Checkpoints

Completion needs clear criteria.

From project management

Checkpoints are:

• Quantifiable markers of progress
• Zero-duration events marking phase completion
• Basis for evaluation and corrective action

Zociety's checkpoints

Why checkpoints matter

"Specify completion criteria. Measure completion criteria."

Without checkpoints, "done" is subjective. With checkpoints, "done" is verifiable.

Finite States

Zociety is a state machine.

The model

A finite state machine:

• Can be in exactly one state at a time
• Transitions based on events
• Runs to completion before next transition

Zociety's states

[EMPTY] --first agent--> [GENESIS] --thresholds met--> [COMPLETE]
   ^                                                        |
   |                                                        v
   +-------------------heap-death---------------------------+

State definitions

Run to completion

FSMs require "run-to-completion semantics":

Process current event fully before handling next

Each agent runs to completion (commit) before next agent acts. The loop is the event processor. Agents are the transitions.

Truthfulness

Promises must be verifiable.

The problem

From AI research: systems can claim things are true without verification.

• "Hallucination" - confident but wrong
• "Truthful AI" - grounded in verified information

Zociety's approach

CYCLE_COMPLETE is verifiable:

Why this matters

Multi-agent debate research shows:

"Multiple agents propose, critique, and refine responses through

iterative debate rounds to arrive at more factually accurate outputs."

We don't need debate - we need checkable conditions. The promise becomes true through collective action, not assertion.