-
Notifications
You must be signed in to change notification settings - Fork 1
Derivation
All entities have their own cannonical history which defines them. Entities will sometimes need to consult that history to respond to events in that context. For example, a well-travelled bridge will eventually break when its load exceeds its current capacity, which is a function of its history.
Properties are the accumulation of history. Properties are also a declared aspect of an entity's concerns. Entities have properties which matter to them in a story-relevant way.
To make decisions, an agent will need to query its own world model to determine how best to further its own goals. This means finding out what its goals are, what could advance them, the risk/reward of such actions and so on. All of these things require that the agent examine its own cannon. This page explains how the agent asks questions and how the cannon answers them.
A question is a pattern matching problem. Example questions:
- What are my goals?
- What are my projects?
- How might I advance X project?
- What risks does step Y pose?
- How much do I care about risk Z?
A question will evoke zero or more answers. The answers will have varying confidence.
Motive: positive or negative
- a condition - matches a hypothetical context to some degree
- incentive - degree to which the agent values this condition
- dis-incentive - contrary-wise
Events the entity experiences will necessarily modify their motives.
Initialization of entities includes pre-populating survival motives like "breath" and "eat".