Decision-Making Under Uncertainty

Author: Vitaly - mr. Koteo (Brisbane Mafia Club)

Mafia is a game of incomplete information.

You never know the full truth. You never act with certainty. Every decision is made under ambiguity, pressure, and limited time.

This chapter explains how to make correct decisions when certainty is impossible.

It is the bridge between:

  • mathematical thinking (previous chapters)

  • real gameplay under pressure


Risk vs Reward

Decision-making in Sports Mafia is not about being right. It is about making the best possible decision with the information available.

A correct decision can lead to a loss. An incorrect decision can win the game.

What matters is not the outcome. What matters is whether the decision was logically justified at the moment it was made.


Decision Quality vs Outcome

Players are not “Red” or “Dark” in your mind. They are probabilities.

Example:

  • Player #4 → more likely Dark

  • Player #7 → less likely Dark

You vote #4. #4 flips Red.

This does not make your decision wrong.

Decisions are evaluated before the reveal, not after.


Risk Depends on Structure

Risk is not just probability — it is also game state.

The same decision has different risk depending on player count.

  • At 9 players → wrong vote is recoverable

  • At 8 players → wrong vote may lead to immediate loss (via 7 → night → 6)

Critical rounds amplify risk.

A 60/40 decision:

  • acceptable at 9

  • dangerous at 8


Expected Value (Practical Form)

You do not calculate EV (Expected Value) explicitly at the table.

But you act according to it:

Choose the action that maximises your team’s chance to win.

This includes:

  • eliminating the most likely Dark

  • preserving likely Red players

  • avoiding moves that lead to parity loss


Risk vs Consequence

Not all risks are equal.

Two decisions may have similar probability, but different consequences.

Example:

  • Voting wrong at 9 → low consequence

  • Voting wrong at 8 → catastrophic consequence

Risk must always be evaluated together with consequence.


Information vs Elimination

Not all decisions aim to eliminate correctly.

Some aim to generate information.

Examples:

  • free voting

  • creating splits

  • delayed voting

  • forcing nominations

These actions:

  • may not remove a Dark immediately

  • but improve future decisions

Information EV vs Elimination EV

Strong Red play balances both.


Red vs Dark Risk Profiles

Red Team

  • minimise long-term error

  • maximise information

  • avoid catastrophic mistakes

Dark Team

  • increase uncertainty

  • force critical errors

  • accelerate losing transitions for Red


Dominated Decisions

Some decisions are always worse than alternatives.

Examples:

  • voting a less suspicious player when a stronger candidate exists

  • following votes without reasoning

  • avoiding responsibility in critical rounds

These should be eliminated entirely.


Regret Minimisation

When uncertain, ask:

“Can I justify this decision after the reveal?”

A correct decision should:

  • be explainable

  • match your reads

  • fit the structure of the game


Core Rule

Choose the decision that maximises team advantage while minimising structural risk.


Decision Trees

Every decision creates multiple possible futures.

This can be visualised as a decision tree.

In theory, trees are large and complex. In practice, strong players simplify them.


Two-Level Thinking

Instead of full trees, use:

  • Level 1: What happens immediately after the decision

  • Level 2: What happens after the next night/day cycle

That is enough for practical play.


Why Full Trees Are Impractical

A full tree includes:

  • all players

  • all votes

  • all future interactions

This is impossible to compute in real time.

Strong players reduce trees to key forks only.


World-Based Thinking

Instead of trees, think in:

  • World A → assumption 1

  • World B → assumption 2

Then test:

  • does behaviour match this world?


Decision Rule

Choose the branch that:

  • is most likely

  • leads to a better structure if correct

  • is least damaging if wrong


Information Paths

Every action generates information.

Strong players think not only:

“What happens if I’m right?”

But also:

“What do I learn from this action?”


Information Value

Information is a resource.

Some actions increase it:

  • free voting

  • forcing nominations

  • delaying decisions

  • creating splits


Red vs Dark Objectives

Red

  • play to learn

  • increase clarity over time

Dark

  • reduce clarity

  • create noise

  • force premature conclusions


Information vs Speed

Fast decisions:

  • reduce information

  • increase risk of error

Slow decisions:

  • increase information

  • but may allow Dark to manipulate

Balance is required.


Information Traps

Dark players can create false information.

Examples:

  • staged disagreements

  • fake conflicts

  • controlled vote patterns

These produce:

  • information that looks real

  • but leads to incorrect conclusions


Connection to Patterns

Information is not raw — it must be interpreted.

Patterns:

  • convert actions into meaning

  • help distinguish signal from noise


Core Principle

The best decision is not always the fastest or most confident — it is the one that improves future decisions.


Psychological vs Mathematical Choices

Mafia is played by humans, not equations.

Even perfect logic can fail if it ignores human behaviour.


When Mathematics Should Dominate

  • critical rounds (8, 7 players)

  • parity risks

  • sheriff-based decisions

  • clear probability advantages

In these cases:

Logic must override psychology.


When Psychology Matters More

  • early game

  • unclear structures

In these cases:

The table’s behaviour affects outcomes more than pure probability.


Table Influence and Leadership

Players do not act independently.

Some players:

  • lead decisions

  • influence votes

  • shape narratives

Ignoring this is a mistake.

Example:

  • correct vote mathematically

  • but table will not follow → decision fails


Common Psychological Factors

  • fear of being wrong

  • avoiding conflict

  • trusting confident players


Red vs Dark Use of Psychology

Red

  • stabilise the table

  • reduce chaos

  • guide decisions logically

Dark

  • increase chaos

  • create doubt

  • exploit confidence gaps


Core Rule

A correct decision that the table will not follow is not a complete decision.

You must consider:

  • not only what is correct

  • but what can be executed


Mixing Intuition Into a Logic-Driven Game

Intuition exists — but it must be understood correctly.


What Intuition Really Is

Intuition is not guessing.

It is:

fast pattern recognition based on experience

Examples:

  • tone changes

  • hesitation

  • unnatural reactions

  • emotional inconsistencies


Beginner vs Advanced Use

Beginners:

  • rely heavily on intuition

  • often without structure

Strong players:

  • use logic first

  • use intuition when logic is incomplete


Intuition Must Be Verbalised

As a Red player:

Intuition should be expressed, not hidden.

Example:

  • “I don’t have structural proof, but behaviour feels inconsistent”

This:

  • adds information to the table

  • allows others to evaluate it


Calibration of Intuition

Strong players:

  • track when intuition is correct

  • adjust trust in their reads over time


Danger: Overconfidence (“Power Play” 💪)

One of the most dangerous states:

Believing you have solved the game completely

This leads to:

  • ignoring contradictions

  • forcing incorrect decisions

  • collapsing team structure


Balanced Use

Correct approach:

  • use logic as foundation

  • use intuition as a supplement

  • never allow intuition to override clear structure


Core Rule

Intuition is useful only when it supports logic — not when it replaces it.


Chapter Summary

Decision-making under uncertainty requires:

  • probabilistic thinking

  • structural awareness

  • controlled risk-taking

  • information management

  • psychological understanding

  • disciplined use of intuition

The strongest players are not those who guess correctly — but those who consistently make the best possible decisions in unclear situations.

Last updated