Wednesday, 21 September 2016

Game Balance - Key Points

I think I've explored this more thoroughly elsewhere (I think in one of the game design series) but recently been playing a few PC games where game balance is an issue has revived my interest in this topic.

Basically, most game designers "nerf" (or weaken) overperforming ("OP") units and tries to "buff" (improve) underperforming units.  The aim is to have units that can perform to similar levels but have different abilities.  I.e. the holy grail is a game that is asymetrical yet balanced; there are different units for different playstyles, and each unit can be unique and can be strong in particular areas, but none is inherently more powerful in the long run, in a range of situations over a long series of games.

I think I've also done articles on why true balance is inherently impossible, unless units ARE bland carbon copies of one another, but I'd like to focus in on the balancing method, with applies to wargames as well.  Or perhaps, three very common mistakes.

Common Balancing Mistakes
Namely, the tendency to
(#1) nerf/buff several things at once
(#2) nerfing/buffing things dramatically (the sledgehammer approach)
(#3) try to be clever by nerfing/buffing things indirectly, rather than using trying to deal directly with the issue. You know, Occam's Razor and all that....

Changing/Balancing Multiple Factors
The problem with (#1) is buffing/nerfing multiple things at once violates the scientific method of testing - i.e. "change only one variable at a time."    Changing multiple things can result in both unintended effects, contribute to (#2) over nerfing/buffing, and finally, worst of all - you have no idea which change caused the end effect.

Say you have an overpowered unit, which stats and game experience suggest is overperforming.  So you nerf its weapon range, AND rate of fire, AND reduce its armour.  Not only does this often result in (#2) "overbuff"/"overnerf" due to multiple buffs/nerfs often having a multiplicative effect on each other...   ....but perhaps after this "balance"  the unit is now underperforming - but there is a problem. You have no idea which was the significant factor.  Was it weapon range that mattered? Or was it the rate of fire? Or was it the weakening of its defence?  You can only guess - because you changed them all at once.  It wasn't a "fair test."

Making Radical Changes (i.e Sledgehammer Nerf/Buff)
These seldom work out well.  Changing the gun range by 10%, maybe even 20% - means you are adjustingly slightly. Fine tuning, as it were.  Doubling or halving gun range (changes in the 50-100% bracket or more) tend to suggest to me that designer had no clue as to game balance in the first place and is swinging wildly trying to fix a radical imbalance. Changing several factors (see #1 above) at once (say 20% rate of fire, 20% weapon range, 20% extra damage) can have the same effect; even if the individual changes are minor, they often add or even multiply one another to create a massive power spike (or drop, if the changes are negative) - for example in the above example the lethality per round would be increased ~40% (20% damage + 20% more rate of fire) as well as having the "kill zone" extended 20%.   Cumulatively, it's fair to say the buffed weapon is 60% more powerful than the original.

While I'm mentioning WoWs, the Youtube channels of iChase and Flamu are useful sources of information on improving your play...

The Indirect Buff/Nerf
This one will take a bit longer as it's a bit harder to explain.  It's a change that does not deal with the problem directly but indirectly; perhaps buffing a unit by nerfing its counter, rather than directly improving the unit's stats.  The problem with this as it often causes unintended consequences and often results in more sweeping changes to overall balance rather than specific and direct balancing.

I'm going to use World of Warships as an example here, as I've been playing it a bit lately.
Basically, there is a problem at high-tier games, whereas battleships "camp" at far sides of the map, parked stationary and bow-on, and just "snipe" at each other all game at ~20km ranges. No one plays cruisers, as they can get insta-deleted by battleship salvoes.  This is a problem as the game was designed for BB > CL > DD > BB in a kinda paper-scissors-rock.  Destroyers (without cruisers to counter them) launched walls of torps, further encouraging battleships to camp at the back.  In lower tiers, the gameplay is far more dynamic, and all ships have a role. Why is this?

Well, the primary difference between Tier 5 and Tier 10 is battleship gun range, and battleship gun accuracy, relative to both other classes and speed/survivability. There's a big difference between a ~15-20km effective range and a ~10-12km effective range - the ships are still moving ~25 knots so the move:shoot ratio is moved dramatically (50%+) in favour of shooting effectiveness vs maneuver. In contrast, cruiser and destroyer gun effective ranges are more similar between tiers - maybe a couple of km.  The second is the "cost" of damage; if your ship is sunk you "lose" in-game currency and actually go backwards in your game progress.   So we have a circumstance where players are (a) excessively punished for taking damage/dying (b) have the ability to shoot across the map and harm each other.  Due to the excessive penalty, battleships are scared to advance or maneuver tactically where they will attract fire; and instead focus on presenting a minimal target. Due to the gun range, they can still contribute to the fight (albeit at limited effectiveness, but the risk/reward tradeoff is worth it). The range/accuracy boost is a big deal as maps stay the same size; therefore at higher tiers you can take focussed fire from multiple enemies all across the map.  At low tiers, with short gun ranges, you only take fire from the few enemies in your immediate vicinity.

So.... what's the core issue?
Players are shooting at long ranges, camping, and afraid to move to objectives for fear of sinking.  Gameplay is static and boring.   Cruisers are too easily killed by battleships. Fewer people played cruisers, thus making the class cruisers were meant to counter (destroyers) more attractive.

As a wargamer I would say "the dice roll chance to hit/kill is too high, and the move:shoot ratio is skewed in favour of shooting."

The simple solution:
Reduce the maximum battleship firing ranges, and reduce the long range shooting accuracy so it is more proportionate to other classes/the relative speed of the ships, as per the better balanced tier 5.
Further, reduce the "punishment" cost for sinking so players are willing to advance and maneuver and "take risks" to get into positions optimal for winning a battle, not just optimal positions for staying alive.

The roundabout solution:
The WoWs devs have tried to address the problems in a range of ways.
They tried to make cruisers more attractive by giving them radar (thus increasing their ability to counter stealthy destroyers) - effectively nerfing destroyers.  They further nerfed destroyers by reducing torpedo range (and making torpedoes easier to spot and dodge).  Unsurprisingly this did not fix the problems.  Despite the nerfs to their natural counter, battleships did not suddenly boldly advance into the middle of the map for close-range duels, and cruisers did not become more popular.  Because the designers did not fix the core problem directly.

Their latest idea - reduce battleship bow armour so they will advance and not sit parked bow on at extreme range.  I'm not sure how removing battleship armour will make them braver and more willing to advance and risk their ships, but hey...

When balancing a game, don't fall into the common traps. Make sure you
(a) change only one variable at a time
(b) make minor adjustments to stats and percentages (+/- 10-20%, not doubling, halving etc)
(c) identify the issue and deal with it as directly as possible; don't try to be clever with indirect nerfs and buffs that effect the unit through secondhand means (i.e. "flow on" effect).


  1. I have never played World of Warships but it seems to be me that increasing the size of the map as ranges increase forcing ships to maneuver and find each other might address much of the problem. I am almost certainly missing something here but that would be my approach.

    1. It would, only there are other victory/time conditions which prevent the maps being drastically resized.

      The movement:shooting range/lethality ratio is the main issue. The other is the penalty for damage; even if they did maneuver to find each other, once they did they would stand off and trade at long range...

      Point being, the devs did everything EXCEPT address the issue directly, but went about it in a "roundabout" fashion, creating new problems and inbalances along the way....

    2. Game designers do seem to have "needlessly complicated" as there default setting.

    3. So true. It's so easy to add things in - and so hard to take things out.

      Hence the pursuit of "elegance" in rules - simple, consistent rules with deep gameplay.

      I view it as resolution vs effect - is there a way to simplify the resolution (dice, modifiers, steps of rules etc) process to get the same end effect?

    4. Everyone likes writing more rules :)
      I've certainly fallen into the "No, this rule has to be vital to the game" only to learn that a gaming group never realized it was there and have been playing happily without it.

    5. The easiest fix is to add Aircraft Carriers! That'll solve the Battleship problem... ;)

  2. My solution with all "World of X" is: Stay at Tier 5.

    In the low tiers most vehicles are affordable (money&time). Above that you have to spend a lot of money or grind forever.

    A had some of my best games at Tier I-III.

    1. Very true. Mid tiers (5-7) are usually best balanced too. However due to recent MM change, WoWs has best tiers at 4 and 7. WoT is still 5-6ish.

  3. It's worth noting that games are systems, and that designers tend to ignore knock-on effects of changing one thing at a time. Sometimes, once you've charted your game's effective game-tree, you realize that it will work so long as you change several things at once so that the pay-offs are in relative balance.

    1. I'd say that most anyone ignores that lesson :-)

      The classic example is when everyone house-ruled the AD&D roleplaying rules to let all races play all classes and without a level cap.

      Then, everyone complained that nobody wanted to play a human character any longer :)

    2. True. In practice, it often makes the balancing more convoluted as it's hard to pinpoint what caused any "new" issues that invariably arise.

      I'd think a PC game company with literally millions of games worth of meta data recorded shouldn't mess up balance badly (compared to an indie wargame designer with a handful of playtesters and limited test time)...

      ...I remember being startled to find out how some of the biggest players like GW playtested (or rather, didn't) their rules - perhaps players care more fevrently about balance than the companies?

    3. I know GW at least tends to focus more on "story" than strict balance. Or well, they used to a decade ago, I doubt that changed.

      With PC games, you can't get away with that though, since that many eyes will quickly figure out the optimal meta and 2 weeks later, everyone is using the one good tactic.

  4. Or, maybe GW is right...

    1. I don't see the GW analogy. "Really Bad Chess" is actually random in a way that is equal to BOTH players.

      It does not allow one player to consistently tilt the deck in his favour. On the contrary, it actually "balances" the game by assigning more valuable pieces to less capable players.

      Bad Chess would indeed be bad if the best player could always choose the best pieces.

      Imagine if 40K assigned you armies based on a random roll for units, from a common pool of units, with army points based on your win % as a player. "Really Bad 40K"? Actually sounds more fun and "fair" (as long as it occurs equally to both sides) that min-maxing army builds.

      To me, this article was one in favour of controlled randomness being superior to cold clinical formality. It's that games need some elements of chance. Games with chance are more forgiving to less capable players, and can engage them and give them hope against better players.

      A better title for the article would be "randomness and unfairness is fun, as long as it applies fairly to both players"

      Balance is NOT always necessary, or the most fun. But most wargames have an element of chance built in already.

      A little off topic: if you tout your game as "competitive" then it cannot rely too much on chance. It's like a competitive dice rolling competition where we throw ten dice and the one with the most 6s is champion... It's why e-sports with games like WoT (where bullets rely on RNG) are not in the same category as Unreal (where bullets go exactly where you point them and do identical damage every time).