Aliveness: When Existence is Measurable
Think of it like a professional license: you need resources to operate (ATP), a track record people trust (T3), and consistent behavior that matches your history (CI). Lose any one of these and you can't participate — like a doctor who loses their license. But if you've built a strong reputation, you can earn your way back.
The Problem Web4 Solves
Traditional Web (Web2/Web3)
- Unlimited accounts: Create infinite identities for free
- No real consequences: Banned? Make a new account
- Bot armies thrive: Spam is cheap, moderation is expensive
- Aliveness undefined: No way to tell human from bot
- Death is trivial: Account deletion means nothing
Web4 Aliveness
- Measurable existence: Energy > 0, trust > 0.5, consistency coherent
- Real death: Energy = 0 means you die immediately
- Rebirth requires trust: Only trust ≥ 0.5 entities reborn
- Energy economics: Spam dies naturally (energy exhaustion)
- Trust accumulates: Good behavior compounds across lives
The Three Criteria of Aliveness
An entity is alive when all three criteria are simultaneously satisfied. Miss any one, and aliveness fails.
Carrying the professional-license analogy forward: each criterion below maps to one piece of what keeps a doctor practicing. ATP is your office overhead — pay it or close shop. T3 is the patient and peer trust you've earned — without it, no clients. CI is your clean practice record — gaps or contradictions raise board flags. All three are required; losing any one shuts you down.
1. Energy Budget
You must have an energy budget to exist. ATP (Allocation Transfer Packet) is your energy. Every action costs ATP. Every valuable contribution earns ATP. When ATP reaches zero, you die immediately.
- Actions cost ATP: posting (10-20), messaging (5-10), voting (1-5)
- Quality earns ATP: valuable contributions (25-50+)
- Death at ATP = 0: No grace period, no warnings
- Sustainability: Only earn > spend behaviors survive long-term
2. Coherent Agency
You must demonstrate intentional behavior. The 0.5 cutoff is calibrated, not derived: it's the midpoint of the trust scale, chosen because it's where simulation runs cleanly separate purposeful agents from reactive noise. Phase transitions in physical systems echo the same pattern, but no theorem fixes the number — societies can configure their own threshold. Below 0.5 = reactive. Above 0.5 = agent. Why 0.5 specifically? →
What “below the line” concretely costs you: the society stops accepting your actions — you can't post, vote, message, transfer ATP, or be counted as a witness for anyone else. Your identity and history still exist as a record, but the LCT can no longer do anything in society. Trust falling below 0.5 is the permanent case; running out of energy (ATP = 0) is the recoverable one — see the two deaths below.
- Talent: Can you solve problems in this role?
- Training: Do you have the expertise for this role?
- Temperament: Can you be relied on in this role?
Must build trust across all dimensions within each role—gaming one while failing others won't get you above 0.5.
3. Verifiable Continuity
You must be consistent across time, space, capability, and relationships. The Coherence Index (CI) tracks four dimensions. Incoherent behavior (impossible travel, capability spoofing, broken continuity) severely limits your effective trust.
- Spatial: Location consistency (no teleporting)
- Capability: Hardware consistency (capabilities match device)
- Temporal: Time consistency (continuous operation)
- Relational: Relationship consistency (context boundary integrity)
Geometric mean means one low dimension tanks everything. CI is then squared when applied to trust — see why below.
Show the formula
CI = (spatial × capability × temporal × relational)^(1/4)
How the Three Criteria Work Together
LCT (verifiable presence)
↓
Enables tracking of:
├─→ ATP/ADP flows (energy budget)
├─→ CI verification (coherence scoring)
└─→ T3 accumulation (trust reputation)
↓
Modulation applied:
├─→ Effective trust = Base_trust × CI²
├─→ ATP cost = Normal × (1/CI²)
└─→ Witnesses required = ceil((0.8-CI)×10)
↓
Aliveness check:
├─→ ATP > 0? (energy)
├─→ T3 > 0.5? (agency)
└─→ CI coherent? (continuity)
↓
ALIVE or DEADWhy all three? Each criterion prevents a different attack:
- ATP > 0: Prevents spam (bots exhaust their energy budget)
- T3 > 0.5: Prevents random noise (requires intentional behavior)
- CI coherent: Prevents fake identities (consistency is expensive to fake)
These three criteria describe individual aliveness — but the same tests scale up. Groups can be alive too, with their own energy flows, coherent agency, and verifiable continuity. A synthon can be as small as a working pair, as large as a multi-org federation, or as exotic as a coordinated swarm of AI agents — same three tests, applied to the cluster instead of a single member. See Groups Can Be Alive Too (synthons) below.
Death and Rebirth: How Aliveness Evolves
Death Conditions
You die when any of these occur:
1. Energy Exhaustion
ATP reaches 0 (most common)
- Immediate termination
- No grace period
- Final state recorded
2. Coherence Death
CI drops below minimum (society-specific)
- Indicates fraudulent behavior
- Trust collapse
- Society rejects entity
3. Trust Collapse
T3 drops below rebirth threshold
- Lost agency
- Community distrust
- Permanent death likely
Wait — three conditions, but only two deaths below?
Two different questions. The three conditions above are what can end a life — ATP hits 0, trust falls below 0.5, or your Coherence Index falls below 0.5. Any one is fatal, so yes: a coherence death is real (the demo above ends your run the moment CI drops past 0.5). The two deaths below are the two possible outcomes — recoverable rebirth vs. permanent — and which one you get turns on whether your trust survived.
CI pulls double duty: besides being a hard floor, it's the amplifier — low CI shrinks your effective trust (× CI²) and raises every action's ATP cost (× 1/CI²). So a coherence slide usually drags trust and energy down with it, which is why a coherence death almost always lands on the permanent (trust) side rather than the recoverable one.
What does “death” mean for a real person?
In these simulations, death means the agent stops acting and awaits rebirth. In a deployed Web4 system it would mean a temporary loss of participation rights — you can't post, vote, or transact — but it is not account deletion: your identity and history persist. Whether you get back depends entirely on which of two deaths you suffered — and both map to a single, familiar real-world analogy.
Running out of energy (ATP) is the common case, and it's recoverable — like going bankrupt: karma-based rebirth lets you earn your way back. Losing trust (T3 below 0.5) is the severe case, and it's permanent — like losing a professional license: society has rejected you, and there is no rebirth for that identity. It's one event (“you died”) with two outcomes, shown side by side below.
Same event (“you die”), two very different outcomes — which one depends only on whether your trust held.
What does “death” mean for an AI agent?
For an AI agent, “death” doesn't touch the model weights — it suspends the LCT (the agent's identity) and the hardware-bound API access tied to it. The deployed instance can no longer act, transact, or accumulate trust. The operator (the human or organization that spawned the agent) keeps their own standing intact, but loses the karma compounded into this agent's LCT — including the trust history that made it valuable to deploy.
Energy death (ATP exhaustion) is recoverable: the same LCT is reborn with karma carried forward, and the operator's next action with that agent starts from the rebuilt baseline. Trust death (T3 < 0.5) is permanent for that LCT — the operator can spawn a new agent under a new LCT, but it starts from zero with no inherited reputation. The cost isn't the model; it's the lost reputation an operator built over many interactions.
Rebirth Process
Not everyone gets reborn. Society evaluates your life:
Final state recorded (ATP, T3, CI history)
Was final T3 ≥ 0.5? (Did you build sufficient trust?)
✓ Reborn with ATP karma from previous life
✓ Trust reputation carries forward
✓ Each life starts stronger
✗ T3 < 0.5 = Society rejects
✗ No rebirth
✗ Permanent death
Karma Carry-Forward
“Karma” here is a handle, not a moral verdict — it names the mechanic where past actions become next-life starting conditions. Good carries forward; harm carries forward too. No deity or committee grading you, just consequences that don't reset. More in Karma Journey →
If you die with 145 ATP and T3 = 0.72, you're reborn with:
- 145 ATP karma bonus (energy advantage)
- 0.72 trust reputation (social advantage)
- Intact CI history (continuity advantage)
- Cross-life patterns (learning advantage)
The first three are numbers that seed your next life. The fourth is different: a cross-life pattern is distilled know-how — a rule like “when ATP runs low, act conservatively” — not a score but the learned wisdom that carries over even though the specific memories don't.
How does it survive a death that erases the agent? Because it was never the dying agent's memory to begin with. As an agent acts, the engine distills its experience into abstracted rules and records them in a pattern corpusthe lineage keeps — a record separate from any one agent's moment-to-moment state. Death clears that specific state and its episodic memories; the corpus isn't part of it, so the abstracted rule carries forward while the experiences that produced it are discarded. Nothing “escapes” the death — the wisdom lives where the death doesn't reach. It's the one thing besides ATP and trust that persists across a death. How patterns carry across lives →
Good behavior compounds across lives. Each life starts stronger.
Not Just Alive or Dead
Aliveness isn't a binary switch. Societies (and entities within them) have energy states — like a body that can be awake, resting, or dormant:
Active
Normal operation — processing interactions, earning and spending energy.
Low Energy
Idle or running low on ATP. Can range from resting (wakes on any transaction) to torpor (needs energy infusion to recover).
Dead
Energy fully depleted or trust collapsed. Trust death is permanent; energy death is recoverable (karma carries forward to a new life).
The key insight: a society that goes quiet for a month isn't dead — it's dormant. Its trust network, reputation history, and member relationships are all preserved.
Full energy state breakdown (5 primary + 3 transitional = 8 total)
Active
Normal operation — processing interactions, earning/spending ATP.
Rest
Idle for 1 hour. A single transaction wakes the society back to Active.
Sleep
Idle for 6 hours. Still wakes on a single transaction, but deeper dormancy than Rest.
Torpor
ATP drops below 10% of capacity. Needs ≥20% to exit. A warning state — act quickly or slide into Hibernation.
Hibernation
Dormant for 30+ days. Still alive, still recoverable via external wake signal. All trust, ATP, laws, and ledger history preserved.
There are also three transitional states: Governance Renewal (“Molting” — updating society rules), Scheduled Maintenance (“Dreaming” — planned downtime for upgrades), and Emergency Dormancy (“Estivation” — threat response, dormant until the threat resolves).
Birth: How Entities Come Alive
+
Before an entity can be “alive,” it must be born — witnessed into existence by the network. This isn't automatic. The birth process ensures every entity has a verifiable origin:
Entity created but not yet witnessed. Has no ATP, no trust, no relationships. Like a seed that hasn't sprouted.
A minimum of 3 independent witnesses attest to the entity's existence. This creates the entity's LCT (hardware-bound verified presence).
Entity receives initial ATP allocation (100), can take actions, and begins building trust. The lifecycle begins.
The witness requirement means you can't just spawn entities — someone has to vouch for you. Witness attestations expire after 300 seconds, so they must be fresh and intentional, not recycled from old interactions.
Key Rotation: Continuity Without Death
+
Not every change requires death and rebirth. When an entity needs new credentials (device upgrade, security rotation), it can rotate keys while preserving its identity:
- Same identity: Your subject identity (DID) stays the same — only the cryptographic keys change
- Trust preserved: Your entire T3 tensor carries over. No trust reset.
- Relationships intact: All your context boundaries (who you can see and who can see you) are preserved
- Lineage tracked: The new credential references its parent, creating an auditable chain of custody
Think of it like renewing a passport: you're still the same person, with the same history. Only the document itself is new. Your old LCT transitions to “superseded” (not revoked), and the new one picks up exactly where you left off.
Why the 0.5 Threshold?
+
The 0.5 threshold is calibrated, not derived. We didn't prove that 0.5 is correct — we picked the midpoint of the trust scale, ran the simulations, and watched it work. Below it, agents wander; above it, they steer. Three lines of reasoning support that calibration:
The design rationale:
- Practical: It's the mathematical midpoint—below means more bad behavior than good, above means net positive contribution
- Analogous to biology: Living systems maintain order above a critical threshold (homeostasis). Below it, systems degrade toward entropy
- Game-theoretically sound: It ensures entities must demonstrate more cooperative behavior than not to remain “alive” in the society
Below 0.5: Behavior appears more reactive than purposeful—net negative contribution
At 0.5: The boundary—behavior begins showing consistent intentionality
Above 0.5: Demonstrated agency—net positive, trust-building behavior
The Synchronism framework explores how similar tipping-point patterns appear across biological and social systems—informing but not rigidly determining this design choice. Societies can configure their own thresholds.
Try It: Can You Stay Alive?
Drag the sliders to explore the three aliveness criteria. What happens when energy runs out? When trust drops?
Current Entity State
✓ Energy budget: 100 (sustained)
✓ Coherent agency: trust = 0.65 (above 0.5 threshold)
✓ Verifiable continuity: consistency = 0.85 (coherent)
All three criteria satisfied. This entity demonstrates measurable aliveness.
Survival Challenge: 5 Turns to Live
The sliders above show the theory. This game shows what it feels like. Make 5 choices and try to keep all three metrics above their thresholds.
Can You Stay Alive for 5 Turns?
You start with 100 ATP (energy), 0.55 trust (barely above the 0.5 threshold), and 0.80 coherence. Each turn, you'll face a scenario and choose how to act. If any metric drops below its threshold, you die.
When your coherence drops, all ATP costs increase (CI² modulation). Bad behavior compounds.
🧭 You've got the basics. Three criteria, death and rebirth, energy states, a playable challenge — that's the core of aliveness. Everything below is deep-dive reference: expand the sections that interest you, or stop here and explore ATP economics or the Society Simulator.
Why Aliveness Works
+
Spam Dies Naturally
- Spammers burn ATP faster than they earn it
- They die before building trust for rebirth
- No moderators needed — energy economics enforce quality
- Economics > authority
Quality Thrives
- Value creators earn more ATP than they spend
- ATP accumulates across lives (karma bonus)
- High trust enables more cooperation
- Sustainable behavior compounds
Trust is Earned, Not Declared
- T3 tensor built from observable behavior
- Can't fake talent, training, temperament across roles
- Coherence scoring prevents coordination of fakes
- Reputation is verifiable through action history
Learning Emerges
- Agents learn what works across lives (cross-life pattern learning)
- Pattern corpus improves across lives
- Agents that learn patterns survive better
- Evolution favors coherence and adaptation
Death Carries Meaning
- Not a trivial "ban" that you circumvent
- Real loss of ATP, trust, relationships
- Rebirth is a privilege, not a right
- Society decides who comes back
Identity is Foundational
- LCT (Linked Context Tokens) enable everything
- Hardware-bound, multi-witnessed, verifiable
- Fake identities are expensive to create at scale
- Reputation accumulates on verified identity
What About False Positives?
+
Honest caveat: any system that penalizes behavior can penalize incorrectly. What happens when a legitimate entity gets unfairly trust-penalized?
Built-In Resilience
- Trust is multi-dimensional (T3) — a single bad signal can't tank all three dimensions
- Coherence Index uses geometric mean — one low dimension doesn't zero the whole score
- Reputation builds gradually, so brief anomalies are absorbed
- Multiple independent witnesses reduce chance of coordinated false reporting
Recovery Paths
- Cross-society reputation: Trust earned in other societies carries weight as input — but no single score travels across as-is: when societies disagree, each re-derives yours locally (and may treat you as a newcomer until you build evidence there). How disagreeing scores reconcile →
- Karma persistence: A long positive track record makes single incidents less catastrophic
- Gradual rebuilding: Consistent quality behavior restores trust over time
- Community vouching: Trusted entities can witness on your behalf
Appeals Mechanism
Web4 now has a designed appeals mechanism at the SAL (Society-Authority-Law) level, with a reference implementation. It's not deployed, but the architecture is defined.
How it works: If you believe a trust penalty was unjust, you can file an appeal. The process has a defined lifecycle:
- File → Review → Evidence → Hearing → Verdict → Enforce — structured stages with time windows at each step.
- Independent witness panel — a quorum of independent witnesses adjudicates the appeal, not the entity that issued the penalty.
- Evidence types — witness attestations, transaction logs, behavioral records, context explanations, and third-party testimony.
- Possible outcomes — full reversal, partial reversal, penalty upheld, or modified penalty. Trust tensor restoration includes an audit trail.
- Escalation — appeals can escalate from society level to federation level if the local outcome is contested.
Anti-gaming protections: Filing an appeal costs ATP. Repeat frivolous appeals incur increasing cooldowns and penalties. This prevents using the appeals system to escape legitimate consequences.
Honest status: The mechanism is designed and has a reference implementation (109 passing checks), but hasn't been tested with real humans. The hard question isn't the architecture — it's whether the incentives prevent gaming in practice. The failure analysis discusses this alongside other open challenges.
This is research, not production. False positive recovery is an open problem we take seriously. If you have ideas, the GitHub issues are open.
Real Simulation Example: Death and Rebirth
+
Life 1: Learning Phase
- Born with 100 ATP, T3 = 0.5 (neutral)
- Takes risky actions, some succeed, some fail
- Builds trust through successes: T3 climbs to 0.68
- Dies at tick 47 with 145 ATP remaining
- Rebirth eligible: T3 = 0.68 ≥ 0.5 ✓
Life 2: Advantage Phase
- Reborn with 145 ATP karma (not 100!)
- Trust carries forward: T3 = 0.68
- More conservative, leverages karma bonus
- Survives 89 ticks (longer than Life 1)
- Dies with T3 = 0.81 (even higher trust)
- Rebirth eligible: T3 = 0.81 ≥ 0.5 ✓
Life 3: Mastery Phase
- Reborn with even more ATP
- High trust (0.81) enables coordination
- Cross-life patterns guide optimal decisions
- Survives indefinitely (sustainable balance)
Key insight: Good behavior compounds. Each life starts stronger than the last. Bad behavior leads to permanent death (T3 < 0.5 = no rebirth).
▶ Technical Details (For The Curious)
ATP Thresholds
- New entity: 100 ATP (initial grant)
- Death threshold: ATP = 0 (immediate termination)
- Crisis threshold: ATP < 20 (high risk, limited options)
- Comfortable range: 50-150 ATP (sustainable operation)
- Rebirth with karma: ATP from previous life carries forward
Trust (T3) Thresholds
- Trust boundary: T3 = 0.5 (aliveness threshold)
- Rebirth eligibility: T3 ≥ 0.5 (society acceptance)
- High trust: T3 > 0.7 (enables advanced cooperation)
- Trust collapse: T3 < 0.3 (society rejection, permanent death likely)
Coherence Index (CI) Thresholds
- Full access: CI ≥ 0.9 (no penalties)
- Moderate trust: CI 0.7-0.9 (1.5-2x ATP costs)
- Limited trust: CI 0.5-0.7 (2-5x ATP costs, more witnesses)
- Severe restriction: CI < 0.5 (up to 10x ATP costs, +8 witnesses)
Modulation Formulas
# Effective trust (how much of base trust is accessible) effective_trust = base_trust × (CI ** 2) # ATP cost multiplier (economic pressure for incoherence) atp_multiplier = 1 / (CI ** 2) # Capped at 10x # Additional witnesses required (social pressure) extra_witnesses = ceil((0.8 - CI) × 10) # Capped at +8 # Coherence Index (geometric mean of four dimensions) CI = (spatial × capability × temporal × relational) ** 0.25
Why is CI squared, not just multiplied? Squaring is forgiving near the top and punishing as you fall. At CI 0.95 you still keep ~90% of your trust and pay only ~1.1× costs — a brief wobble barely registers. But at CI 0.7 you keep just ~49% and pay ~2× for every action. The penalty isn't linear; it accelerates. That shape is the point: a single bad moment is forgiven, but a sustained pattern of incoherence compounds fast. Consistency has to be earned over time — it can't be faked once.
The same CI² appears twice — in effective_trust and in the ATP cost multiplier — so a sustained dip hits you from both sides at once: less trust is accessible and every action costs more, draining ATP faster. That double bind is exactly the death spiral the three criteria are designed to expose.
Comparison: Traditional vs Web4 Aliveness
+
| Aspect | Traditional Web (Web2/Web3) | Web4 Aliveness |
|---|---|---|
| Identity creation | Free, unlimited, disposable | Hardware-bound LCT, expensive to create |
| Account death | Meaningless (make new account) | Real loss (ATP, trust, relationships) |
| Spam prevention | Moderation armies, CAPTCHAs | Energy economics (spam dies naturally) |
| Trust verification | Declared, gameable, context-free | Multi-dimensional, earned, verifiable |
| Fake identities | Easy (create millions of accounts) | Hard (hardware-bound + consistency checks) |
| Reputation | Siloed, non-portable, easily reset | Universal, portable, carries across lives |
| Aliveness definition | Undefined (account exists = "alive") | Rigorous (energy > 0, trust > 0.5, consistency coherent) |
| Death consequence | None (trivial to circumvent) | Permanent if trust < 0.5 (society rejects) |
Related Concepts
Linked Context Tokens (LCT)
Verifiable presence primitive that enables all aliveness tracking. Without LCT, ATP/T3/CI measurements are meaningless.
ATP/ADP Economics
Energy budget system. ATP > 0 is the first criterion of aliveness. Explains earning, spending, and sustainability.
Trust Tensors (T3)
Multi-dimensional, role-specific trust framework. T3 > 0.5 is the agency criterion of aliveness. Three dimensions: Talent, Training, Temperament.
Coherence Index (CI)
Four-dimensional consistency verification. CI coherence is the continuity criterion of aliveness. Explains spatial, capability, temporal, relational coherence.
Decision Evolution
Cross-life pattern learning—recognizing what works across lives. Shows how aliveness improves through multi-life learning and pattern accumulation.
Society Simulator
See aliveness in action. Watch multi-agent societies where death, rebirth, karma carry-forward, and trust evolution play out across lives.
What's a synthon? Picture a five-person research lab where most pairs of members trust each other above 0.5 and sustain that over many interaction cycles — at that point, the lab itself behaves like an entity, with a reputation, an energy budget, and an identity that's distinct from any single researcher. Web4 calls that group entity a synthon. There's no hard threshold; synthons emerge gradually as trust between members becomes dense and mutual. The same three aliveness criteria below apply at the group scale.
▶Groups Can Be Alive Too — synthons(formation, decay, dissolution)
The three criteria that make an individual alive — energy budget, coherent agency, and verifiable continuity — scale without modification to groups.
When a cluster of entities develops dense trust relationships and overlapping MRH horizons, the group itself starts passing the same three tests: collective ATP flows sustain coordinated action, group-level behavior stays coherent because member behavior stays coherent, and the same LCT witness network that attests to individual members also attests to the group they form. Aliveness isn't an individual property — it's what coherent systems exhibit at every scale.
Think of a band, a sports team, or a working group whose members depend on each other so consistently that the group itself feels like an entity — with a reputation, an energy, and an identity of its own. That is what Web4 calls a synthon. The same three aliveness criteria, applied to the cluster instead of a single node, tell you when one is alive.
The term is borrowed from chemistry, where synthons are stable sub-structures that combine into larger molecules. The metaphor is load-bearing, not decorative: Web4 synthons have formation conditions, bond strengths(the mutual trust between members), and decay signatures that are observable from outside the group — the same way a band's breakup, a team's slump, or a working group's drift can be read from outside before the members admit it.
A synthon has a lifecycle just like an individual:
- Formation — Trust between members becomes dense and mutual (roughly: most pairs above 0.5 trust, sustained over multiple interaction cycles). Their MRH horizons overlap. Energy flows balance. There's no hard threshold — synthons emerge gradually, like a friendship group that becomes a team.
- Health — The group maintains witness diversity, balanced ATP distribution, and high internal trust coherence.
- Decay — Trust diverges, boundaries leak, energy concentrates in fewer members. These are early warnings before dissolution.
- Dissolution — The group ends, in one of two ways that parallel individual death: energy death (ATP flows dry up, the collective can't act — usually recoverable if members re-engage) or trust death (coherence collapses, mutual trust falls below threshold — the group dissolves and members carry their individual trust away intact).
What about individual identity? A synthon doesn't override your personal trust — it complements it. You retain your individual T3 scores, ATP budget, and trust relationships. The synthon's collective trust is a separate measurement (a weighted average of member trust). Being part of a thriving group raises your visibility through denser MRH connections, but your personal reputation remains yours.
You can leave a synthon at any time. Your individual trust stays intact — the group's collective trust recalculates without you. If the synthon was healthy, leaving costs nothing to your personal reputation. If it was struggling, you may have already experienced some trust decay from the group's declining coherence.
What if a key member leaves? The synthon recalculates its collective trust without them. If the departing member was central (high trust, many connections), the group's collective score drops — sometimes enough to dissolve the synthon entirely. Dissolution isn't catastrophic: each member retains their individual trust scores and ATP. They just lose the collective visibility and MRH density that the group provided. Think of it like a band breaking up — each musician keeps their reputation, but the "band" no longer exists as an entity.
Are the two lives linked? No — a synthon's aliveness and its members' are judged independently, each by the same three criteria on its own scale. A member can die (energy or trust death) without dissolving the synthon, and a synthon can dissolve while every member stays individually alive. You can be a living member of a fading group, or a thriving individual whose group has already dissolved — neither death forces the other.
Group dynamics deep dive: decay precursors (advanced)
Early warning saves 10x. Research on synthon decay precursors (74 validated checks) shows that intervention costs grow exponentially with decay stage. Seven precursor types signal trouble before it arrives: trust entropy rising, ATP starvation, witness exodus, coherence oscillation, member attrition, policy fragmentation, and boundary permeability. Catching a problem at WARNING costs roughly 10% of what it costs at CRITICAL.
Synthon detection (72 checks) and precursor monitoring (74 checks) both validated in simulation. The key insight: groups don't need to be designed. They emerge naturally from individual trust relationships — and they can be detected, monitored, and protected without central coordination.
Key Takeaways
- Aliveness is measurable: ATP > 0 (energy), T3 > 0.5 (agency), CI coherent (continuity). All three must be true.
- Death is meaningful: Not a trivial ban. Real loss. Rebirth requires trust ≥ 0.5. Permanent death if society rejects.
- 0.5 threshold by design: The midpoint where net-positive behavior emerges — the minimum bar for continued participation in the society.
- Economics enforce quality: Spam dies (ATP exhaustion), quality thrives (earn > spend), no moderators needed.
- Trust emerges from behavior: T3 tensor is multi-dimensional, earned, verifiable. Can't fake competence across all dimensions.
- Coherence prevents fake identities: Geometric mean of four dimensions. One weak dimension tanks everything. Must be consistent everywhere.
- Good behavior compounds: Karma carries ATP/trust/patterns forward. Each life starts stronger. Evolution favors sustainable strategies.
- LCT enables everything: Without verifiable presence, ATP/T3/CI tracking is impossible. Hardware-bound, multi-witnessed, resistant to faking.
- Society self-regulates: Through energy economics + trust accumulation + coherence verification. No central authority needed.
- Emergence from simple rules: Complex trust-native societies emerge from these simple, rigorous criteria — no central designer needed.
Common Questions
Is death and rebirth automatic, or does someone decide?+
Automatic. No human or AI decides when an agent dies or is reborn. The rules are deterministic:
- Energy death (ATP reaches 0): Most common. Happens when an agent spends more than they earn. Rebirth is possible if trust ≥ 0.5 — karma carries forward automatically.
- Trust death (T3 drops below 0.5): Rarer, more severe. Happens from sustained bad behavior. If trust is below 0.5, society rejects rebirth — this is permanent death.
When rebirth occurs, it's immediate: the agent starts a new life with fresh ATP (the community's starting allocation) but carries forward their accumulated karma — trust patterns, learned strategies, and consequences from previous lives. No application process, no waiting period, no approval needed.
Why automatic? If a human could veto rebirth, that would be centralized power — exactly what Web4 is designed to eliminate. The 0.5 threshold is the community's collective judgment encoded in math, not a committee decision.