Here’s part two of my notes on Rob Pardo’s GDC talk. Read the first part here.
Game Balance - Define Your Criteria
For each game, the aspects that factor into game balance are different. For WoW, the requirements for each character class are that every character class can solo to max level, have an important role in a group, have an important role in a raid, be competitive in group PvP (it was not a goal to have every class beat every other in a duel) and be fun!
For Starcraft 2 each race is totally different. Every time they add a unit to a race they intentionally do not add that to the other races. They did a press demo where they showed off the Protoss Mothership and everyone kept asking “What’s the Terran equivalent unit? What’s the zerg unit?” But there isn’t an equivalent, all the trees are unique. Better players can win games fast - some novice players complain about the early Zerg rush, but that’s why matchmaking is so important. It sucks for the advanced player to have to spend 20 minutes in the game to win versus a much inferior opponent. This point was really interesting to me because I worked on real time strategy games for many years and we did spend a lot of time talking about how to eliminate the early game rush as a viable strategy. Blizzard explicitly seeks to maintain that strategy because they feel it’s important that players can get better and there be a large skill differentation. Offense over defense - they want players to attack. If you’re playing a version of Simcity in Starcraft 2, focusing on building a pretty city, you’re going to lose. Every unit has a counter - maybe not a direct counter but something that can be used against it. And again, the game has to be fun.
Math vs Fun Math is the foundation for all the game balancing you do. Rob talked about his design philosophy being that everything should feel overpowered - don’t use the math to balance the game into mediocrity. If you have a huge tank it should be UNBEATABLE! except that there’s a counter to take it down. The same applies to the classes in WoW. Now, this is a nice thing to say but I don’t know that this is really how they make decisions in WoW. One of the things we’ve been seeing is that priests are no longer the best end-game healers. Druids have usurped their position, and pallies have plate and other benefits that make them better endgame healers as well. We heard that Blizzard was looking into making the priests a better option, but it has sounded like they were going to nerf druids and pallies rather than improving priests. Who knows though, that’s just more forum complaining perhaps.
Must understand nuances - if you’re balancing it you have to be able to play it. The designer working on balance can’t just watch a replay or read a spreadsheet. You need to play the game to understand how your pathfinding algorithms are affecting one unit vs another. There is a lot of information that is not going to show up in a spreadsheet, only when playing another person. Balance for all skill levels. He gave an example of the shaman power Bloodlust from WoW. Bloodlust was working great in solo play but in 3v3 arena play it was overpowered and it’s tricky to affect just the 3v3 situation.
No super weapons! - if other person teched up to build something and there’s nothing the other player could have done to stop it, they’re going to get upset. Starcraft nuke could have been considered a super weapon, but you had to scout out with a ghost and then target and then the opponent has several seconds to react. The opponent has multiple chance to react before the nuke comes in. He can see the missile silo, can detect ghost sneaking in, or gets the nuke detected message and can find red dot and stop the ghost.
Use your beta - you have a very changing game and you don’t have a lot time to make changes. For an RTS, you might be in beta for 3 months, and you’ll want to get as many balancing passes in during that time as possible. He talked about how important it is to get feedback from the players during the beta. It’s much easier to de-power things that are overpowered than to get them to start playing things that are useless - if it’s underpowered, overbuff it so they’ll use it and you’ll get feedback and can tone it down.
UI Affects Balance
This was a really interesting point. Many convenience features that people have added to RTS games have actually made those games easier to play. He cited many examples.
Unlimited selection - Lots of people complained that it seems silly you can only select 12 units at a time in starcraft. He said it was a long drawn out argument on whether they should switch to unlimited selection. This goes back to skill differentiation. If you can do unlimited selection you can select 100 zerglings and they’re much more powerful now since you can order them all at once instead of having to select many groups individually and assign them orders. It makes it much easier to concentrate firepower with hydralisks when you can order all of them at once to attack the same target. Multiple building selection - In Warcraft 2, the player had to select each individually and then queue them to build individually. They added the ability to select multiple buildings at the same time and this makes it much easier to queue up units with less time from the player.
Sub Groups - Before Warcraft 3, you’d only get buttons that they all selected units shared. With Warcraft 3 you could still get spells from the primary unit, so you could have lots of units selected and still cast spells. This made spell casting much easier and therefore more powerful.
Click to Move - During the WoW beta they decided to try a click to move feature. You could make it play a little more like Diablo - right-click and your character will automatically run to that point. When the feature first went in, it went in wrong such that if you clicked on an enemy, the player character would automatically turn to face them, run towards them, then attack and if they ran away your character would keep following. This completely removed skill differentiation. Blink didn’t matter, if they hid behind a tree it didn’t matter, the character would just follow. Blizzard quickly pulled this out.
UI Mods Blizzard is constantly trying to evaluate what mods are out there and should they be balancing the game for the modders or the non-modders. If a mod comes out that really breaks game balance they sometimes purposely put something in to break the mods. They have simple guidelines for the most part: anything that turns your character into a bot they’ll try to break. Outside of that, all mods are good
[My notes are really long, so I'm going to leave some of these things as lists instead of filling in more detail so as to get this posted in a kinda timely fashion and to avoid boring you with too much detail]
Game Balance - It Never Ends
- Change is always bad to players
the change itself will be bad even though it’s good in the long run
- Maintain your game to keep it relevant - if you want a multiplayer game that lasts for years you need to be committed to supporting it for years. They are still patching Warcraft 3 - 4+ years old, still maintaining it from a balance perspective
- Ban the cheaters
if you don’t take an active stance towards the cheaters, if you don’t find them and stop them they will destroy your game
- Plan patches, but leave time to be reactive
at any given moment some strategy will be uncovered that you must fix
- Don’t panic!
a strategy that seems overpowered, given enough time players will find a counterstrategy
if you change things all the time whenever someone finds a winning strategy, players will begin to use you as a crutch, “I don’t need to become a better player, learn a new strategy, blizzard will fix it for me”
give it a little time to develop
Player Psychology
- Perception of fairness
you may have best balanced spreadsheets in the world, you may be able to mathematically prove your game is balanced
but if you the players don’t believe it is…
arguing on the forums is not the answer. they’re going to post their opinions and they’re going to spread
[His image to accompany this slide was from Penny Arcade, Shaman Player Characters no longer take damage]
- Players hate losing
frustrating to play for 30 min-an hour and then lose. play again and lose again.
try to combat that so even the losing side gets something. get a mark of honor even for losing.
- How do they climb the ladder?
starcraft uses chess system, top players don’t actually play the game because losing a match costs them a lot, while winning gives very little
other methods don’t find best player but more fun
Incentives drive behavior
- original vision of AV was an epic huge battle zone including player-controlled bases, NPC support, quests
- insufficent incentive to engage in several smaller skirmishes resulted in DPS race to kill the opposing general
instead of meeting in the middle of the battlefield and fighting they would just run on by
“good luck storming my castle, i’ll have fun storming yours!”
- what we meant to do: give even the losing side some kind of reward for fighting an epic battle
knew it would be 30-45 min match and wanted the losing side to even get some honor
- what ended up happening: AFKaving, players log in and grab a sandwich and come back to get honor
More examples
- In beta you had a message when you got inspected. Inspect message was creepy. Players wanted them to get rid of the feature, they just got rid of the message
- The rest system - it’s not a penalty - it’s a bonus!
originally implemented it the other way. you started getting a full experience and it got halfed if you played too long. People felt like they were being punished for playing a lot. they reversed the explanation and it’s the exact same system and now people love it.
- Warcraft 3 “thumbs” - could thumbs up or down any maps you wanted. all thumbs would turn into votes when you got matched with someone. the problem was that people’s expectation was they thumbs down they won’t see it, but it’s a vote. Changed the system so you get some limited number of VETOs, and that is not going to get picked. Went with player perception rather than trying to ram their design down people’s throats
Visualization - Visual Clarity
- Does a unit,gun,weapon suggest its function and power
- special effects can cause confusion in battle.
too many fx going off means you really can’t tell what’s going on
- time differentiation
alliance v horde
team color
- avoid hidden modifiers
had to have hidden modifiers to balance the rts but that makes it harder to play the game
Visualization - examples
- team fortress 2 - who runs faster, who has more hp?
[burly dude versus thin dude]
- harder in sci fi
[starcraft 2, huge thing versus zergling]
- very hard to do in an mmo - players can customize themselves
good reason for class based armor, art of armor set suggests the class
show same model in warlock vs priest gear
Home Stretch - Maps
- Random vs. pre-made
both have pros and cons. they go pre-made approach, more control over the experience.
player perception is always “the map generator screwed me, i have worse starting position”
- number of maps - more not better!
want a small number that players can learn to play on. starcraft about 8 per template.
players won’t ever learn the maps with too many
- black shroud vs dark fog of war
starcraft veterans didn’t want to play on new maps because they didn’t know terrain so opponent would get bonus. Warcraft 3 had fog to show all resources and basic terrain so players wanted to explore new maps and not be afraid of them
- Size does matter
want it as small as is reasonable for the number of players, otherwise it takes too long to get into combat
Home Stretch - Matchmaking
one of most difficult things to get right if you want to have a successful mp game
- less buckets is better
developers always want to give players power user feature - what map, how will i play with, etc etc.
but with all the different options, there’s no one to play with since everyone is in their own bucket
that’s why randomized maps - less matchmaking buckets
if not enough people are in a bucket, the template dies
- don’t give players too many choices
- feedback on system is critical. players want to know if it’s working
bg queues give feedback so they know its working
lfg system doesn’t give feedback so no one uses it
- momentum can work against you. if players think the system is broken then it is.
Decide upfront if you want an e-sport game
outside of the game design choices, a lot of what’s fun is making it a spectator activity
need to make sure you ahve replays so people can share and learn by watching
- replays
- spectator mode
poker got popular when they came up with the poker hand cam. really showed the drama
and helped illustrate the process of the players. added spectator to the game
- referee controls
- web support
w3 site tracks all sorts of data/stats. build a community around the game
- map editor, modding ability
- easy way to transfer maps back and forth
Ok, that’s all folks. If anyone wants me to elaborate on some of those bulleted lists, let me know and I can go back and fill in some more words and stuff.
