My next book will be about learning at scale, both machine learning and organizational learning. That means I need a model of learning that works for people, machines, aliens, and spiritual beings, should they suddenly appear and start disagreeing. Subject to change, currently I've broken disagreements and learning down into three areas:
- What exactly are we trying to do?
- What tool do we need to do that and monitor that it's being done?
- How do we build that tool?
Example Activities and Conversations
- Pure math
- Computer Program
- Political/Religious Speech Delivery
- Historical Relationships
- Historical Narratives
- AI Restaurant Recommendation
- Legal Proceeding
- Startup Growth
- Marketing Plan
Disagreements, Conversations, and Activities Evolve Over Time
Identifying and managing these changes form an important part of learning and problem-solving.
- Rise of Stalin
- Development of a hard science
- Creation of a number system or calculus
- Political argument
- Public health policy
- Simple tool-making
- (proven) Economics
- Riot control
- Modeling complex systems
- Protein folding discovery and mfg
- Creating recommendation models
- Speech writing
- The more created things involved, the farther right you go
- The more observed things involved, the farther left you go
- The more complex, complicated, and potentially unclear, the higher up
- The more rote, simple, predictable, and mechanical, the lower down
This takes us one step closer to Pivot Questions, a necessary part of modeling learning at scale. It categorizes the kinds of arguments, conversations, and other activities that drive any kind of learning.