DON'T want to do. If you're playing a thief, you don't need to worry about picking abilities that are thief-y in order to be a successful thief. The class does this for you.
3. Conflict-Solving: Because a class relates to a conflict in the mechanics of the game and in the world, a class tells you what kinds of situations your character excels in and which they're better off taking a back seat in. If you're playing a thief, you don't try to mix it up in a massive melee with trained warriors. You're not there to kill, you're there to steal...the conflict you excel in is in remaining undetected, gaining access, and absconding with the goods successfully. Leave the dirty work to the meatheads.
4. World-Building: Classes are useful as world elements, because they define the kind of world the character lives in. If there are thieves in play, everyone needs to worry, at least a little, about their stuff going missing all of a sudden. There may be guilds of thieves as well as churches of clerics and colleges of wizards. They appear as NPC's, as helpful allies, as unfortunate enemies. They appear as an identifiable archetype in the world.
5. Mechanical Bucketing: Classes are a way to divide up mechanics so that everyone doesn't need to learn all the moving parts. The archetypal D&D wizard may have to deal with resource rationing, spell planning, and advanced decision-making, but doesn't need to learn much about weapon types and armor weight. The typical D&D fighter is probably much more interested in those elements, but doesn't have to bugger about much with rationing and planning: they can react on the fly.

Looking at that (non-comprehensive) list, I can see a few areas where my experiment didn't quite hold up. While the class was understandable, was part of the world, and gave you everything you needed to hold up that character type, it didn't have a clear "conflict" it was solving, and the system itself homogenized mechanics enough so that the bucketing became irrelevant.

So what we're looking at improving would be the mechanical definition of the class, which should improve both of those metrics. The exact thing that I mentioned that was historically in tension with the story definition of the class. As D&D has showed us, there's nothing necessarily inherent in the story idea of a trained warrior that implies that they don't have to ration or plan their ability use (What's up, Martial Dailies?), and there's nothing necessarily inherent in the idea of a thief that implies they roll a die for success (percentile dice, or a skill check).

This can be perhaps addressed with a new moving part. Much like how equipment, HD, and saves weren't a part of last week's class automatically, we can tie in a moving part that, for now, we'll call the Core Mechanic. The core mechanic is interchangeable and customizable: a class comes with one, but it's something you can replace about the class.

As an example, lets use a mechanic that I probably wouldn't actually use these days: % Chance, the original thief mechanic of "roll less than your % to accomplish a given task from the following list." Gaining levels might give you higher success, or more abilities on your list, or both. So, on a thief, you have things like "Hide in Shadows" or "Tell a Lie" or "Escape Bonds" that let you do your Thief-y things. On a fighter, perhaps you separate out weapons, so you have things like "Sword Attack" and "Arrow Attack" and "Deflect Blow" and "Stay Standing." Because we're using the old "% Chance" mechanics, we're not too worried about your enemy's skill -- succeed on your roll, and you do the thing you rolled to do, maybe the DM makes an ad hoc adjustment, and we're done.

Hypothetically, that works. We'll see it in action on the class I presented last week below.

I'm Not Sold On Specificity

Specific I find a truly horrible goal unless you are intending to write a very focussed game

a giant book full of overly specific classes is something I find personally kind of repulsive. It reminds me far too much of my old Palladium days.
...
My preferred take on classes is for a class to be relatively broad in its overall mission statement, but then allow for significant flexibility in the assignment of its particulars.

But let the players play the broad class and define their own unique approaches within it.

A flexible broader class - and a good muliticlass system - lets a player expand their character into the unfolding story...rather than require the two to be in synch from the get go.

Specificity, I knew, was going to be kind of a controversial design principle. D&D has wrestled with specificity over the years in things like kits, prestige classes, and in classes themselves, and it's always been problematic, because a tightly defined character is a character without much inherent flexibility, and the arc of D&D history has been toward flexibility, with things like feats and skills and extra spells and subtypes within types. Going against that flow would rankle some hairs, and I though it would be useful to see where the friction occurred.

I'm confident that specificity, as a principle, achieves the goals of making the class understandable, decisive, and part of the world. It solves the problem inherent in calling a class, say, "Fighter." "Fighter" isn't really an archetype, it isn't a specific kind of character, and it tells you very, very little about the world (okay, there's fights in it?). By those criteria, "Fighter" doesn't cut it as a class. "Rogue" falls into the same boat, and "Wizard" does, to a large degree, too. "Cleric" a little less so, but if you view it through the 2e lens, it can become meaninglessly vague, too. Specificity means that there is a certain kind of character you are choosing to be when you grab a class. It has a direct correlation to something in the story of your own game. When your "rogue" can be anything from a spy to a locksmith to a scout to a backstreet thug to a rakish swashbuckler, it doesn't say much about the kind of character you are to call yourself a "rogue."

My own likely-misplaced confidence aside, it's clear that "you can make your own class from scratch if these are too narrow!" isn't something that these commentators really embraced as a viable solution to the dilemma of specificity. In truth, all the flexibility and adaptability anyone could want is found in that big "every class ability is compatible with every other class ability" rule -- can't get much more flexible than that! But what about that solution might be untenable for some folks?

This problem may be one of perception. In presenting "three traits of a good class," I may have obscured the idea that these good classes are dependent on a good underlying construction system, and hidden some of the adaptability. It was clear to some posters (those mentioning that this made the system nearly classless), but not to others (those mentioning that specificity means you lose significant flexibility -- in this situation, no, you wouldn't, because the system is ALSO a la carte). It may also be an issue with my perception: it's not clear to me how this system would stop people from expanding their character into the story as they go, because you can a la carte abilities to represent only the abilities present in the story, however it develops. They're seeing obstacles I'm not, so one of us isn't talking to the other.

In this situation, it's often useful to make the intent much clearer by example, so that way, at least the problems become more...er...specific. A true example of the system in play might be a little too much for this article (though it's good fodder for the future!), but I can perhaps point at it by including an example of how it might play out.

This Is Not What A "Class" Is
A few posters presented their own ideas for what the traits of a good class are in their minds:

- It does something distinct.
- It does something well.
- It needs to be customisable enough.

So what should classes be in my opinion?

• Bold
• Inspiring/Inspired
• Flexible

So here are the three things that broader classes do right:
1. Accommodate emergent (rather than focused) gameplay.
2. Encourage all players to participate in the various arenas of the game (if well-designed).
3. Provide a common table language (and ease of player character creation/leveling).

This kind of feedback is interesting because it provides me some insight into what people are looking for, and also potential things to highlight in my system. Does this system allow for classes that do something distinct? Something well? Are the characters customizable "enough"? Can my classes be bigger, "cooler", and flexible? Does the system in which their embedded allow for emergent gameplay without over-specialization and a common terminology?

Honestly, there's a few of those areas where this system is pretty weak. This doesn't feel like a systemic problem to me, though. The "Core Mechanic" concept and a focus on the conflict the class solves should help provide something more distinct and strong, while the overall system keeps individual characters from being pigeonholed too tightly and allows for people to literally pick abilities as they go if the pre-set option isn't appealing. Boldness in the mechanics are mostly a matter of establishing the binary swing, something tough to do without, and things like common terminology and whether one is "inspired" to play as the class comes as a matter of individual taste and specific campaign world-building.

Iteration!
And so, that's lead to the version of the class (as an example of what you can make within the system) attached here.

Iteration is a literally endless process, so I don't expect this to be perfect. I just expect it to be better and, where it fails, to do so in a way that shows me where I can improve. That's iteration. You can't be perfect -- no such thing. It's always a journey, never reaching its destination. What do you think of this now? How is it better? How does it still not quite meet the needs? Let me know down in the comments!View attachment 59014

 

[Argyle King]

I know I mention the game a lot, but -serious- take a look at GURPS: Dungeon Fantasy. I think it's a very good example of how you can have something like classes, but still be flexible and allow player choice. While I'd never expect D&D to take a path which was quite that flexible, it did seem as though early 5th Edition playtest packets may have gone a few steps in that direction. I had hoped they did, but the last few seem to have gone a different way. If nothing else, it's interesting to look at how somebody else did it, and compare notes to what D&D has done over the various editions.

http://www.sjgames.com/gurps/books/dungeonfantasy/
http://www.sjgames.com/gurps/books/dungeonfantasymonsters1/


Another game I've played a lot recently (and another one I've been mentioning) is Star Wars: Edge of The Empire. The game has classes, but abilities are gained through talent trees. It occured to me yesterday (while playing) that talent trees weren't too radically different from feat chains. The main difference is that talent trees allow for multiple paths to reach an ability. With the huge amount of feats in D&D, it might be more difficult to pull off multiple paths, but I think it's an interesting idea. What if I could pick up feat Z by one of two ways? Maybe one player progresses through X, Y, and then Z. Maybe a different player progresses through A, B, X, and then Z without needing Y. One of the interesting things about this is that two players can be playing the same class, but have characters who feel very different in play. In theory, D&D allows for this too; unfortunately, in the past two editions, there have been too many choices which were "no-brainer" choices or seen as "the one true way." I believe part of the reason isn't due to the classes per se, but is tied to how D&D level progression works, but that's a different topic. Either way, I view EoTE to be another game which D&D could take some cues from, and it's also a game which handles things far different than GURPS, so the juxtaposition of styles is interesting. Looking at both and seeing the two ways of doing things when it comes to handling, class, characters, and character abilities is enlightening as well as being pretty good food for thought.

http://www.fantasyflightgames.com/edge_minisite.asp?eidm=232


edit: Links provided for references

 

[Shayuri]

I think my main disagreement with you is in what our criteria for a 'good' class is. For example, I don't see a particular sort of conflict as being a necessary part of class design. Nor do I think a class needs to tell us anything in particular about the setting other than it's a setting where the class exists.

I prefer to move those things farther down, to where choices about the character are made. Less a question of class, than of selecting specific class features, feats, and so on. In my mind, character creation is usually a matter of moving from broad generalizations to specific manifestations.

A class is therefore a very broad, general thing...second only to race. It's later on, as the character becomes specific, that the things you're talking about take place.

Having said that, I suspect that you and I aren't as far apart on this as I'm making it sound. Our main differences may just be semantic, since a lot of what I'm talking about involves selecting optional class features...which is to say, picking things off of a 'menu' of choices available to anyone of a given class.

That's not so far from what you're talking about, except that it's done on a character-by-character basis, rather than generating an entire class unto itself that consists of those specific options. And since there's nothing in what you're saying that forbids it from being done on a character-by-character basis...yeah. Largely semantics.

 

[Argyle King]

I think my main disagreement with you is in what our criteria for a 'good' class is. For example, I don't see a particular sort of conflict as being a necessary part of class design. Nor do I think a class needs to tell us anything in particular about the setting other than it's a setting where the class exists.
.

I don't think you were responding to me, but you said something which caught my attention. I believe the same thing. Often, one of the problems I have with D&D is that it assumes too much about how I handle a problem. I think a lot of the classes, feats, and options which get derided as being "bad" or "underpowered" get that reputation not because the idea is bad, but because the idea is bad in the context of the system. If I may steal some of the popular language being used to describe Next, I'd say I believe those options were designed with pillars in mind which the editions of their time didn't support very well.

To be fair, I can't really blame D&D. It took me a while to realize it, but, in many cases, I was trying to use the particular editions of D&D in a way they weren't really intended (or I at least believe they weren't intended) to be used. Each edition that I'm familiar with (which includes mostly 3rd and 4th) has some ability to stretch to other styles, but, overall, I found that a lot of problems I was having were due to the fact that I was trying to do things which often fell outside of the game's expectations. That is something to consider when it comes to class design as well. If you want a class to be designed in such a way to highlight certain features, it helps if that class exists within a game which supports the highlighting of those features.

 

[I'm A Banana]

That's not so far from what you're talking about, except that it's done on a character-by-character basis, rather than generating an entire class unto itself that consists of those specific options. And since there's nothing in what you're saying that forbids it from being done on a character-by-character basis...yeah. Largely semantics

I think I'd say that if we start in different places, with different approaches and different needs, but end up being able to use the same system, then I've pretty much hit the goal.

 

[Jhaelen]

I think my main disagreement with you is in what our criteria for a 'good' class is. For example, I don't see a particular sort of conflict as being a necessary part of class design. Nor do I think a class needs to tell us anything in particular about the setting other than it's a setting where the class exists.

I'd like to give an example to illustrate these criteria:
In the DSA rpg, there are hundreds of classes. The reason why there are so many is that the designers really tried to allow players to play absolutely every individual that might have a place in the game's setting. What they didn't do though, is to think about what a pc of any of these classes is supposed to do or contribute in the context of a typical adventure set in their game world. A party consisting of a baker, a shepard, a catamite, and a pearl-diver may have trouble dealing with almost any kind of conflict that might be expected to come up in an adventure. So, are these good classes?

Neglecting the setting when designing classes works better. It's what D&D's been doing at least since 2e: The core rules define a (small) set of (broad) class concepts that can work in a multitude of (somewhat similar) settings. Then, in setting-specific supplements they show how to use and modify these concepts to work well within the setting. This can be done by introducing new feats, alternate class features, prestige classes, etc.
There are some obvious advantages to such an approach.

Imho, the question is, if I know from the beginning what kind of setting I'm going to play in, wouldn't it then make more sense to incorporate everything required for them to fit-in, right from the start?
Many rpg systems are doing this if the system is meant to be used only for one particular setting, e.g. FFG's 'Star Wars RPG' or 'The One Ring' for Middle-Earth based games. FFG even decided to split their RPG system into three 'sub-systems' to account for different campaign types played in the same setting.
Again there are some obvious advantages to such an approach.

I'm of the opinion you have to decide for one or the other approach. D&D hasn't always taken a well-defined path there. AD&D 1e had a bunch of classes that carried the baggage of belonging to a very specific cultural group that didn't necessarily have to present in the setting that was used: Assassin, Monk, and Druid are all examples that didn't work well as 'generic' classes, especially because of the way they worked mechanically: all of them implied the existence of a certain kind of organization.

 

[I'm A Banana]

party consisting of a baker, a shepard, a catamite, and a pearl-diver may have trouble dealing with almost any kind of conflict that might be expected to come up in an adventure. So, are these good classes?

Yeah, that's exactly the kind of thing that paying attention to the "conflict" helps to avoid, I think. Just saying "You can play a baker!" or "You can play a cultist of the Crystal Dragon!" doesn't necessarily give you something useful to do when you play the game.

I'm of the opinion you have to decide for one or the other approach. D&D hasn't always taken a well-defined path there.

What do you think of the proposed idea of having very specific classes embedded in a highly adaptable system, like the proposal? The "organic growth" example shows that it'd be trivially easy for a setting to introduce a new Core Mechanic (for instance) that changes how a given class works. You can imagine Dark Sun offering things like Defiling, Preserving, Loyalty, and Psionics, for instance. Eberron might offer Dragonmarks or Artifice. Those could be Core Mechanics and/or actual classes, so that if you have a Templar class that gets its own unique abilities and has a Loyalty Core Mechanic, you could either take that core mechanic and go to a different class (to perhaps represent Templars of different regions, or different kinds of Templars), and you can also take the templar class and swap out the Core Mechanic (perhaps the Templars of Gulg take the Spirit Binding core mechanic instead!).

I think it might help give people obvious things to change that have a big impact on how the game plays, at a level independent of class -- it essentially defines how you get the resources you're spending on things.

 

[Neonchameleon]

That was a vast improvement on the last one Looking at that character pathway I might want to play it for a few sessions, depending on the DM and pitch. But there's still plenty of room for improvement even at a fundamental level. Notably:

I'm still unhappy with very little player choice in levelling up - and looking at how you say multiclassing works, anyone who found 4e classes too samey is probably going to have at least as much problem with that class structure - and some of us who liked the 4e class structure are going to have significant problems. If you can multiclass the way you say without cherry picking being ridiculously easy and effective, you're basically back into the realms of point buy rather than class based with multiple pools for points (see WoD for an illustration).

And asboth DM and player I don't want the default assumption to be that the DM creates the classes for multiple reasons. Firstly that I want to be able to play the game out of the box and don't want the faff. It's overhead I don't need, and I'm very much an improv DM - and believe in general that lowering the barriers to entry for DMs is a good thing in general. Secondly as a player I seldom like playing a pregen (and that's ultimately what these optionless classes amount to in my eyes - pregens). It's my character - for the DM to cross that line is for them to cross into storytelling rather than DMing territory. Thirdly as a DM I want to see what the players come up with. I want to give them guidelines as to the world, then I want them to add their own richness to it. I certainly won't have fleshed out every nook and corner. Worlds are complicated, chaotic, and confusing, and more heads working on them makes them feel more alive.

 

[I'm A Banana]

And asboth DM and player I don't want the default assumption to be that the DM creates the classes for multiple reasons.

I think I've successfully made the system so that you don't have to have that assumption. The "organic growth" example shows a player basically deciding to crib an ability from these guys. Speaking functionally, the abilities are free-floating. If a group would prefer to just do whatever, they can open up the selection so that you can pick any character ability, meaning each player can customize their OWN vision of what their specific character looks like, from literally every ability and core mechanic and background and basic ability set in the game (this would make the Compendium or something like it VERY valuable for these groups! )

But if another table is a more traditional top-down kind of table, they've got printed classes to choose from, too, so that they don't HAVE to make a million choices to play a certain type of character (thus leveraging the big benefits of a class-based system). And when new classes gets printed, groups that go more open still get a brand new pile of abilities, backgrounds, basic abilities, and core mechanics that they can re-assemble as they desire.

 

[Jhaelen]

What do you think of the proposed idea of having very specific classes embedded in a highly adaptable system, like the proposal?
[...]
I think it might help give people obvious things to change that have a big impact on how the game plays, at a level independent of class -- it essentially defines how you get the resources you're spending on things.

I think that would work for me, but I'd really like to see it in action. I'm still wondering how much say a player should get when picking bits to create a class.
When I started my 3e campaign I met with each player individually before the first session and asked them what they were interested in playing, then we basically built the character in tandem, with me making suggestions and explaining options.

Maybe it would be possible to see the class only as a starting point. It often happens during game play, that a player discovers she'd like take her character in a different direction than she initially thought. In such a case, the system shouldn't put any obstacles in her way. So, there should either be a way to 'retrain' or 'multiclass'. Perhaps classes could even be designed in smaller increments. Assuming the system uses the concept of levels, you could initially restrict yourself to define class abilities only for the first few levels. Then, after reaching the last level of that increment, if the player wants to continue on that path, you design the next increment, etc.
Writing this, I suppose this would be similar to D&D's BECMI approach... hmm.