Task Allocation Without Instructions

Gordon's Discovery: Interaction Rates

Ants switch tasks based on the rate of interactions with other ants doing those tasks. Local encounter rates encode global demand.

1. An ant performing task A encounters other ants
2. She detects their task by their odor
3. If she frequently encounters ants doing task B, she may switch
4. If she rarely encounters ants doing task A, she may leave

Task Switching Thresholds

Not all ants are equally likely to switch tasks. Some switch easily (low threshold), some are stubborn (high threshold). This variation is crucial for colony stability.

The mix creates stable yet adaptive behavior. If all ants had low thresholds, the colony would oscillate wildly. If all had high thresholds, the colony couldn't adapt.

Implemented in TypeDB: 6 Inference Patterns

We translated these biological mechanisms into TypeDB 3.0 inference patterns — rules that fire automatically, classifying entities, switching states, and allocating tasks without centralized control. Each pattern maps to a mechanism from this chapter.

1. Classification — Chemical signatures: ants identify task type by reading cuticular hydrocarbons. TypeDB functions read multiple attributes and classify entities into tiers.
2. Quality Rules — Response thresholds: ants have different thresholds for switching. Inference rules fire at different thresholds, creating mutually exclusive bands that prevent oscillation.
3. Hypothesis Lifecycle — Probabilistic switching: ants accumulate encounters before committing. Hypotheses accumulate observations until statistical significance triggers a state transition.
4. Task Management — Task allocation: ants find available work through local information. The negation pattern detects tasks with no blockers. Pheromone trails on edges make tasks attractive or repelled based on colony memory.
5. Contribution Tracking — Interaction rates: Gordon measures encounter frequency to understand colony demand. Aggregate functions measure value flow between contributors.
6. Autonomous Goals — Emergent behavior: the colony adapts without instructions. Objectives spawn from frontiers and curiosity signals — no central planner.