Happy new year. Blogging begins again here. Things are afoot.
In light of my recent thoughts on developing a new approach to dialogue systems (or, more accurately, thanks to a recommendation from the designer I'm working with on a nursing simulation side project) I was approached recently to produce a short interactive dialogue demo for a middleware firm based on some of those ideas.
Their core system, though, is a traditional dialogue tree, and it got me to thinking - if we're limited to the usual tools and can't factor in any procedural elements then what practical guidelines can we follow to avoid the usual problems around motivation, depth of simulation and reward? Here are some thoughts:
- Keep character and player motivation consistent. This is the single biggest one. If the player is factoring in knowledge his character doesn't have (eg xp rewards or that all paths are valid) then the decision presented (eg talk your way in or scale the wall) isn't legitimate. Naturally this requires careful balancing with subsystems like xp and relationships.
- If asking the player to choose between expressing his personality and pursuing personal gain make it clear what the outcomes are. We've all played countless dialogue-tree-driven games, each with its own subtle differences. Does pissing off my companions affect their skill in combat? Can I talk my way into losing a quest? Whether in the form of a guide at the start or signposting within, there needs to be a careful balance between exposing the systems that float the game (thereby forming a motivation gap between player and character) and preventing the player from feeling like he's been cheated when the game behaves differently than expected.
- Signpost opportunities for manipulation. Encourage the player to read the other character and select intelligent responses separate to other decision making.
- Keep the conversation moving forward and on-topic. Don't waste text on unnecessary dead ends, encourage every decision to be a meaningful one.
- Don't use a morality system. It's guaranteed to conflict with many players' own ideas.
- Use stats sparingly. Don't rely on skill checks to replace the challenge of choosing a particular approach, but do use them to reflect social standing and character knowledge.
- If all else fails, don't use subsystems. Subsystems like xp, relationship or personality sliders can be invaluable tools in simulating conversation, but they also threaten to separate player from character, rendering decisions unsatisfying. As long as this line only affects the next line a lot of the issues (and complexities) dissolve.