Joe Osborn, Mike Sennott

No Comments »

No comments yet.

RSS feed for comments on this post. TrackBack URL

Leave a comment

You must be logged in to post a comment.


Acknowledging Player-Generated Stories

Filed under: Ensemble,Vikingr — Tags: , @ 21:26

The specific social, combat, and economic mechanics at the core of Vikingr were all selected for their narratively interesting and universal nature as well as their appearance in source materials such as the Icelandic family sagas. Rather than merely adapt, say, Njálssaga to a video game, I wanted to capture the rules and systems that were behind all of the sagas of the Icelanders. My process involves going through the sagas, identifying patterns that recur across a variety of stories, and formalizing them in such a way that a computer could identify them in gameplay terms.

Initially, I thought that teaching Vikingr to comprehend these half-mechanical, half-narrative concepts was a necessary step on the way to a holy grail: the generation of new story content based on player-character backgrounds and histories. After my first round of prototypes, I have realized that the key thing is not generating new randomized content from templates and Bayesian belief networks—it’s simply recognizing and acknowledging the narrative freight with which players load their characters and play sessions. Some of this can be specified a priori (players can customize a Viking’s appearance, or assert that a particular family member had an earlier career as a brewer), and some can be easily inferred based on a player’s habits (a Viking who never wears a shield can be labeled with the title “The Reckless”; two Vikings that always travel together can be hit with performance penalties when separated).

Quests and achievements are both powerful tools for maintaining player engagement. I suspect that this is because the former express narrative events in terms of game events and the latter transform game events into tellable narrative accomplishments. Vikingr represents two main advances beyond these mechanisms: first, its equivalent of “achievements” are reintegrated into the game design, providing concrete player outcomes (either mechanical or aesthetic) besides sitting in a list in a menu somewhere; second, the game can examine a player’s activity and automatically provide a relevant “quest”-like structure that is in keeping or in competition with the inferred goals of the player. This would benefit from an example.

Let’s say that game-year after game-year, the player does not go on any Viking raids and instead farms and hunts year-round. The game might notice this and perform the following actions:

  1. First, acknowledge his intentions by calling him a “farmer” and giving him a boost to production and perhaps a deficit to combat skills;
  2. Second, encourage him to diversify by exerting pressure on these food supplies: freeloading relatives could come by and demand hospitality, for example. Refusing to provide it could lead to a new title: “stingy”.

To sum up:

  1. Ensemble permits the storage of complex relationships and data at an arbitrary level of detail and specificity;
  2. Vikingr employs the data from (1) to calculate and administer “achievements” based on game-inferred and player-specified goals, recognizing the stories players carry in their heads (all this according to an extensible suite of designer-provided rules);
  3. Vikingr may then trigger simple quests to encourage or discourage player behaviors based on (2). In a sense, this represents a move from a “quest tree” to a “quest forest”, with context-appropriate objectives for players’ personal goals (and an emphasis on socially-motivated quests rather than pre-authored narrative dumps).

Vikingr does not attack problems of text or story generation. Instead, it primes players to interpret possibly-unrelated game events as related ones by recognizing just enough of their intention and input to suggest that connections are the norm rather than the exception. It should provide experiences that vary widely between players, since content is recombined based on player activity decoupled from game progress. The real “content” of this game consists of rules for recognizing player intentions, player-specifiable goals, and triggered quests, narrative, and game events.