Cooperative Game Theory

From BloomWiki
Revision as of 14:47, 23 April 2026 by Wordpad (talk | contribs) (BloomWiki: Cooperative Game Theory)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

How to read this page: This article maps the topic from beginner to expert across six levels � Remembering, Understanding, Applying, Analyzing, Evaluating, and Creating. Scan the headings to see the full scope, then read from wherever your knowledge starts to feel uncertain. Learn more about how BloomWiki works ?

Cooperative Game Theory is the study of how groups of people can work together and fairly "Split the Pie" once they succeed. While standard game theory focuses on "Me vs. You," cooperative game theory focuses on "Us." It asks: "If we team up to make $1,000, how much does each person deserve?" By using concepts like the "Shapley Value," we can find mathematical ways to reward people for their unique contributions, ensuring that coalitions stay together and that everyone feels they are being treated fairly. It is the math of teamwork, mergers, and social justice.

Remembering

  • Cooperative Game Theory — A branch of game theory where players can form binding agreements and work in coalitions.
  • Coalition — A group of players who agree to work together to achieve a better outcome than they could alone.
  • The Core — The set of all possible ways to split the "Grand Prize" such that no sub-group has an incentive to break away.
  • Shapley Value — A mathematical formula for assigning a fair "Payoff" to each player based on their "Average Marginal Contribution."
  • Nucleolus — A specific solution that minimizes the "Unhappiness" of the most dissatisfied coalition.
  • Transferable Utility — The assumption that players can "Give" money or value to each other to balance out the rewards.
  • Characteristic Function — A mathematical rule that defines the "Value" of every possible group (coalition) in the game.
  • Lloyd Shapley — The Nobel Prize winner who defined the primary solutions for cooperative games.
  • Grand Coalition — The coalition that includes every single player in the game.

Understanding

Cooperative game theory is understood through Fairness and Coalition Stability.

1. The Power of the Group: In a cooperative game, the "Value" is not found in individual moves, but in the "Synergy" of the group.

  • If Alice can bake 10 cakes alone and Bob can bake 10 cakes alone, but together they can bake 30 cakes, the "Coalition Value" is 30.
  • The question is: who gets the extra 10 cakes?

2. The Shapley Value (The Fair Split): Shapley's genius was to look at every possible order in which people could join the team.

  • If a team already has a great coder, a second coder might not add much value.
  • But if the team has no coder, the first coder adds "Infinite" value.
  • The Shapley Value averages these "Joining Bonuses" to find a number that truly reflects how "Essential" you are.

3. The Core (The Anti-Revolt Rule): A split is only "In the Core" if it is so fair that no group says, "Hey, if we three left this team and started our own, we'd make more money." If a split is in the core, the group is stable.

Grand Coalition vs. Fragmentation: Cooperative game theory tries to find the conditions under which everyone stays in the "Grand Coalition" (like a unified government or a massive merger) rather than splitting into smaller, warring tribes.

Applying

Modeling 'The Shapley Value' (Splitting a reward fairly): <syntaxhighlight lang="python"> def calculate_shapley_simplified(val_a, val_b, val_ab):

   """
   How much should A and B get?
   A's value = Average of (A alone) and (Team - B's contribution)
   """
   # A joins first: adds val_a
   # B joins first: A adds (val_ab - val_b)
   shapley_a = (val_a + (val_ab - val_b)) / 2
   shapley_b = (val_b + (val_ab - val_a)) / 2
   
   return {"A": shapley_a, "B": shapley_b}
  1. Alice makes $10. Bob makes $10. Together they make $50.
  2. Without Bob, Alice only adds $10. With Bob, she adds $40 ($50-$10).
  3. Shapley: (10 + 40) / 2 = $25 each.

print(calculate_shapley_simplified(10, 10, 50)) </syntaxhighlight>

Cooperative Landmarks
The Shapley Value (1953) → The most famous formula in the field, now used to decide everything from "Who pays for a shared runway" to "How to rank the importance of different genes."
Gale-Shapley Algorithm → Used to match medical students to hospitals and kidneys to patients in a way that is "Stable" (no two people would prefer to be matched with each other instead).
The EU Voting Power → Using game theory to calculate how much "Say" each country has in the European Union based on their population.
Carbon Credit Allocations → Deciding how many "Pollution Rights" each company deserves based on their historical output and their efforts to clean up.

Analyzing

Non-Cooperative vs. Cooperative
Feature Non-Cooperative (Nash) Cooperative (Shapley)
Basic Unit The Individual The Coalition (Group)
Question "How do I win?" "How do we split the win?"
Key Mechanism Competition / Bluffing Agreements / Fairness
Result Stable Equilibria Fair Allocations

The Concept of "Marginal Contribution": Analyzing what a group "Loses" if you leave. This is the ultimate measure of your value in a cooperative game. If the group's profit drops by $100 when you quit, you have high leverage.

Evaluating

Evaluating cooperative game theory:

  1. Fairness vs. Power: Should the person who worked the hardest get the most, or the person who is most "Essential" (even if they worked less)?
  2. Complexity: Calculating the Shapley Value for 100 people requires checking 100! (a massive number) of combinations. Is the math "Too Hard" for real life?
  3. Selfishness: Does it assume everyone is a "Rational Actor" who only cares about money? (In real life, people often accept "Unfair" splits to keep friends).
  4. Stability: Can a group ever be 100% stable, or is there always someone who wants more?

Creating

Future Frontiers:

  1. AI Profit Sharing: Using Shapley values to pay "Data Contributors" for their specific contribution to a massive AI model like ChatGPT.
  2. Decentralized Organizations (DAOs): Building software that automatically splits the revenue of a company among all "Gig Workers" using cooperative game theory.
  3. Climate Change Treaties: Designing treaties where countries are "Fairly" rewarded for reducing emissions, making the Grand Coalition of the whole Earth stable.
  4. Fair Algorithmic Ads: Deciding how much each "Website" in a chain of clicks deserves from an ad sale.