Elder Game: Classes vs. Open Skill Systems

A few years ago, I was interviewing for a systems designer gig. The design interviewer had been at Turbine many years prior, so we vaguely knew each other from there, but she needed to know how much I’d grown as a game designer since then.

She asked:

“When you came to Turbine, how did you feel about classless systems like Asheron’s Call 1 has?”

My answer: “I loved them and wanted to make all games that way.”

“Do you still feel that way?” My answer: “No, I wouldn’t inflict that upon a game that I wanted to see succeed in the long run.”

This was the correct answer, if you’re keeping track at home, and I got the job.

AC1′s system of free-form skill purchases was a lot of fun at first and there’s no denying it. But my experience showed that they had serious problems, including:

- Lack of diversity: there is usually one or two perceived “ideal” setups and most people choose them. This makes it quite difficult to create grouping scenarios with divergent roles.

- Inability to predict power levels: creating fun content is already quite difficult. Adding in the notion that people of the same “level” might be wildly different in their ability levels makes it even harder.

- Difficulty adding new verbs: it is tricky to add new open-ended skills to a system without sending everybody into a tizzy.

- Difficulty balancing: the lack of granularity makes it very hard to fix overpowered skill combinations. The already-hard problem of balancing becomes much harder.

- Poor expectation management: it is quite hard to teach players what they will end up doing later in the game.

and many more.

I hesitate to say this because the general designer-blogosphere will likely disagree with me. I’m not trying to pick a fight; I’m just telling you all how exceptionally wrong you are!

Now before you get all frothy, it’s true that each limitation of open-skill systems can be worked around. Templates help with expectation management. Balancing content requires some clever mechanisms, but it can be done. You can use skill dependencies to avoid overly cheaty builds. To encourage diversity, you can heavy-handedly add “role skills” into your mix, with some gating mechanism (such as skill-point cost). And on and on.

And it seems like many beginning designers weigh the options and go “Yeah… I can make this work!” And you can. Sort of. But you’re forgetting two things:

You can’t perfectly balance an MMO. There are a million times more players than there are of you. So players instantly find the loopholes, and you spend a vast amount of time working against them. This is still true with classes, but when players are “locked in” to their class, they can’t go too far wrong. Yes, it’s a designing with a safety net. And yes, it’s necessary, in order to keep balance issues from damaging your game’s lifespan.

(And yes, balance issues can accelerate the ruin of a game. Champions Online’s mess was… well, a mess. And it’s still pretty messy last I checked.)

Open-ended skills sure sound more fun on paper, though, don’t they? For a large number of gamers it seems like the perfect system. Who could argue that you need to take the cop-out approach? Anybody who is entrusted with designing a multi-million-dollar game, that’s who.

Free-form skill choice is much harder than classes. That’s what it boils down to. You can massage free-form skill choices in a million ways, but you’re doing extra work just to get the usability levels that simple classes have built in. Thus, there’s less time to design other features. And even worse, it means there’s less time to balance your game — yet your game needs even more balancing than a class-based game does.

The even bigger cost, the deal-breaking cost, is over the lifespan of the game. The live team will have far fewer resources than the production team, and you’re saddling them with an extremely fragile beast. If your live game desperately needs new game systems, it really really sucks for the designers to instead have to focus on enormous player-hemorrhaging balance issues. And trust me, your balance issues will be enormous at first, and they will be a lot more damaging than classes ever are.

What I Am Saying Is Not Obvious. This is a case where theory and reality have a wide gulf. That’s one reason why this is so controversial. If you design systems on paper, you are quite likely to decide that the difficulty between classes and open-ended systems is pretty much a wash. That’s not reality, and that’s why this was such a good interview question.

It’s a trick question in some ways, because if you’re a reasonable sort of designer, you don’t often go around declaring game mechanics to be unworkable. You waffle. You know that if this is true, then that is true. If this is the case, then that might work. But here, the waffling would have shown a lack of experience. Sure, there are scenarios when a free-form system is the right choice. But that would be a pretty contrived scenario compared to the vast number of MMOs where it’s the wrong choice.

The Exception?

Would I ever make a game with free-form skills? Oh sure. Indie ones, ones where I knew that failure was a possibility and that was okay. I would also assume the game has these qualities:

- largely solo-oriented,

- a strong PvP focus (or some other “generic content”, as opposed to tons of hand-crafted quests)

- any group content is designed for a “gang of very similar bad-asses”, not a large group with diverse abilities and roles.

Depth Illusions

Those qualities I listed happen describe EVE and Asheron’s Call, by the way. They are both successful games, given the history of MMOs. But the thing you may not realize is that both games vastly under-delivered on their promise. At first it seemed like there would be a million crazy character builds, but it turns out, nope: just a few. It’s all smoke and mirrors, and to add insult to injury, you have to use an add-on or a web-page to remember the “right” way to level your character, because the facade of free-form-ness actually gets in the way more than it helps. And trust me, the developers spent a LOT of time working around the balance limitations of their skill system. That means updates come a lot slower than they otherwise could.

There are lots of other possibilities besides strongly typed classes and very generic open skill lists, of course. So there’s a continuum here. But it’s not a linear one. Moving just an inch towards “open skill system” brings massive amounts of extra work. And annoyingly, any compromise will create rabid angry fanboys who claim you’ve failed… before they even try it.

Asheron’s Call 2 was an innovative system of skill trees mixed with gating skills that led to particular roles. Ultimately it inherited the major flaws of both skill trees AND classes, and was quite expensive to maintain, yet it didn’t get “credit” for how open it was, because… drumroll… “it had classes”.

Guild Wars has very open-ended classes that took a ton of time to balance, and it shows in how the rest of the game was rather underdeveloped at launch. But despite a masterful class design by very experienced designers, they fell into the same trap as everyone else does: there is just one “right” way to play most classes, or maybe two; the vast number of theoretically-viable options are ignored as inferior.

Was it really wise to put that much effort into all these unused skills? Just to let people think the system was “open ended” for a short time? Maybe so — that’s a defensible argument, anyway, one which the developers of Magic: The Gathering have in fact explicitly defended. But if you accept this premise, you need to know what you’re doing here: you’re intentionally creating an illusion of complexity that will wear off in six months, tops. If it takes you twelve months to make it, you’re betting a LOT of resources on that illusion. For a card game, okay. For an MMO, you’re losing a lot of crucial polish time on a transitory feature.

The worst part about Guild Wars’ story is that their design was clearly a resource trade off, trading polish for a larger-than-normal number of viable character builds… but they still get “points off” from irrational fanboys who think it would be way better without those stupid constraints. Frustrating. (You can’t really make those people happy because they don’t really know what they’re talking about. Don’t try.)

Really! Free-Form Is Harder Than It Looks

I admit this isn’t quite an open-and-shut case. But too many designers (especially armchair designers) think you can just write this item on your MMO’s feature list. “Oh we’re gonna have a classless skill system, and crafting, and auto-leveled PvP fighting, and…”

"Vaaaaporwaaaarrrrrre! And if you do ship the game, very depressing retention!"

Whenever I see this in a high-level design, I hear a little sound in my head. It sounds like a raven cawing “vaaaaporwaaarrrre“. It is a very sad sound.

It’s VASTLY more expensive, in the long run, to create and maintain a free-form system.

Of course, in theory you can make it work and still have time left over for everything else. In theory you might also be the 1 person every 25 million years who ends up being Batman. Why not play the odds?

You know what, if you’re a master RPG designer I’m sure you can do this. It’s true that no MMO has done this particularly well — their “classless” systems tend to collapse into just a few viable templates, which are no better than classes — and it’s true that games that focus on the openness of their skill system suffer from a lack of polish in other areas. But the history of MMOs is still young.

Just do me a favor and don’t try this as your first MMO design. I’m pointing out a trap here to help you avoid it.