Game Designers in Detail: Anna Kipnis

Game Designers In Detail is an interview series between a current NYU Game Center MFA student and a PRACTICE speaker.  PRACTICE is the NYU Game Center’s annual conference that takes a close look at the concrete challenges of game design, creating a context for conversation among game designers of all types.

Eric Teo: From your Indie Tech Talk 2014 at NYU, its fascinating to learn of the various dialog engines that Double Fine uses. At which point in a game production life cycle should designers start dialog design and what are the different dialog design types that designers can consider for their games?

Anna Kipnis: I personally believe that dialog in your game is equally important as content like animation or visual effects, and could even be more important than those, if your game relies on a narrative unfolding.  If the dialog in your game is something interactive (dialog trees, player input-reliant interfaces, etc), then it’s a gameplay mechanic in its own right and should be thought about as early as possible — during the game’s pre-production as part of the overall design.  It could be that you’ll need to do some R&D if you’re trying to create an original dialog interface, and something you would need to iterate on quite early in the game’s production to get it right.  The process of getting dialog into a game is surprisingly time-consuming, so even if it’s something that is not the focus of your game, leaving plenty of time for it during production is important.  In general, I believe dialog should be treated as if it was a gameplay mechanic — with content systems on board very early on, represented in the game as early as possible, with scratch assets if need be, and honed throughout development to make it feel deeply woven into the game.

As for what designs you could have for dialog, aside from the established approaches, it is an area of design that is wide open for innovation.  Some of the things that have been done to much success in the past have included:

  • Linear, interstitial story-telling or cutscenes triggered by success/failure (e.g. the light story development you’d see in a fighting game between fights, or generally, games that use brief cutscenes to introduce story context or motivations, but never appears during the interactive portion of the game)
  • Dialog that is triggered based on a direct interaction from the player, which could be primed to react differently based on player progress (e.g. coming up to an RPG character and hitting the interact button, or interacting with an object in the world that plays back a recording or a log of something that happened before)
  • Dialog that is triggered based on an indirect interaction from the player, for instance, when an NPC discovers a game state change at a later time (games like the Sims, where the AI discover objects that the player has placed in the world over time and responds in Simlish)
  • Dialog trees that provide the player with choices for how to respond and could potentially have gameplay gating effects (as in adventure games, where you could receive an item that helps you progress). The game Kentucky Route Zero also used dialog trees as a way to let the player choose the back story for some of the events, creating a rich experience, while not necessarily having impact on the game’s internal state.
  • Dialog trees or an otherwise underlying system that uses game verbs to affect a branching narrative (TellTale games like The Walking Dead, Quantic Dream games like Heavy Rain, the recent Her Story, and Wheels or Aurelia)
  • Dialog trees that have simulation consequences (as in games like Mass Effect, Dragon Age, or KOTOR, where the choices you make change the underlying simulation that informs the game what direction you’d like the narrative or the player stats to take)
  • Procedurally-driven dialog that generates unique sentences on the fly, used to accentuate the believability of a character that is constantly speaking (e.g. Boyd from Psychonauts, who rants non-stop about conspiracy theories)
  • AI-simulation dialog that serves it up to the player based on prior player decisions, an underlying simulation, randomization, or some game state (e.g. Prom Week, the Versu engine games, like Blood and Laurels)
  • A pervasive narration that responds to the player’s actions in the game (e.g. in Stanley Parable, where the narrator comments on the in-game actions the player is taking)
  • Dialog that is served in a non-linear manner, based on some game state, randomization, or even AI agency, usually in the form of “events” to which the player may often have to respond with a choice, and which may impact the underlying game state or simulation (e.g. games like Massive Chalice, King of Dragon Pass, Crusader Kings II, Princess Maker 2, 7 Grand Steps, which thrive on giving the player a unique experience each time they play through the game)

These can all be text-only or fully voiced, with real or fake language. One catch is that the more dynamic the dialog, the more difficult it becomes to have fully-voiced characters. Voiced dialog introduces a lot of complexity to the game as is, be it a new requirement to support many, many lines of dialog that take up a lot of space, the possible necessity to include subtitles, adding lipsync to your characters, or recording distinct voices when your game has many characters. It can also create surprising problems in contrast to text-only dialog.  The reason for this is that humans are very good at hearing glitches in audio (when two lines are not smoothly played back to back, for example, or when one interrupts the other in an awkward way, or when a line repeats too often, or a lack of variety in sound, and so on).  These effects may not even be noticeable to you until you replace your text-only dialog with voiced.  With dynamic dialog, you may want to seamlessly combine lines of content, maybe even with player input, which is relatively easy to do with text, but can be harrowing with recorded lines.  One approach has been to lean on text-to-speech implementations (for instance, in the game Animal Crossing).  Unfortunately, those systems are as of yet no replacement for talented voice actors who can bring a lot of emotional heft to a line delivery.

Even with the amount of experimentation and progress there has been with dialog systems over the years, the field is still quite young.  My feeling is that we will see a lot more games with dynamic or emergent narrative, capturing the experience of having a conversation with someone, or otherwise feeling more fully immersed in the game world.  I very much admire all the developers making new strides in this area and really look forward to playing their games!