Skip to main contentSkip to navigation
ThisIsHowItWorks.in

Where complex ideas unfold at human pace

Primary

  • Atrium
  • Map
  • Pieces
  • Series
  • Search

Secondary

  • Archive
  • Index
  • Library
  • Fragments

Meta

  • About
  • Principles
  • Lexicon
  • Questions
  • Resources

Connect

  • Instagram
  • Discord
  1. Home
  2. /The Infrastructure of Belief
  3. /02 · Scaling Problems II — Stranger Problems
Map

Scaling Problems II — Stranger Problems


SERIES 2: SCALING PROBLEMS

Phase 2.2 — Stranger Problems: How to Trust People You'll Never See Again

Why Dunbar Mechanisms Fail at Scale

Let's revisit what worked in small groups and see where it breaks:

Mechanism 1: Personal Reputation

How it worked (small groups):

Bob cheats Alice       ↓ Alice tells everyone       ↓ Everyone knows Bob is a cheater       ↓ No one cooperates with Bob       ↓ Bob is excluded       ↓ Cheating doesn't pay

Why it fails (large groups):

Small Group (<150)Large Group (>500)
Alice knows everyoneAlice knows maybe 150 people
Word spreads to whole groupWord spreads only within Alice's network
Bob is known to everyoneBob is known only to his subgroup
Exclusion is group-wideBob just moves to a different neighborhood
Reputation is inescapableReputation is local and fragmented

The structural breakdown:

Bob cheats Alice in neighborhood A       ↓ Alice's network learns (150 people max)       ↓ Bob moves to neighborhood B       ↓ No one there knows his history       ↓ Bob cheats again       ↓ Reputation system has failed

The cheater's advantage scales with anonymity.

Mechanism 3: Reciprocal Altruism

How it worked (small groups):

I help you today       ↓ We'll interact again (high probability)       ↓ You help me tomorrow       ↓ We both benefit over time

Why it fails (large groups):

The one-shot interaction problem:

Stranger A helps Stranger B       ↓ Will they ever meet again? (Probably not)       ↓ B has no incentive to reciprocate       ↓ B defects       ↓ A learns not to help strangers       ↓ Cooperation collapses

The math:

Group SizeProbability of Re-encounterReciprocity Viability
50 people~95%✅ Stable
150 people~60%⚠️ Weakening
500 people~20%❌ Failing
5,000 people~2%❌ Collapsed

Without repeated interaction, reciprocity doesn't work.

Mechanism 5: Leveling Mechanisms

How it worked (small groups):

  • Mockery and ridicule of would-be dominators
  • Everyone participates in collective suppression
  • Assassination as last resort (entire group cooperates)

Why it fails (large groups):

The coordination problem:

Bob tries to dominate       ↓ Some people notice and mock him       ↓ But many don't know Bob       ↓ And those who do can't coordinate response       ↓ Bob builds power base in one subgroup       ↓ By the time the whole community notices, he's too strong       ↓ Leveling has failed

The structural outcome: In large groups, ambitious individuals can build power before collective suppression can coordinate.

This is why hierarchy becomes permanent in agricultural societies.

The Cooperation Dilemma in Anonymous Settings

Consider this scenario:

The Traveling Merchant Problem:

Merchant A travels to distant village
Merchant B (local) offers trade
Neither will ever meet again
Each could cheat the other

The payoff matrix:

B HonestB Cheats
A HonestBoth gain (+5, +5)A loses, B wins big (-10, +15)
A CheatsA wins big, B loses (+15, -10)Both lose (-5, -5)

The rational choice (one-shot game): Cheat

The outcome: No trade happens (both expect cheating)

The problem: There are genuine gains from trade (+5, +5) that neither can access because trust is impossible.

What's needed: A mechanism that makes cheating more costly than honesty.

Solution 2: Standardization

The problem without standards:

Farmer A offers "a basket of wheat"
Merchant B: "Is that a large basket or small basket?"
Farmer A: (uses small basket)
Merchant B feels cheated
No repeat business

The solution:

Authority sets standard weights and measures       ↓ Baskets must be regulation size       ↓ Inspectors check compliance       ↓ Violators punished       ↓ Everyone knows "one basket" = specific amount       ↓ Trade becomes reliable

What this requires:

  • Central authority to set standards
  • Enforcement to maintain standards
  • Punishment for violators

Early evidence:

  • Mesopotamian grain measures (standardized pots)
  • Egyptian cubit rods (standard length)
  • Indus Valley weights (uniform across region)
  • Chinese weights and measures (Qin dynasty standardization)

Why this is partial:

  • Only standardizes measurement, not honesty
  • Requires effective enforcement (expensive)
  • Someone must police the police

Solution 4: Kin and Ethnic Networks

The logic:

Can't trust strangers       ↓ But can trust people from my group       ↓ Shared identity = shared incentives       ↓ Cheating your own group = severe social punishment

Historical examples:

  • Jewish merchant networks (medieval Europe)
  • Armenian trading networks (Silk Road)
  • Chinese diaspora networks (Southeast Asia)
  • Gujarati traders (Indian Ocean)
  • Lebanese merchant families (Mediterranean)

Why this works:

  • Reputation travels within ethnic/kin network
  • Exclusion from network is catastrophic (can't just join another)
  • Shared norms and enforcement mechanisms
  • Multi-generational (reputation lasts beyond one lifetime)

Why this is partial:

  • Only works within the network
  • Creates inter-group distrust
  • Can enable monopolies and exclusion
  • Breaks down if network grows too large

What None of These Solutions Fully Solve

All partial solutions share common limitations:

Problem 1: The Enforcement Gap

Someone must enforce contracts, standards, and punishments.

But:

  • Enforcers are also strangers (can't trust them either)
  • Enforcers have power (can abuse it)
  • Who enforces the enforcers? (infinite regress)

This is the principal-agent problem:

Principal (ruler/community) hires Agent (enforcer)       ↓ Agent supposed to serve Principal's interests       ↓ But Agent has own interests (enrichment, power)       ↓ Agent can exploit position       ↓ Principal can't monitor Agent perfectly       ↓ Agent behavior diverges from Principal's interest

Example:

  • King appoints tax collector
  • Tax collector supposed to collect 10% for king
  • Tax collector collects 15%, gives king 10%, keeps 5%
  • King can't audit every transaction
  • Tax collector enriches himself

This problem appears everywhere:

  • Police (supposed to protect, can extort)
  • Judges (supposed to judge fairly, can take bribes)
  • Officials (supposed to serve, can embezzle)
  • Bureaucrats (supposed to implement, can obstruct)

What's needed: Some way to make agents act in principals' interests even when principals can't watch them.

Problem 3: The Collective Action Problem

The logic:

Group faces threat (famine, invasion, flood)       ↓ Coordinated response needed       ↓ But coordinating 1,000+ people is hard       ↓ Who decides what to do?       ↓ How to ensure compliance?       ↓ How to prevent panic/defection?

Small group solution:

  • Discussion until consensus
  • Everyone knows the plan
  • Social pressure ensures compliance

Large group problem:

  • Discussion doesn't scale (too many voices)
  • Information doesn't spread quickly
  • No consensus mechanism
  • Compliance can't be monitored
  • Defection cascades (if some flee, others follow)

Example: Flood Defense

  • River about to flood
  • Must build levee quickly
  • Need 500 workers, coordinated effort
  • Who directs the work?
  • What if some neighborhoods refuse?
  • What if panic spreads?

What's needed: Command structure + rapid information flow + enforcement

What This Explains

This framework clarifies:

Why cities are fundamentally different from villages:

  • Different trust mechanisms
  • Different enforcement needs
  • Different coordination challenges

Why markets required institutional support:

  • Can't function on reputation alone at scale
  • Need enforcement, standards, contract law
  • Require dispute resolution mechanisms

Why "social capital" breaks down in large groups:

  • Thick trust doesn't scale
  • Thin trust requires institutions
  • Institutions require enforcement

Why anonymous societies feel less cohesive:

  • Mechanisms designed for small groups don't work
  • Impersonal institutions feel cold
  • Nostalgia for "community" = nostalgia for thick trust

Why fraud increases with city size:

  • Anonymity enables deception
  • Reputation doesn't follow you
  • Enforcement is imperfect

The Limits of This Analysis

What this explains:

  • Why small-group mechanisms fail at scale
  • The specific trust problems that emerge
  • The coordination gaps that appear
  • What functions institutions must serve

What this doesn't explain:

  • The specific forms institutions took
  • Cultural variation in responses
  • Why some solutions worked better than others
  • The role of ideas and belief systems

What this doesn't evaluate:

  • Whether large-scale societies are better or worse
  • Whether anonymity is good or bad
  • Whether we should return to small groups
  • Whether current institutions are optimal

We're describing problems, not prescribing solutions.

What's Next

We've established: 1. Agriculture created surplus and hierarchy 2. Scale broke small-group coordination mechanisms 3. Stranger problems require formal institutions

But institutions face their own problems:

Problem 1: Institutions are made of people. People have incentives to abuse institutional power. How do you prevent institutional collapse from corruption?

Problem 2: You can't coerce millions of people simultaneously. Dungeons are expensive. Surveillance is limited. What makes people obey without constant force?

Problem 3: Bureaucracies need to coordinate. But bureaucrats are strangers to each other. How do large organizations function internally?

Problem 4: Information must flow upward (reports) and downward (commands). But information can be distorted, delayed, or lost. How do empires maintain information flow?

Problem 5: Why do some institutional forms spread while others collapse?

These are institutional formation problems.

Next question: How do you build organizations that don't immediately fall apart from internal contradictions?

Next explainer: "Big Men, Chiefs, and Kings: The Gradient of Formalization"

(Beginning Series 3: Institutional Formation)


PreviousScaling Problems I — The Surplus TrapNextScaling Problems III — Violence Specialization

The Suitcase

Take this piece with you—works offline, no internet needed.

↩ Return to The Infrastructure of Belief⌂ Ascend to The Observatory