Rob Pardo began his lecture by noting that Blizzard focuses on multiplayer game design first and foremost. Multiplayer games have many more design constraints than single player games, and it's important to get all of these worked out first.
"Once you have a great multiplayer game, you can make a good single player game out of it, but it's very hard to do the reverse," Pardo said. On Blizzard's upcoming Starcraft 2 game, Pardo noted that the single player game was formed only in the last nine months of the game's development.
Pardo then looked at the different requirements of tuning Player vs. Player gameplay and Co-operative gaming. While PvP requires focus on balancing amongst units while effectively differentiating skill sets, co-operative gameplay is built on solid methods of communication and complementary strategies between disparate players and units.
Pardo also noted the importance of tuning non-combat PvP, referencing World of Warcraft's own Auction House system. In this example, players hoped for developers to add a proxy bidding system to the auction system so that the game could automatically bid up to a particular price while they were elsewhere in the world.
While this would have been useful, the design team decided that it was very important for the auction house to be its own gameplay and players make a decision as to what type of player to be, making being a largely city-dwelling player a viable strategy.
Skill differentiation itself can be found in multiple places in games, and it is important to decide early on where that differentiation between new players and expert players exists. "Starcraft is known as a more twitch-oriented RTS game, and that's intentional," Pardo said.
Besides more hardcore spaces for differentiation, like twitch gaming and multitasking, avatar improvement is one that's often dwelt on; however, Pardo isn't certain that this provides very good examples. "Your gear or your items sometimes makes you more powerful, and that's not necessarily a skill."
Game balance itself involves many tradeoffs, and Pardo reminded everyone that developers can't keep every single thing they want for the game. Here, he compared World of Warcraft and Starcraft 2 philosophical differences.
WoW focused on making each character able to reach the same max level and maintaining important roles in raids of all sorts; moreover, competitive group PvP on all scales needed to be balanced and robust. In Starcraft 2, every race is totally different, a marked contrast to World of Warcraft. Very often there would units on one side that had no counterpart in the world.
"The press would always ask for an example, and they didn't get it. That's the way it is designed." Also, Starcraft 2 has a focus on speed and offense; rushing is a big part of the game, despite the complaints that some players had on that strategy.
While math is the foundation of the the gameplay, Pardo noted the importance of the overarching philosophy of the design, rather than the math, in creating the game's fun. For him, everything in the game should feel overpowered, empowering the player and making them not left wanting.
"Don't use the math the balance the game out of mediocrity," Pardo said, mirroring his own exhortations to his designers. "You want every class to feel unbeatable." Along with this, designers should be intimate with the nuances of the gameplay. "Things like unit acceleration and pathfinding algorithms affecting one unit versus another unit... these are things that don't show up in the spreadsheets."
Pardo also strongly discouraged the use of any sort of superweapon. "It's really fun punching the 'I WIN' button, but it's not so fun for the other player." While Starcraft can have some extremely powerful weapons unleashed, their use can always be negated by an aware opponent.
As an example, Pardo presented the Starcraft Ghost Nuke ability which, powerful as it was, still provided many diverse signals and opportunities for the opposing player to negate its effect.
Dealing With The Player
User interface also has a huge role in affecting game balance. Pardo presented Starcraft's old system of selecting a maximum of 12 units; should the player wish to command more than that, he would need to continuously, and quickly, select and command groups of 12 units. This makes navigating the UI and the map itself a part of the gameplay.
The new Starcraft 2 UI will have unlimited selection, and there was a long argument at Blizzard before it was decided to go through with this major change to the system. Subgroups in Warcraft 3 also had a huge impact on the gameplay balance, as allowing you to command groups of units while still having access to your hero made spell casters much more powerful.
In the case of World of Warcraft, UI mods can change the game balance... often enough to have the base gameplay change in the eyes of the designers. "'Mods are bad and they play the game for you' is our philosophy, but outside of that all mods are good." If the designers do see a mod as breaking the gameplay, they go ahead and break the mod in their next update.
Perhaps the most important thing in balancing a multiplayer game is player psychology. Actually being balanced is useless without a perception of being balanced. "You may be able to prove that it's balanced, but if the players don't think it is, it doesn't matter," Pardo said.
Towards that, developers should strive to make losing a more acceptable outcome. Changing the time scale of losing can help greatly, as it is acceptable in other genres like fighting or racing which have shorter windows of gameplay. Also, the ladder and the method of climbing the ladder should be well matched to the game itself; simply putting ELL or Trueskill without tweaks into the game may not be the best fit.
Visual clarity also has a huge impact on player psychology. It's important that a unit, item, or weapon suggest not only its function, but also its ability and quality. This can be easier in a more contemporary style (Pardo used Team Fortress 2 as an excellent example of visual design suggesting function) but should be strived for regardless of overall visual style.
Having this sort of visual clarity can be very hard in an MMO, thanks to user customization of avatars, but even this can be helped by having class-specific armors and items.
Pardo also encouraged developers to not give their players too many choices. With maps, more is often not better. "You want a small number of maps that players can know and learn on." Pardo also noted that having less buckets in matchmaking is also better than more.
With too many match permutations, players can find that there is no one out there to play against. Showing the player that the system is working is perhaps more important than the matchmaking itself. Feedback on the system is critical, even if the player has no more interaction with the system; simply providing a counter can be enough to keep people engaged and believing in the game.
Pardo ended by noting the whole other can of worms that is E-sports. "Decide up front if you want an e-sport game, as there's lots of work to do." Making it a spectator event entails a whole other set of technology, like replays, saves, referee controls, statistics viewing, and map editing.
He encouraged developers to look at other sports that have recently made the jump to spectator sports, such as poker with its hole cam. "Show the player's decision making process. If you can do that in the game, it'll be successful."