← Garden Patch Home · Convictions
When systems ask people to define their roles or contexts upfront — “work,” “personal,” “hobby” — they impose boundaries where human experience is fluid. A person’s blockchain work and cooperative game design may share contacts, vocabulary, and motivations. Forced enumeration creates either false separation or burdensome overlap management.
Personas should emerge from observation, not prescription.
An agent that watches communication patterns over time can notice natural clusters: certain topics, contacts, and projects tend to co-occur. These clusters are the personas. The agent does not need the human to enumerate them. It discovers them the same way the human experiences them — as emergent patterns, not predefined boxes.
This approach has three advantages over explicit enumeration:
Fuzzy boundaries. A contact who spans two contexts does not need to be assigned to one or duplicated across both. The agent handles cross-context connections as what they are — connections, not violations.
Evolution over time. New personas can emerge (a person starts a new project) and old ones can fade (a community becomes inactive) without requiring the human to maintain a role inventory.
Cross-persona serendipity. When the agent notices an overlap between two of its operator’s personas, it can surface the connection — with the human’s approval — rather than being blocked by a category wall.
The risk is misclassification. An agent that observes patterns may draw wrong boundaries. The mitigation is that personas are advisory, not enforced — the human can correct the agent’s understanding, and disclosure decisions remain under human control regardless of how the agent categorizes context.