hermes-brain/projects/passepartout/strategy/game-theory.org

:PROPERTIES:
:ID:       72db9428-d6e4-472d-9693-49335f888e48
:CREATED:  [2026-05-25 Mon]
:END:
#+title: Game Theory
#+filetags: :passepartout:strategy:adoption:game-theory:

Why the adoption trajectory is structural, not just plausible. This
document models the key strategic interactions as formal games with
payoff matrices and Nash equilibria. Phase numbers refer to the
[[id:6d2e3f4a-5b6c-7d8e-9f0a-1b2c3d4e5f6a][Adoption]] document.

The central claim: the trajectory toward verified computing is the unique
stable equilibrium of a multi-player game, not one of several possible
outcomes. This document models the five core games — enterprise adoption,
regulator commitment, hyperscaler defense, social protocol platform
competition, and geopolitical competition between nation states — and
shows that each has a unique Nash equilibrium that
reinforces the others.

* Game 1: Enterprise Adoption (coordination with network effects)

Two firms in a regulated industry choose whether to adopt a gate or
maintain their current compliance process. Their payoffs depend on what
the other firm does and on the regulator's stance (set independently in
Game 2).

**Strategy sets:** {Adopt Gate, Maintain Current Audit}

**Payoff parameters:**

Let:
- C = annual compliance cost under current audit ($200K-$1M)
- G = annual gate subscription cost ($50K)
- R = risk of audit failure (probability × penalty, ~$500K/yr expected)
- D = competitive disadvantage if rival adopts gate and you don't (productivity gap, estimated 2-5% of compliance-adjacent revenue)
- N = network effect benefit if both adopt (shared gate rules, interoperable proofs, collective regression suite)

All values are per-firm per-year.

**Payoff matrix (firm A row, firm B column):**

| | B adopts gate | B maintains audit |
|---|---|---|
| A adopts gate | (-G + N, -G + N) | (-G, -D) |
| A maintains audit | (-D, -G) | (-C - R, -C - R) |

**Nash equilibria:**

1. Both adopt gate is a Nash equilibrium if G - N < C + R. Since G ≈ $50K,
   C ≈ $200K-$1M, R ≈ $500K, and N is positive, this inequality holds
   strongly — adopting the gate strictly dominates maintaining the audit
   when both firms adopt.

2. Both maintain audit is also a Nash equilibrium if D < C + R - G. If
   the competitive disadvantage of being the only non-adopter is small,
   both firms can get stuck in the current audit even though adoption
   would make both better off. This is a classic coordination failure.

**What breaks the coordination failure:**

The two equilibria exist because each firm's payoff from adopting
depends on the other firm's choice. But the game changes when:

- **The regulator encodes a rule as a gate.** This removes the audit
  option for both firms — adoption becomes mandatory. The game collapses
  to a single outcome.
- **Insurance differentiates on verification.** The compliance cost C
  becomes C + I (insurance premium surcharge for unverified), where I
  can be 2-10x of C. This pushes D < C + R + I - G even for the first
  adopter, making unilateral adoption dominant.
- **One firm has a large enough D that they adopt unilaterally.** If
  the competitive disadvantage of being the only non-adopter exceeds the
  adoption cost, the first firm switches, and the second firm follows
  because -G + N > -D. This is how the flywheel starts.

**Key insight:** The coordination failure exists only in the absence of
regulatory mandate or insurance differentiation. Both of these are
themselves equilibrium outcomes of separate games (Game 2 and Game 3
interaction). The enterprise adoption game does not create a structural
barrier — it creates a timing problem that the other games resolve.

* Game 2: Regulator Commitment

A regulator chooses how to enforce a rule. An industry of N firms has
already chosen their adoption status (partly determined by Game 1).

**Strategy sets:** 
- Regulator: {Encode as gate, Maintain paper-based enforcement}
- Fate chooses the political environment: {Pro-gate, Anti-gate}

**Payoff parameters:**

Let:
- E = enforcement effectiveness (paper: 0.3, gate: 0.95 on a 0-1 scale)
- L = legitimacy cost of choosing the "wrong" approach (0 if aligned with
  political environment, k > 0 if misaligned)
- T = political turnover risk (probability that the next administration
  reverses the decision)
- B = benefit to regulator of career-defining legacy (gate: +M, paper: 0)
- S = surveillance-value loss to the state when data becomes invisible to
  bulk collection (paper: 0, gate: -S, where S varies by regime type)

**Regulator's payoff function:**

U(gate) = αE + βB - γL(gate|environment) - δT - εS
U(paper) = αE(paper) - γL(paper|environment) + δT(paper)

Where α, β, γ, δ, ε are the regulator's preference weights.

**Key parameter: S (surveillance-value loss)**

For a democratic regulator in an EU-style system, S ≈ 0 or small — bulk
surveillance is constrained by law. The gate's privacy properties are a
feature, not a cost. The regulator's choice depends primarily on E, B,
and L.

For an authoritarian regulator, S is large. The gate makes their
regulated entities invisible to bulk collection. Even with high E and B,
the payoff may be negative. The regulator's dominant strategy is to
maintain paper enforcement or ban gates.

**Nash equilibria by regulator type:**

1. **Democratic/pro-rule-of-law regulator:** U(gate) > U(paper) when
   α(E_gate - E_paper) + βB > γ(L_gate - L_paper) + δ(T_gate - T_paper).
   Since E_gate >> E_paper, B > 0, and S ≈ 0, the gate dominates for
   any reasonable parameter range. Unique equilibrium: encode gate.

2. **Authoritarian/high-surveillance regulator:** U(gate) may be
   negative if εS > α(E_gate - E_paper) + βB. Unique equilibrium:
   maintain paper or ban gates.

3. **Captured regulator:** If incumbents (Game 3) successfully lobby,
   γ(L_gate|pro-incumbent) can be made large. The equilibrium depends on
   whether the capture is stronger than the enforcement benefit.

**How Game 1 and Game 2 interact:**

If Game 1 reaches "both adopt gate" without the regulator, the
regulator's payoff from encoding increases further — there is now proven
infrastructure, lowering L and T. This makes encoding more attractive
even for a borderline regulator.

If Game 1 is stuck in coordination failure (both maintain), the
regulator can break it by encoding — this is the most leveraged
intervention. A forward-leaning regulator who encodes early creates the
condition for Game 1 to tip.

**What changes the equilibrium:**

- A high-profile AI harm event increases B and decreases L for the
  regulator choosing gate (the public demands action).
- An incumbent lobbying campaign increases γ(L_gate) if the regulator
  is vulnerable to capture.
- A regime change can flip S from 0 to large or vice versa.

* Game 3: Hyperscaler Defense

An incumbent hyperscaler (AWS, Azure, GCP) chooses how to respond to the
gate architecture. Enterprise adoption level a ∈ [0,1] is a parameter.

**Strategy sets:** 
- Hyperscaler: {Fight, Accommodate, Co-opt}
  - Fight: lobby against gate standards, bundle a fake "verified" wrapper,
    use procurement lock-in to deter adoption
  - Accommodate: support gate instances on their infrastructure, capture
    revenue from the compute that gate instances need
  - Co-opt: acquire or clone the gate provider, offer gate-as-a-service
    on their terms

**Payoff parameters:**

Let:
- R_h = hyperscaler's annual revenue from the enterprise segment that
  gates target ($B+ per hyperscaler in regulated markets)
- S_h = share of R_h that depends on unrestricted data access
  (surveillance advertising, training data extraction, platform lock-in)
- P_h = cost of fighting (lobbying spend, engineering cost of fake
  wrapper, reputation damage if exposed)
- A_h = revenue from accommodating (compute for gate instances, storage
  for proof logs — lower margin but insulated from competition)
- C_h = cost of co-opting (acquisition price or clone engineering cost)

**Hyperscaler's payoff by strategy and adoption level a:**

| Strategy | a < 0.1 (Phase 0-1) | a > 0.1 (Phase 2+) |
|----------+---------------------+---------------------|
| Fight | -P_h + (1-k)R_h (preserves most revenue but spends on lobbying) | -P_h + (1-k')R_h (lobbying less effective; revenue erosion accelerates) |
| Accommodate | A_h × a (small but growing revenue from gate compute) | A_h × a + residual non-gate R_h (gate revenue grows, non-gate revenue declines) |
| Co-opt | -C_h + R_h (full control but high cost) | -C_h + R_h (but harder to execute — gate ecosystem has network effects) |

Where k, k' are revenue erosion rates from gate adoption (k < k').

**Nash equilibria:**

1. **At low adoption (a < 0.1):** Fight dominates if -P_h + (1-k)R_h >
   A_h × a and -P_h + (1-k)R_h > -C_h + R_h. The hyperscaler tolerates
   the lobbying cost because it preserves most of their business model.
   This is the equilibrium in Phases 0-1.

2. **At high adoption (a > 0.1):** If the gate ecosystem has enough
   momentum that A_h × a + residual > fight payoff, and co-opt is
   infeasible (network effects of the open regression suite, AGPL
   license, community), then Accommodate becomes dominant. The
   hyperscaler accepts the two-tier outcome and captures what they can
   from gate compute.

3. **Co-opt is only dominant if C_h < P_h + kR_h — the hyperscaler must
   believe they can acquire the gate provider cheaper than fighting or
   losing revenue. The AGPL license and the open nature of the
   regression suite make co-opt difficult (you can buy the code but you
   can't buy the community).

**Key insight:** The hyperscaler's optimal strategy changes with
adoption level. At low adoption they fight because the cost is low and
the revenue preservation is high. At a threshold adoption level, the
fight becomes more expensive than accommodation. This threshold is
determined by S_h — the share of revenue dependent on unrestricted data
access. A hyperscaler with high S_h (Google, whose advertising business
depends on data) will fight harder and longer than a hyperscaler with
lower S_h (AWS, whose revenue is more infrastructure-based).

**What changes the equilibrium:**

- A high S_h means the fight threshold is higher — the hyperscaler
  tolerates more revenue erosion before accommodating.
- AGPL license and community ownership raise C_h (cost of co-opting)
  and lower the probability of co-opt succeeding.
- If adoption jumps suddenly (regulator encode in Game 2), the
  hyperscaler may skip the fight phase entirely and move directly to
  accommodation.

* Game 4: Social Protocol Platform Competition

The social protocol does not compete with any single platform. It offers
an alternative to the entire paradigm of centralized internet services —
a single protocol that replaces 20+ products across social graph,
publishing, video, messaging, e-commerce, payments, contracts, identity,
code hosting, collaboration, and freelancing. See the [[id:1bc22b89-d3eb-4f6d-bcfc-2b0c19c8ed8f][Social Protocol
Competitive Landscape]] for the full platform map.

**Why this is a different kind of game:**

Unlike the institutional games (binary adopt/don't-adopt decisions by
firms), the social game is a multi-front platform competition where:

1. The protocol competes against 20+ incumbents simultaneously, each with
   its own network effects and switching costs.
2. No single incumbent can copy the bundle — Meta cannot offer portable
   identity (it destroys their surveillance model), Google cannot offer
   private messaging (it destroys their ad model), Stripe cannot offer
   contracts and social, DocuSign cannot offer payments.
3. The protocol's value proposition is not "Twitter but better" — it is
   "one account replaces every platform you use."
4. Users do not switch all at once. They join for one use case and
   discover the rest. This is a sequential expansion game.

**The entry vector game entry choice:**

The protocol must choose which use case to lead with. Each candidate has
different payoff parameters:

| Entry vector | Cold start | ARPU | Bundle necessity | Competitive response | Failure mode |
|-------------+------------+------+-----------------+---------------------+--------------|
| Organized communities (HOAs, clubs, PTAs) | Solved (groups exist as units) | Low ($20-100K fees) | Full | Negligible (no incumbent targets this segment) | No community finds PMF |
| Community refugees (banned subreddits, nuked discords) | Solved (arrive together) | Low-medium | Partial | Medium (Reddit/Discord defensive) | Migrations don't stick |
| Creators (OnlyFans, Patreon, adult content) | High (individual migration) | High ($1-5M fees) | Partial (identity + content + payments) | High (OnlyFans, Stripe, payment processors) | Creator acquisition cost too high |
| Freelancers (Upwork, Fiverr) | Low (scattered, no density) | Medium | Full (contracts + payments + reputation) | Low (Upwork network effects weak) | No dispute resolution trust |
| Developers (GitHub) | High (open-source communities exist) | Low | Partial (code + identity) | Very high (Microsoft/GitHub moat) | Developer habit inertia |

**The entry vector game payoff matrix (protocol chooses primary vector,
fate chooses incumbent response strength):**

Let:
- U_x = users acquired through vector x at Phase 0
- C_x = engineering cost to ship the features vector x requires
- R_x = revenue from vector x users
- I_x = incumbent response strength (0 = none, 1 = existential fight)
- S_x = stickiness (probability user stays for other capabilities after
  joining for vector x)

Expected protocol payoff for choosing vector x:
  P(x) = U_x × (R_x + S_x × future_value_of_expansion) - C_x - D_x

Where D_x is damage from incumbent response (lobbying, FUD, feature
parity, price cuts) weighted by I_x.

**Nash equilibria by entry vector:**

1. **Organized communities (Phase 0):** I_x ≈ 0 (no existing platform
   serves this segment well). U_x is bounded but guaranteed (groups
   exist). S_x is high (full bundle forces exposure to all capabilities).
   Unique equilibrium: this is the dominant entry strategy.

2. **Creators (Phase 0 parallel):** I_x is high (OnlyFans, Stripe,
   payment processors have incentive to block). But R_x is also high
   (highest ARPU). U_x is unpredictable (depends on creator acquisition
   cost). Multiple equilibria: high-I/low-U leads to "do not enter,"
   low-I/high-U leads to "enter." The outcome depends on whether payment
   discrimination against adult content is acute enough to overcome
   switching friction.

3. **Freelancers (Phase 2):** I_x is low (Upwork's network effects are
   weak — both sides multi-home). But S_x requires the reputation graph
   from Phase 0-1 to exist first. This is a sequential game — the
   payoff from enter in Phase 2 depends on having entered Phase 0-1
   successfully.

**The asymmetric bundle advantage:**

The protocol's structural advantage is that it competes with 20+
incumbents, but no incumbent competes with the protocol. Each incumbent
faces a different threat:

| Incumbent | Threat level | Can adapt? | Adaptation would require |
|-----------+-------------+------------+--------------------------|
| Meta (Facebook, Instagram) | Existential | No | Abandon surveillance advertising and portable user data — their entire business model |
| Google (YouTube, Gmail, Google Docs) | High | No | Abandon data mining for ads — their entire business model |
| Microsoft (GitHub, LinkedIn, Office) | Moderate | Partial | Accept decentralized identity — but Windows/Office lock-in is different from data mining |
| Twitter/X | Medium | No | Algorithmic feed is their product — giving up curation control destroys their ad model |
| Stripe | Low | No | Cannot offer social, contracts, and identity — outside their competence |
| Discord | Medium | Possible | Can add more features but cannot offer portable identity or zero-fee payments |
| Substack/Medium | Medium | Possible | Can improve creator tools but cannot match zero-fee + censorship resistance |
| OnlyFans/Patreon | Low | No | Payment discrimination is structural (banking regulations) — not a choice they can undo |
| DocuSign | Low | No | Contracts + payments + social is outside their competence |
| Reddit | Low | Possible | Decentralized moderation would destroy their ad business |

Key insight: of the 20+ incumbents, most (Meta, Google, Twitter, Stripe,
DocuSign) cannot adapt because adaptation requires abandoning their
business model. A few (Discord, Substack, Reddit) can make marginal
improvements but cannot match the bundle because they don't control the
other layers. Zero incumbents can match the full protocol.

**Geopolitical dimension: free speech and association infrastructure:**

The social protocol is not just a consumer product. Its architecture
makes it naturally censorship-resistant and surveillance-resistant:

- Speech on the protocol cannot be removed by any government because
  there is no central server to issue takedown orders to. Each user
  controls their own PDS; content is addressed by CID, not hosted on
  a platform URL.
- Association on the protocol cannot be surveilled because the relay
  network routes by DID, not IP. The state can see that a message was
  sent but cannot determine who sent it to whom without controlling
  the entire relay graph.
- Organisation on the protocol cannot be disrupted because Collective
  Personas exist cryptographically — there is no office to raid and
  no server to seize.

This makes the protocol a direct geopolitical tool. For citizens in
restrictive regimes (China, Russia, Iran, Belarus, Myanmar, Eritrea,
and dozens of others), the protocol offers the first infrastructure
that enables free speech and association that their government cannot
control, surveil, or shut down.

This changes the entry vector analysis:

| Entry vector | Previously framed as | Geopolitical frame adds |
|-------------+---------------------+------------------------|
| Community refugees | Banned subreddits, nuked discords | Political dissidents, exiled journalists, banned NGOs — users fleeing state censorship, not just platform moderation |
| Creators | OnlyFans/Patreon refugees | Journalists and writers under threat in authoritarian regimes — their incentive is not just revenue but survival |
| Organized communities | HOAs, clubs, PTAs | Exile communities, diaspora groups, cross-border activist networks — users with an organizational structure that cannot exist on any centralized platform |

The geopolitical entry vectors have different payoff parameters than
the consumer ones. Their cold start is solved not by group density but
by necessity (the alternative is censorship or imprisonment). Their
stickiness S_x is near-maximal because the protocol is not a convenience
but a lifeline. Their ARPU may be low but their **strategic value** is
high — a single dissident community successfully using the protocol
demonstrates censorship resistance to millions.

**How this changes Game 5 (geopolitical competition):**

The legitimacy cost L_i for a state that bans the protocol was modeled
in Game 5 as a general cost. But the social protocol's free speech
dimension makes L_i much larger than a standard trade restriction:

- A state that bans the protocol is not just blocking a technology — it
  is telling its citizens "we are afraid of you speaking freely."
- The protocol makes the ban visibly ineffective — citizens who can
  access the relay network retain their speech. The state either
  accepts visible defiance or invests in internet shutdowns, both of
  which carry high legitimacy costs.
- The protocol's verification layer means that a banned instance can
  still *prove* it is operating correctly. A state that claims to have
  shut down gates cannot fake compliance.

This creates a new dynamic in Game 5: states that ban the protocol face
a legitimacy cost that grows with the protocol's adoption in free-world
jurisdictions. The more citizens in free countries use the protocol, the
more citizens in restricted countries know what they are missing, and
the higher the legitimacy cost of the ban.

**The interaction between Game 4 (social) and Game 5 (geopolitical):**

| Direction | Effect |
|-----------+--------|
| Game 4 → Game 5 | Social protocol adoption in free-world jurisdictions increases L_i for restrictive states (citizens in restrictive regimes see what they're missing; free-world users amplify dissident content). |
| Game 4 ← Game 5 | A restrictive stance in Game 5 (China bans gates, Russia blocks relays) makes the social protocol's censorship resistance the *only* option for citizens in those countries, accelerating adoption as a human rights tool. |
| Game 4 → Game 2 | Social protocol's demonstrated censorship resistance gives democratic regulators evidence that verification is compatible with free speech, lowering their L(gate|political) in Game 2. |

The net effect: the social protocol's geopolitical dimension does not
just make it more resilient — it creates a positive feedback loop with
the geopolitical game that no purely consumer platform can generate.

**Tech industry alignment: natural allies:**

The protocol creates winners and losers in the tech industry. Companies
whose business model depends on surveillance advertising (Meta, Google,
Twitter/X, TikTok) are structural losers — the protocol's verification
layer blocks their data extraction. But companies whose business model
is compatible with or enhanced by verification are structural winners.
These companies have an incentive to embrace the protocol as a
competitive weapon against their surveillance-dependent rivals.

| Company | Business model | Gate compatibility | Incentive to embrace | What they gain |
|---------+---------------+-------------------+---------------------+----------------|
| Apple | Hardware (iPhone, Mac) + services (iCloud, App Store) | High — privacy is their brand, Secure Enclave is already a proto-gate | Strong — verification structurally disadvantages Meta and Google, Apple's main competitors in services and AR | New hardware differentiation ("the verified iPhone"), new services (iCloud PDS, gate appliance), brand reinforcement |
| Microsoft | Enterprise software (Office, Azure) + OS (Windows) | Medium — Azure can host gates, but Windows telemetry conflicts | Moderate — enterprise clients want verification, but Windows surveillance revenue is small | Azure differentiation for regulated industries, Office integration with gate proofs |
| Cloudflare | Edge infrastructure (CDN, DNS, security) | High — already privacy-forward, Workers platform could host gate logic | Strong — gate-compatible edge services are a new product category | New revenue from gate relay and verification edge services |
| Samsung | Hardware (phones, TVs, appliances) | Medium — Android allows gate installation, Knox security platform | Moderate — can differentiate Galaxy as "gate-compatible Android" | Hardware differentiation, B2B enterprise sales |
| IBM | Enterprise services + consulting | High — compliance is their core market | Strong — gate rule consulting is a natural service line | New consulting practice, mainframe integration |
| Amazon/AWS | Cloud infrastructure + advertising + retail | Mixed — AWS can host gates, but Amazon advertising depends on data | Conflicted — AWS division may embrace, advertising division fights | AWS can capture gate compute revenue but loses advertising data access |

**The Apple scenario as a case study:**

Apple's incentive to embrace the protocol is the strongest of any major
tech company because:

1. Apple's revenue does not depend on surveillance — 80%+ comes from
   hardware and services that work identically with or without data
   extraction. The protocol's privacy properties cost Apple nothing.
2. Apple's brand is built on privacy ("what happens on your iPhone
   stays on your iPhone"). Integrating a DID-based identity system and
   gate-compatible Secure Enclave makes this claim verifiable rather
   than aspirational.
3. Apple's main growth competitors (Meta in AR/VR, Google in services
   and AI) are structurally threatened by verification. Meta's entire
   business model collapses if advertising cannot extract user data.
   Google's AI advantage depends on unrestricted data access. The
   protocol harms them more than it harms Apple.
4. Apple already has the hardware foundation: the Secure Enclave, M-series
   chips with隔离 zones, and iCloud Keychain are all proto-gate
   infrastructure. Shipping a gate-compatible iPhone would be a
   feature update, not a new product.

Apple's optimal strategy: embed the protocol's DID system into Apple
ID, ship gate-compatible hardware in the next iPhone generation,
position iCloud as a PDS hosting service, and market the entire stack
as "privacy you can prove." This would instantly put gate capability
in 1B+ existing devices and create a distribution channel no other
protocol project has ever achieved.

**The strategic consequence for the five games:**

If a major compatible company (Apple, Cloudflare, or IBM) embraces the
protocol:

- **Game 1 (enterprise adoption):** Apple's endorsement massively lowers
  G (gate cost) because enterprise buyers already have Apple hardware.
  The coordination failure breaks faster.
- **Game 3 (hyperscaler defense):** A compatible hyperscaler (Cloudflare,
  Azure) providing gate-compatible infrastructure creates an
  accommodation path for enterprise users, reducing the fight incentive.
- **Game 4 (social protocol):** Apple's distribution gives the social
  protocol a built-in user base of 1B+ devices. The cold start problem
  in every entry vector is solved by the installed base.
- **Game 5 (geopolitical):** A US-based company (Apple) embracing the
  protocol makes it harder for the US to restrict it. Apple's
  lobbying power counterbalances Meta and Google's anti-gate lobbying.

The protocol does not need any major company to embrace it — the
trajectory is viable without them. But if even one does, the timeline
shortens from decades to years.

**Sequential expansion game:**

The protocol expands through categories sequentially, not simultaneously.
Each phase builds the conditions for the next:

| Phase | Vector | Capabilities shipped | Enables next phase by |
|-------+--------+----------------------+-----------------------|
| 0 | Organized communities | Full bundle (identity + content + payments + contracts + governance) | Proving the bundle works for real coordination |
| 1 | Refugees + creators | LSAT (paywalled content), Lightning subscriptions | Building reputation graph and creator audience |
| 2 | Freelancers | SCAL stack, arbitration guilds | Reaching contract marketplace critical mass |
| 3 | Institution crossover | Enterprise-grade verification | Leveraging network for institutional products |

This is not a choice between vectors — it is a sequence where each
phase unlocks the next. The protocol can fail at any stage if the
current vector never reaches critical mass.

**How the social game interacts with the institutional games:**

The social and institutional games are independent in their causal logic
but connected through resource flows:

- **Institutional → Social:** Compliance revenue would accelerate the
  social protocol's development timeline, but the architecture is
  self-developing after Stage 2 — the bootstrap loop means engineering
  does not depend on external funding. Revenue is amplificatory, not
  necessary.
- **Social → Institutional (late stage):** At Phase 3+, the social
  network's users become the distribution channel for institutional
  products. Institutional verification tools sell as fulfillment orders
  to a network that already exists.
- **Social → Institutional (ongoing):** Social contract disputes generate
  real-world edge cases for the collective regression suite, making the
  certification more valuable for institutional buyers.

**Asymmetric coupling:** The two sides are largely independent in both
directions. Institutional success can exist without social success
(compliance sales work with zero social users). Social success can exist
without institutional success — the social protocol reaches critical
mass through pure platform competition, and its development is not
blocked by lack of institutional revenue because the architecture is
self-developing after Stage 2 (the agent writes its own code; the
bootstrap loop makes engineering self-sustaining). The coupling is
amplificatory, not foundational: each side's success makes the other
more valuable, but neither depends on the other to survive.

* Game 5: Geopolitical Competition

Nation states do not act in isolation. Each regulator's choice (Game 2)
is shaped by what other states are doing, creating a strategic game
between major geopolitical blocs. This game determines whether the
protocol remains unified or fragments into incompatible networks.

**The players:**

| Player | Surveillance dependence | Regulatory sophistication | Gate incentive | Key constraint |
|--------+------------------------+--------------------------+----------------+----------------|
| EU | Low (GDPR limits bulk collection) | High (AI Act, NIS2, eIDAS) | Strong — gates solve enforcement at scale | Must maintain coherence across 27 member states |
| US | High (NSA, FBI surveillance apparatus) | Fragmented (sectoral agencies, no unified AI law) | Conflicted — financial regulators want verification, intelligence community wants visibility | Tech incumbents (Meta, Google) lobby against; surveillance state fights gates structurally |
| China | Total (social credit system, mass surveillance) | Top-down (state dictates standards) | Negative — gates block surveillance entirely | Would develop state-controlled "verified" alternative with backdoors |
| UK | Medium (Investigatory Powers Act) | Medium (emerging AI safety framework) | Positive — post-Brexit wants regulatory leadership | Must balance US alliance with EU regulatory alignment |
| India | Medium (growing surveillance infrastructure) | Developing (DPDP Act 2023) | Positive — leapfrog to verification without legacy audit industry | Wants sovereignty; may resist a standard defined by EU or US |

**Strategy sets:**

Each state chooses one of:
- **Promote:** Encode gate rules, subsidize adoption, advocate as international standard
- **Tolerate:** Allow gate instances but don't mandate; let the market decide
- **Restrict:** Ban or backdoor gate instances; require state-approved verification
- **Compete:** Develop an alternative state-controlled verification standard

**Payoff parameters for state i:**

Let:
- S_i = surveillance-value loss from gate adoption (0 for EU, high for China)
- E_i = enforcement benefit from automated regulatory compliance
- T_i = trade dependency on states with incompatible stances
- A_i = alliance alignment pressure (cost of diverging from key allies)
- L_i = legitimacy gain/loss from gate stance — includes not just public
  opinion and human rights reputation, but the specific cost of being
  seen to block free speech and association infrastructure. This
  parameter grows with the social protocol's adoption in free-world
  jurisdictions, because each new user in a free country makes the
  ban's censorship intent more visible to citizens in restricted
  countries.
- K_i = first-mover advantage if state defines the standard (tech industry
  attracted, standard-setting fees, geopolitical influence)

**Payoff function:**

U_i(stance) = αE_i - βS_i(stance) - γT_i(stance) + δA_i(stance) + εL_i + φK_i(stance)

**Key equilibria by scenario:**

1. **EU promotes, US tolerates, China restricts — the fragmentation
   equilibrium.** EU encodes gates as the AI Act enforcement mechanism.
   US allows gate instances but does not mandate them (compromise between
   financial regulators wanting verification and intelligence community
   opposing it). China bans all gate instances that it cannot surveil and
   develops a backdoored "verified" alternative. Global enterprises
   operate dual configurations: EU-compliant gate instances for European
   operations, non-verified for China, mixed for US. The protocol
   fragments into two tiers: "free world" protocol and "China-controlled"
   alternative, but the free-world protocol remains functionally unified
   because the EU and US standards are technically compatible (even if US
   doesn't mandate enforcement).

2. **The EU also restricts — the worst case.** If the EU, under pressure
   from member states with surveillance priorities, mandates data
   localization or backdoors for gate instances, the protocol's core
   value proposition (verification cannot be bypassed by the state) is
   broken. The EU and US would converge on a weakened standard that
   preserves some surveillance capability, and the protocol loses its
   structural advantage. This scenario requires a significant shift in
   EU governance toward surveillance — possible but unlikely given GDPR
   and current AI Act trajectory.

3. **EU promotes, US promotes, China isolates — the best case.** The US
   resolves its internal conflict in favor of financial regulators and AI
   safety advocates (triggered by a high-profile AI harm event that makes
   gates politically necessary). US and EU jointly promote gates as the
   international standard. China isolates itself with an incompatible
   alternative. Global enterprises standardize on the EU-US protocol.
   This is the scenario that maximizes the protocol's growth trajectory.

4. **Cold War standoff — US restricts, EU tolerates.** The US
   intelligence community wins the internal debate and bans unrestricted
   gate instances (requiring backdoors for law enforcement). The EU
   tolerates but does not mandate. The protocol survives in the EU but
   loses the largest enterprise market (US-based global firms). Growth
   slows significantly but does not stop — the EU market alone (450M
   people, major regulated industries) is enough to sustain the
   trajectory.

**The race dynamic:**

The first major jurisdiction to encode gates as a regulatory standard
gains first-mover advantage (K_i). The standard they define — rule
format, proof requirements, certification process — becomes the
default. Late adopters must either adopt the existing standard (ceding
some sovereignty) or fragment the market. This creates a race:

- **EU is best positioned.** AI Act enforcement begins 2026-2027. The EU
  can encode gate rules as the compliance mechanism for high-risk AI
  systems. If they do this before the US resolves its internal conflict,
  the EU defines the standard and the US becomes a standard-taker.
- **US could catch up.** A single AI harm event that causes measurable
  financial damage ($1B+) could shift US political will. If the US then
  moves faster than the EU through executive action, they could define
  the standard instead.
- **China cannot win the race but can fragment it.** China cannot adopt
  the protocol without losing surveillance. But they can fund an
  alternative standard, pay developing countries to adopt it, and create
  a bifurcated global market.

**How Game 5 interacts with Games 1-4:**

- Game 5 determines the parameter S (surveillance-value loss) for each
  regulator in Game 2. A state that chooses "restrict" in Game 5 has
  high S in Game 2, shifting the regulator's equilibrium.
- Game 5 determines whether enterprises face compatible or incompatible
  requirements across jurisdictions, which affects their Game 1 payoff
  (N = network effect benefit is lower if the protocol fragments).
- Game 5 outcome affects the hyperscaler's S_h (if gates are legal in
  the US, AWS can accommodate; if banned, they must fight).
- Game 4 (social) is the most resilient to geopolitical fragmentation —
  individual users can still participate in the protocol regardless of
  their state's stance, as long as the relay network operates across
  borders.

**The stability claim for Game 5:**

The protocol has a structural defense against geopolitical fragmentation
that no previous decentralized technology had: the gate itself.
Verification is self-authenticating — a gate instance in a non-EU
jurisdiction can still prove compliance with EU gate standards. The
protocol does not need states to agree; it only needs each state to
allow at least one path for verification to exist. A state that bans
gates loses the economic benefits of verification (lower compliance
costs, AI safety, efficient contracts) without gaining enforcement
capability (because banned gates will exist in other jurisdictions
anyway). This creates a ratchet: once verification is adopted by enough
economically significant jurisdictions, non-adopting states face a
growing competitive disadvantage that eventually exceeds their
surveillance benefit.

* The combined game: why the trajectory is stable

The five games interact through their parameters:

1. Game 1 (enterprise adoption) can have two equilibria, but Game 2
   (regulator commitment) and the insurance loop (Game 3 consequence)
   both break the coordination failure by changing the dominant strategy
   for the first mover.

2. Game 2 (regulator commitment) has a unique equilibrium for
   democratic regulators: encode gates. The only failure mode is
   authoritarian suppression or capture, both of which are
   jurisdiction-specific — they fragment the protocol but don't stop it
   globally.

3. Game 3 (hyperscaler defense) has a threshold that depends on
   adoption level. Since Games 1 and 2 push adoption past that
   threshold, the hyperscaler's optimal strategy shifts from fight to
   accommodate. They do not defect back to fight once adoption is high
   because the sunk cost of fighting has already been incurred and the
   revenue base has already shifted.

4. Game 4 (social protocol) is structurally independent of the
   institutional games — it does not depend on any enterprise outcome.
   Its trajectory is driven by the asymmetric bundle advantage: 20+
   incumbents, zero of whom can match the full protocol. The binding
   constraint is not competition but status quo inertia — the protocol
   must demonstrate one use case that is dramatically better, then
   expand from there.

5. Game 5 (geopolitical competition) determines whether the protocol
   fragments or remains unified. The default equilibrium is
   fragmentation between surveillance and rule-of-law blocs, but the
   free-world protocol remains functionally unified because
   verification is self-authenticating — states don't need to agree.
   The ratchet dynamic means non-adopting states face growing
   competitive disadvantage over time.

**The stability claim:**

The only way the trajectory fails is if one of the following parameter
conditions holds:

- **Game 1 parameter failure:** G + N > C + R even with insurance
  differentiation. This would require gate costs to be higher than
  compliance savings — unlikely given the 4-20x cost ratio, but
  possible if the gate implementation is expensive.
- **Game 2 parameter failure:** εS > α(E_gate - E_paper) + βB for all
  major regulators. This requires a coordinated shift toward high-
  surveillance governance across all regulated markets — possible but
  requires a global authoritarian turn.
- **Game 3 parameter failure:** S_h ≈ 1 and C_h < P_h + kR_h — the
  hyperscaler both depends entirely on unrestricted data and can
  successfully co-opt the gate provider before the ecosystem reaches
  critical mass. The AGPL license and open community are the defense
  against this.
- **Game 4 parameter failure:** every entry vector fails to demonstrate
  a use case dramatically better than the status quo, OR the social
  protocol never achieves critical mass in any vector. This is possible
  if the bundle is too complex for even organized communities (the
  lowest-friction vector).
- **Game 5 parameter failure:** the EU shifts toward surveillance
  governance, adopting data localization or backdoor mandates for gate
  instances. This breaks the protocol's core value proposition in its
  most natural market. This scenario requires a significant shift in EU
  political trajectory — possible but unlikely given GDPR and current
  AI Act direction.

If none of these parameter conditions hold, the unique equilibrium
across all five games is: gates win on the institutional side, and the
protocol has a viable trajectory on the social side via at least one
entry vector. The two sides reinforce each other but do not depend on
each other for survival — each can succeed independently, and together
they compound.

* Failure modes as parameter shifts

The failure modes from the [[id:6d2e3f4a-5b6c-7d8e-9f0a-1b2c3d4e5f6a][Adoption]] document can now be re-expressed as
parameter changes:

- **Agent quality insufficient:** G increases (gate requires too much
  human correction, raising effective cost). If G + N > C + R, Game 1
  reaches "both maintain audit" and the regulator's encoding in Game 2
  becomes less attractive (E_gate drops because gates are unreliable).
- **State suppression:** εS dominates for all major regulators. Game 2
  equilibrium shifts to "maintain paper/ban gates" in significant
  jurisdictions. Protocol fragments.
- **Bootstrap loop stalls:** Same as agent quality — G stays high for
  too long. Game 1 fails to tip.
- **Neither adoption reaches critical mass:** Games 1 and 2 both fail —
  enterprise adoption stalls at Phase 1 (Game 1 coordination failure
  unresolved) and social protocol never achieves network effects
  (different game, not modeled here).