UNIT II

An Introduction to Voting
The Game of Trust - Nicky Case's Interactive Essay
• Nicky Case's interactive essay "The Game of Trust" provides an insightful and
engaging exploration of voting systems and game theory.

• It emphasizes how trust plays a central role in making collective decisions and how
different voting methods influence the outcomes of group decision-making.

• This interactive essay is an excellent resource for

 understanding the complexities of voting mechanisms and
 the strategic decisions involved in a group setting.
Key Concepts in "The Game of Trust"
Voting and Decision-Making
• Voting is one of the most common methods of making group decisions. It's often
used in democracies, elections, and committees.

• In its simplest form, voting is a way for individuals to express their preferences over
a set of options or candidates.

Note: However, voting is not as straightforward as it might seem, because there are
many ways to aggregate these preferences into a collective decision.
Types of Voting Systems

 Plurality Voting

 Ranked Voting

 Borda Voting

 Approval Voting
Types of Voting Systems
 The essay introduces various types of voting systems, each of which has different impacts on the fairness,
accuracy, and strategic behavior in the decision-making process.

 Plurality Voting (First-Past-The-Post)

• How it works: Each voter selects one candidate, and the candidate with the most votes wins.

Simple to understand and easy to implement.

Can lead to strategic voting (where voters may vote against their true preferences to avoid undesirable
outcomes).

It also leads to the spoiler effect, where third-party candidates can split the vote and help elect the
less-preferred candidate.
Plurality Voting (First-Past-The-Post - FPTP)
Example: U.S. Presidential Elections

In the U.S., each state votes for a candidate, and the one with the most votes in a
state wins all its electoral votes (except in Maine & Nebraska).

• Issue: Spoiler Effect

• In 2000, Ralph Nader (Green Party) received ~97,000 votes in Florida.
• Many believe that if Nader had not run, most of his voters would have chosen Al Gore
(Democratic Party), which could have prevented George W. Bush (Republican) from winning.
• This shows how a third-party candidate can split votes and affect the final outcome.

💡 Note: Plurality voting is simple but can lead to unfair results due to the spoiler
effect.
Vote Splitting in Plurality (FPTP) Voting
What is Vote Splitting?
• Vote splitting occurs when two or more similar candidates divide votes among themselves, allowing
another candidate (who may be less popular overall) to win.
Example: Vote Splitting in India (Hypothetical Case in Telangana)
• Consider a constituency election where three candidates are contesting:
• Candidate A (TRS)
• Candidate B (Congress)
• Candidate A wins, even though 60% of
• Candidate C (Independent, but ideologically similar to Congress)voters preferred someone else.

Candidate Votes Received Vote Share (%)

• If Candidate C had not contested, B
A (TRS) 40,000 40% could have won with 60% of votes.
B (Congress) 35,000 35%

C (Independent) 25,000 25% • This is a classic case of vote splitting,

where opposition votes are divided,
leading to an unintended winner.
Ranked Voting (Instant-Runoff Voting)
How it works:

• Voters rank candidates in order of preference.

• If no candidate gets a majority, the lowest-ranked candidate is eliminated,
• and votes are redistributed based on the next preference.

More nuanced and reduces the spoiler effect.

More complex, and voters might strategically rank candidates to avoid a less-
preferred outcome.
Ranked Voting (Instant-Runoff Voting - IRV)
Example: Australian House of Representatives Elections

 Voters rank candidates in order of preference.

 If no candidate gets more than 50% of first-choice votes,
 the candidate with the fewest votes is eliminated, and their votes are transferred to
their next preference.
40 people: A > B > C
Hypothetical Example: 35 people: B > C > A
25 people: C > B > A
•100 voters rank candidates A, B, and C:

• Since no candidate has >50% (A has 40%, B has 35%, C has 25%), C is eliminated.
• The 25 votes for C are transferred to B (their next preference).
• Final Count: A = 40, B = 60 → B wins!

💡 Note: IRV prevents the spoiler effect and allows voters to rank their preferences honestly.
Borda Count

How it works:
• Voters rank candidates, and points are awarded based on the rankings (e.g., first
place gets 3 points, second place gets 2, etc.).
• The candidate with the highest total points wins.

Encourages voters to rank all candidates, reducing the chance of a vote

being wasted.

Vulnerable to manipulation if voters can anticipate how others will vote.

Borda Count
Example: Deciding a Team Outing Location
• Suppose a company votes on where to go for a team outing:
• Choices: Beach, Mountains, Amusement Park, Museum

1st place = 3 points

2nd place = 2 points
3rd place = 1 point
Total Points: 4th place = 0 points
• Beach = 3+2+1 = 6
• Mountains = 0+3+2 = 5
• Park = 2+1+3 = 6
• Museum = 1+0+0 = 1
Winner: Beach & Park (tie)
💡 Note: Borda Count allows compromise candidates to win rather than extreme
choices.
• Each team member ranks their preferences, and points are assigned.
Approval Voting

How it works:

• Voters can approve of as many candidates as they wish. The candidate with
the most approvals wins.

Simple and encourages honest voting.

It can result in a compromise candidate winning, which may not reflect

the preferences of the majority.
Approval Voting
Example: United Nations Secretary-General Selection
• In Approval Voting, each voter can approve multiple candidates.
Hypothetical Example:
• Three candidates: X, Y, Z
• 100 voters can approve one or more candidates:

Winner: Y (70 approvals)

• This allows a broadly acceptable candidate to win rather than one with extreme
support.
💡 Note: Approval voting prevents vote splitting and selects a consensus
candidate.
Conclusion
Plurality Voting (FPTP):
 Simple but suffers from the spoiler effect.

Ranked Voting (IRV):

 Prevents spoilers but is more complex.

Borda Count:
 Encourages compromise candidates.

Approval Voting:
 Helps select a broadly accepted winner.
Strategic Behavior in Voting Systems

• In each system, voters might behave strategically to influence the outcome.

• For example, in plurality voting, voters might choose the less preferred
candidate who has a better chance of winning over their least-preferred
candidate (this is known as tactical voting).

• In contrast, ranked voting systems reduce strategic behavior by encouraging

voters to rank their true preferences without fear of "wasting their vote."
The Game of Trust: The Role of Cooperation

• One of the key insights from the essay is the importance of trust in
collective decision-making.
• Voters must trust that other voters will act honestly and not
manipulate the system.
• In strategic voting systems, the game of trust refers to how much
faith each participant has in others' behavior, especially when they
know that strategic behavior might lead to less favorable outcomes
for the group.
The Paradox of Voting Systems

• The essay also touches on the paradoxes and complexities inherent in different voting
systems, such as:

• The Condorcet Paradox: When voting preferences are cyclical, and no candidate is the
winner by a majority, leading to no clear winner.

• Arrow’s Impossibility Theorem: A formal result in social choice theory which states that
no voting system can convert individual preferences into a collective decision that
satisfies certain fairness criteria (e.g., non-dictatorship, unanimity).
Why is "The Game of Trust" Important?

[Link] and Understanding: The interactive nature of the essay helps readers grasp
the real-world implications of different voting systems and how their decisions can
influence the collective outcome.

[Link] into Voting Systems: It provides an understanding of how systems can be

manipulated, and how trust among participants is essential for a fair decision-making
process.

[Link] Voting: It shows how people may strategically vote based on anticipated
outcomes, leading to distortions in the true preferences of the group.
Conclusion
• Nicky Case's interactive essay on voting systems and the game of trust
offers a deep dive into the complexities of voting mechanisms.

• By presenting different voting systems and highlighting how strategic

behavior influences outcomes, the essay encourages a deeper
understanding of fairness and social choice.

• It’s a valuable resource for anyone interested in game theory, political

science, or collective decision-making.
Arrow's Impossibility Theorem (Arrow's Theorem)
• Arrow’s Theorem is a fundamental result in social choice theory, formulated by
economist Kenneth Arrow in 1951.
• It states that no ranked voting system can satisfy all of the following fairness conditions
simultaneously when choosing among three or more candidates.
• The Three Key Fairness Conditions

A fair voting system should satisfy these three conditions:

 Unrestricted Domain (Universality)

The voting system should work for any possible set of individual preferences.

 Pareto Efficiency (Unanimity)

If every voter prefers X over Y, then the system should also prefer X over Y.

 Independence of Irrelevant Alternatives (IIA)

The ranking of X vs. Y should not change if a third candidate (Z) enters or leaves the
race.
What Does Arrow's Theorem Say?

 No voting system that ranks candidates (other than a dictatorship) can satisfy
all three conditions at the same time when there are three or more candidates.

This means:

 Any ranked-choice voting method will have some unfairness or contradictions.

 The only system that always satisfies these conditions is a dictatorship, where
one person decides the outcome.
Example 1: Choosing a Restaurant (IIA Violation - Spoiler
Effect)
Scenario
• Three friends (Alice, Bob,
and Charlie) want to decide
where to eat. They vote
between three options:
Using Ranked-Choice Voting (Instant Runoff
• A (Pizza 🍕) Method)
• B (Burgers 🍔)
• C (Sushi 🍣)
Each voter picks their 1st choice:
1. Pizza (A): 1 vote (Alice)
Their initial rankings are:
2. Burgers (B): 1 vote (Bob)
3. Sushi (C): 1 vote (Charlie)
Contd..

 Since no one has a majority, the last-place option (C, Sushi) is eliminated.

Charlie’s second choice (Pizza 🍕) gets transferred:

1. Pizza (A): 2 votes (Alice & Charlie) → Pizza wins

Pizza (A) is the winner!

Now, Suppose Sushi Was Never an Option
• Now, Let assume only Pizza (A) and Burgers (B) remain

 With only A and B left, Burgers (B) wins!

PROBLEM

• Just because Sushi (C) was removed, the final winner changed!
• When C was present → Pizza (A) won
• When C was removed → Burgers (B) won

 This violates the Independence of Irrelevant Alternatives

(IIA) because the ranking between A and B changed just
because C was removed, even though no one changed their
A vs. B preferences.
Example 2: Presidential Election (Real-World Spoiler
Effect)
 U.S. Presidential Election, Florida (2000)
• George W. Bush (Republican)
• Al Gore (Democrat)
• Ralph Nader (Green Party)
 What Happened?
1. Many voters preferred Gore over Bush, but some of them also liked Nader (who was ideologically closer
to Gore).
2. Nader got 97,421 votes in Florida, taking votes away from Gore.
3. Bush won Florida by just 537 votes, giving him the presidency.

 If Nader had not run, most of his voters would have picked Gore, and Gore would have won!

 This is a real-world case of Arrow’s Theorem in action. The presence of an "irrelevant" candidate
(Nader) changed the outcome between Gore and Bush.
Conclusion

Arrow’s Theorem proves that no ranked-choice voting system is

perfect.

because it either allows vote splitting, ignores majority preference,

or changes results based on "irrelevant" candidates.
Gibbard-Satterthwaite Theorem

• The Gibbard-Satterthwaite Theorem is a fundamental result in social

choice theory that describes the limits of voting systems when voters have
strategic incentives to manipulate the outcome.

• It was independently proven by Allan Gibbard (1973) and Mark

Satterthwaite (1975).
Theorem Statement
• If a voting system has at least three candidates and always chooses a single winner, then
at least one of the following must be true:
The system is dictatorial
• A single voter (dictator) always determines the winner, ignoring others.
The system limits choices
• The system is not fully expressive (e.g., restricting voters to only two options).
The system is manipulable
• There exists at least one situation where a voter can benefit from voting dishonestly
(strategic voting).
• Note:
👉 Every fair and expressive voting system with 3+ candidates is vulnerable to strategic
manipulation!
Example 1: Strategic Voting in Political Elections
UK General Elections (First-Past-The-Post System - FPTP)
• Scenario
• Three main parties:
• Labour Party (L) (Center-left)
• Conservative Party (C) (Right-wing)
• Liberal Democrats (LD) (Center-left but weaker)
A left-leaning voter prefers:
LD (Liberal Democrats) > L (Labour) > C (Conservatives)
Strategic Voting Effect
• The voter wants LD to win but knows they have little chance of beating the
Conservatives.
• If they vote truthfully for LD, their vote might be "wasted," and Conservatives (C) might
win.
• Instead, they strategically vote for Labour (L) to stop Conservatives from winning.
Contd..

👉 Arrow’s Theorem Problem

The voter does not vote for their real preference (LD) and instead
manipulates their vote to avoid a worse outcome.
Example 2: Reality Show Voting (American Idol, Bigg Boss,
etc.)
Scenario
• Three finalists: A (Strong Contestant), B (Moderate), C (Weak)
• Voting system: The contestant with the fewest votes is eliminated.
• Fans of contestant B fear that A is too strong.

Strategic Voting Effect

• Instead of voting for B (their real favorite), they vote for C (weaker contestant).
• Goal: Ensure that C makes it to the next round and A gets eliminated.
• Later, they can defeat C more easily.
Contd..

Gibbard-Satterthwaite Problem

👉 Instead of voting for their real preference, voters manipulate their

choice to increase their chances of winning later.
Example 3: Hiring Committees & University Admissions
Job Hiring for Professors at a University
• Scenario
Three candidates apply for a professor position:

A (Highly Qualified but Unpopular)

B (Moderately Qualified, Friendly)
C (Weakest Candidate, Close to Some Voters)
• The hiring committee uses ranked-choice voting.
Strategic Voting Effect
• Some committee members fear that A is too competent and might challenge their
authority if hired.

• They rank B first and C second, deliberately ranking A lower.

• Result: A is unfairly eliminated, and B gets the job.

Contd..

Gibbard-Satterthwaite Problem

👉 Instead of choosing the best-qualified candidate (A), committee

members manipulate their votes to favor someone they personally
prefer (B).
Conclusion on Gibbard-Satterthwaite Theorem

 Any voting system with at least three candidates can be manipulated

through strategic voting.

 Voters often don’t vote honestly—they vote to influence outcomes in their

favor.

 Systems like Ranked Choice, Approval Voting, or Runoff Elections reduce (but
don’t eliminate) manipulation.
Domain Restrictions and Multi-Winner
Elections
 Voting systems often face problems such as strategic manipulation,
vote splitting, and unfair outcomes due to Arrow’s Theorem and the
Gibbard-Satterthwaite Theorem.

 Domain restrictions and multi-winner elections attempt to address

some of these issues.
What Are Domain Restrictions?

 Domain restrictions are conditions placed on voters’ preferences to

avoid paradoxes or manipulation in elections.

 The goal is to simplify the decision-making process and ensure that

election outcomes are fairer.
Common Domain Restrictions

Single-Peaked Preferences

Voters have a single favorite option, and their

preferences become less favorable as they move
away from this option.

Example: In an election on tax rates (low,  Eco

 Ris
medium, high), a voter might prefer medium,
 Poli
but dislike both extremes.
Single-Crossing Preferences

Voters can be arranged in a left-to-right spectrum,

where their rankings change at most once.

Example: In a debate on environmental policies, voters

might shift from supporting industry to supporting
conservation at only one point in the spectrum.

Voters do not continuously shift between supporting industry and conservation;

 instead, they remain with industry until a specific threshold
Euclidean Preferences

Voters' preferences can be mapped in a geometric space (like left-wing vs.

right-wing on a political spectrum).

Example: In a city council election, voters prefer candidates geographically

closer to them.
better understand local issues
are more accessible
foster trust
and are seen as more likely to prioritize the community’s needs
Why Do Domain Restrictions Matter?

👉 They reduce strategic manipulation because voters’ preferences

follow predictable patterns.

👉 They make fair outcomes possible even under ranked voting systems.
Incentive Design in Crowdsourcing

Crowdsourcing platforms rely on distributed individuals to perform

tasks, contribute knowledge, or solve problems.

Effective incentive mechanisms encourage higher participation and

improve the quality of responses
Types of Incentives in Crowdsourcing
Monetary Incentives 💰

• Fixed Payment: Workers get a predefined reward per task (e.g., Amazon
Mechanical Turk).

• Performance-Based Payment: Rewards increase based on accuracy,

speed, or effort.

• Bidding Systems: Participants bid for tasks, ensuring competitive pricing

Non-Monetary Incentives 🏆

• Reputation Systems: High-quality contributors receive better rankings (e.g.,

Stack Overflow badges).

• Gamification: Points, leaderboards, badges, and levels motivate

engagement.

• Recognition & Social Rewards: Highlighting top contributors in a

community.
Intrinsic Motivation 🚀
• Skill Development: Users participate to learn and improve expertise (e.g.,
Kaggle challenges).

• Altruism & Social Impact: Some contribute to help others or support a cause
(e.g., Wikipedia).

• Enjoyment & Challenge: Fun and engaging tasks can encourage voluntary
participation.
Key Challenges in Incentive Design

• Balancing Cost & Quality: Higher payments attract workers but may not
guarantee quality.

• Avoiding Malicious Behavior: Incentives should prevent fraud, bot

participation, or spam.

• Task Allocation & Fairness: Ensuring fair distribution of tasks and rewards.
Models for Incentive Mechanisms

• Game-Theoretic Approaches: Design incentives based on rational agent

behaviors.

• Auction-Based Models: Dynamic pricing for task allocation.

• Reputation-Based Systems: Long-term incentives tied to credibility.

Game-Theoretic Approach
Kaggle Competitions

• Kaggle offers prize money and rankings to attract high-quality data

scientists.
• Top participants compete rationally to maximize their rewards
(money + reputation).

Note:
•If the platform sets rewards too low, workers will not participate.
•If rewards are too high, the platform overpays.
•Game theory optimizes rewards for a win-win situation.
Auction-Based Model: Dynamic Task Pricing
Concept: Tasks are allocated through bidding mechanisms, where workers propose
prices, and the lowest reasonable bid wins.

Reverse Auction in Amazon Mechanical Turk

• Task Poster (Platform): Posts a task (e.g., data annotation).
• Workers: Submit bids to complete the task.
• Winner: The lowest bidder gets the task, ensuring cost efficiency.

Note:

Ensures competitive pricing.

Prevents overpaying workers but may lead to underbidding (low quality).
Platforms must set minimum price limits to ensure fair worker compensation.
Reputation-Based System
Workers build trust and credibility over time. Higher reputation leads to better task
opportunities and higher rewards.

Stack Overflow & Upwork Reputation Systems

• Stack Overflow: Users gain reputation points for high-quality answers.
Higher reputation = More visibility.

• Upwork: Freelancers with better reviews receive higher-paying jobs.

Note:
 Workers with higher reputation get access to premium tasks.
 New workers struggle initially but can build credibility over time.
 Ensures quality contributions while reducing spam and cheating.
Real-World Examples

• Amazon Mechanical Turk: Microtask payments for human intelligence

tasks.

• Kaggle: Competitions with prize money and ranking incentives.

• Stack Overflow: Gamification through reputation and badges.

Future Trends

• Blockchain-Based Incentives: Smart contracts for fair and transparent

reward distribution.

• AI-Driven Personalized Rewards: Dynamic incentives based on user

behavior.

• Hybrid Incentives: Combining financial, reputational, and intrinsic

motivators.