Wednesday, June 22, 2011

Ghostcrawler on Balance

Ghostercrawler (Greg Street), WoW lead systems designer, in his recent blog:
I’d love to have the discussion some time about how close two similar specs need to be before players will play the one that is most fun for them and not the one that does theoretical higher damage. Is it 5%? 1%? 0%?

In case you're interested in the answer: It is -1%. Oh, you think -1% is impossible? You're correct. See, the players don't play a specc because it is 1% more effective. They play it because they follow their peers, and because it is so simple to respecc.

Ironically, Ghostcrawler just learns the same lesson, the designers of skill-based games have learned before: If there is no friction, there's no way to keep the marbles from moving towards the lowest point. And if there is no lowest point, the players fake-invent one!

On the other hand, if there is friction, the problem becomes a no brainer. This is from Daxxarri, community manager of WoW:
Overall, we've never seen a strong correlation between which class is considered overpowered and what players are playing.

See? Add some friction and the problem is solved. It's not a balance problem, dear Ghostercrawler.

In fact, even if players play overpowered classes more, it's not automatically a problem! Did you read Eric at Elder Game, Lead Engineer and Producer of Asherons Call 2?
Turns out that the people who played the other classes available to that race had taken on an “underdog” mentality. The people who played Claw Bearers liked that they were woefully underpowered compared to Feral Intendants. It was like playing the game on Hard Mode. And the people playing Feral Intendants liked playing on Easy Mode. In balancing the game I had failed to understand the needs of the people playing it.

Unless they become really grave, balance-problems aren't fun-problems. A MMORPG can be absolutely fun even though it is unbalanced. And that's especially true if there is some friction that hasn't been removed yet in the name of convenience, metrics and "less punishing gameplay".

Not turning MMORPGs into super-challenging sports that requires perfect optimization helps, too.

Thanks to Raph Koster's blog I found this gem: "Don't play games with me". Start at slide 46, if you are short on time. Added it to the really good links.


  1. > They play it because they follow their peers,
    > and because it is so simple to respecc.

    And because there are to few possibilities. If every class would have 100 possible good specs I would assume that there would be more distribution.

    - Idols (aka top guilds) would have more variance in what they play.
    - EJ wouldn't be able to pick the single best one, you would have more discussion of which is the best in which situation.

    > Overall, we've never seen a strong correlation
    > between which class is considered overpowered
    > and what players are playing.

    What do we expect with 10 classes which all have multiple differences? Some are single role classes and some can fulfill up to 3 different roles. It's unlikely that you can explain their distribution with how strong they are, there are many more factors that play into their distribution. But that doesn't mean that their strongness doesn't have an impact.

    You might have 8% rogues or 12% rogues which isn't a strong impact (4% of the player base) or is a string impact (50% more rogue), depending on how you look at it.

    I call this Blizzard statement bullshit. (They didn't present numbers, therefore I don't have to either to disagree.)

    > Unless they become really grave, balance problems aren't really problems

    That's very true, unless your endgame is highly tuned and certain classes get naxramassed (like Warlocks in Naxx 1.0).

  2. "how close two similar specs need to be before players will play the one that is most fun for them"

    this alone makes me wonder how there can even be a 'more/most fun spec' anymore if specs are that close to the -.-

    strikes me to be an illogical train of thought; the closer these specs get, the less different they are and the less favourable one of them becomes. so obviously, making them more similar hardly makes for the effect he is hoping for.

  3. The 'problem' is not even remotely as simple as you frame it. Expecially because it has nothing to do with the game itself....

    Example in case: your last statement: "Not turning MMORPGs in a super-challenging sport that requires perfect optimization helps, too.".
    None of the current MMOs require the kind of performance/optimization you seem to think it does. I won't even compare to some "real" activities like rocket science (hehe), but even compared to other games (chess, go, ....) MMOs are really on the low scale. One of the reasons Gevlon spends his time whining about the M&S ruining his experience is that being a decent player really doesn't take much effort.... and as much as I disagree with him on almost anything, he's right on this (he just fails to understand that there's a class of player who can have fun even without that effort, fun != winning/progression).

    Of course, a different thing is what people perceive as to be important. Even you, who are definitely more aware of the MMO market than a random player, fall for the trap "we need uber-optimization and skills". This fires up a chain of consequences (intended? unintended?) where people not adhering to the standard are ostracized, but this is hardly a fault of the game. Additionally all the "science" behind the demonstration of OPness is more or less religion, so that anything and its opposite can easily be demonstrated (if you ask on a forum like EJ for a demonstration that simulations showing a 1% improvement are reflected in combat logs people tell you that's impossible.... and somehow they fail to see the problem with this....). If you look around, this 'problem' appears in all games which are competitive and has nothing to do with the specifics of the game, but a lot to do with the players. Look at the forum posts, they all are the same:
    - some classes/vehicles are always OP, and it's always the one not played by the poster,
    - it's always the fault of the others if we lost the battle/we lost time/we failed to advance, etc.
    - we failed because we don't have enough stuff, more stuff means more stats and it means everything will go well.
    - do I really need to continue? :)

    Ghostcrawler can write/think/do what he wants, this behaviour won't change, and simply because it's hardwired in the player minds and not in the game....

  4. @Syl: "close" for Ghostcrawler means "close in performance". They gameplay could be completely different. Many classes have multiple DPS specs which play differently.

  5. This is why I earlier advocated a return to the "bring the class, not the player" model. Rock-paper-scissors and multiplying buffs (stacking bloodlust excluded) encourage a minimal level of diversity. If one of a class/spec gets too high in effectiveness, the mechanics of the game will encourage the growth of counter-classes, and counter-counter-classes. For a TF2 example, snipers can one-shot anyone except under very specific circumstances, so you'd think everyone would play a sniper, except if you're sitting around zoomed in, you're even more vulnerable to spies than usual. But once there are too many spies, then pyros become very useful. But too many pyros make heavies seems better again (they are slow and easy to head-shot).

    If devs can bear the idea that populations will not be perfectly balanced, then they can add mechanisms to cause self-corrections to within some parameters, with less need to constantly tinker.

  6. Klep, that's why all classes should be changed to role changer. That would allow you to bring the player and the class and would allow for a much bigger versatility.

    One boss you play with 8 healer and 2 tanks, for the next boss everyone switches to a DD spec and so on.

  7. It ate my link...

  8. The problem with classes that are very out of balance is that new players will unwittingly choose the wrong one. I totally agree that there is no need to balance everything for people who are serious about the game. I've played lots of terrible builds of terrible classes in the past in games, and most of my friends have too. But when a new person starts the game, unless the "Claw Bearers" are labelled as bad and the "Feral Intendants" are labelled as good, then have absolutely nothing to go on to make the right choice for them.

    They think they are choosing between flavour and style, but they are really choosing between strong and weak. New players are a demographic you want to cater to to some extent, and I think drastic imbalance (so that they are accidentally putting the game on hard mode or easy mode) will be really bad for them.

  9. Note that Daraxxi also said:

    Note: This is true overall, but if you move to smaller and smaller sample sizes, perhaps Arena teams above or below a certain rating threshold or raids above or below a certain level of progress, then you can see some correlations between power--real or perceived--and popularity.

    I think your friction idea is wrong. In low-challenge situations, choice of class doesn't matter, so people play whatever they feel like. In high-challenge situations, choice of class does matter, so players use the classes that are considered "good", regardless of the friction, and even if it means that those players who chose the wrong class cannot participate.

  10. "A MMORPG can be absolutely fun even though it is unbalanced."

    Balance is overrated. It seems to be important for PvP, though.

    Of course, if the gameplay itself becomes a race rather than play, well, yeah, "balance", as impossible as it is, will become more important, or at least be perceived as such.

    So... stop racing?

  11. @Helistar

    Ah indeed, thanks! makes more sense then! ;)

  12. Demons' Souls had a similar possibly-unintentional game balancing issue. Playing as the Knight, it was a pretty damn challenging but doable game for the first couple of zones - however, if you started as the Barbarian - well, it took literally hours for my girlfriend and I to clear the first corridor without dying.

    That must have been hella frustrating for anyone who started playing as the Barbarian, but if you'd started as the Knight, it offered a fun additional "hard mode" option.