Fr Tr => Ss a= Sop Inspired by: Vertasium eo >, ©To start off, democracy is a type of 0 government where the majority of = + decisions are made by the “people.” Things like who will be the leader, what policies to implement, where will funds go, etc. L Now, how would you feel if a Nobel Prize winning Mathematician comes and tells you that you may not be (mathematically) living in a democracy? Cty And that’s the case. Mathematics refutes the way democratic leaders are elected! Let’s help you understand all developments, considering 100 voters and 3 candidates competing for election -Nase Simple rule - whoever has 50% votes, wins. But what if Alice, Bob and Charlie get 30%, 40% and 30% votes, respectively? Solution - “First Past the Post” Whoever has the Countries like India, the highest votes wins. So, US and the UK use this Bob could win despite system due to its simplicity not having a majority. and quick outcome. However, according to Duverger's Law, this suffers from “the spoiler effect”. a Let’s say Alice, Bob and Charlie get ° 40%, 40% and 20% votes, respectively. Now, what if Alice’s connections 5 intentionally bring a least-preferred candidate David, who somehow takes 5% votes from Bob. Alice could win! Vote for me! Source: Dahl, R. A. (1991), Cox, G. W. (1997), Yale & Cambridge University PressInstant Runoff Voting (IRV Selection by elimination. w Voters rank candidates, lowest-voted O O © candidates are eliminated, and votes are reallocated until a majority wins. (h) (~~ (=\ Considers 2nd preference of Charlie's voters) ALICE 4O% + 5% = 45% BoB 35% 35% + 20% = 55% (wins) CHARLIE 25% (eliminated) x As there's no majority in first-choices, Charlie with least votes gets eliminated and his voters’ second-choice is reallocated. Bob, with the highest reallocated votes, wins. Countries like Australia and Ireland use this method as it eliminates the spoiler effect. However, this method is complex and takes time to implement. Besides, the highest Ist preference candidate may not win, for instance, Alice.Each voter ranks the candidates and then the candidate with the highest rank-based score wins. For example, let's assume each Ist rank has 2 pts, 2nd has 1 pt and 3rd has 0 pts. XN ALICE YO votes (BoBy =| ‘CHARLIEY 30 votes IST RANK (X) 30 votes 2ND RANK (Y) 10 votes 50 votes YO votes 3RD RANK (2) 50 votes 20 votes 30 votes TOTAL SCORE 90 points 100 points (X*2 + Ys 2*0) Winner This was proposed by French mathematician Jean-Charles de Borda, back in 1770. A similar format is adopted by The Pacific island nation of Nauru. While it considers the broader audience's preference, it does encourage strategic voting. For instance, many voters could have deliberately ranked Alice (the highest Ist ranker) as 3rd to boost their favorite Bob's chances of winning. Source: McLean, |, & Urken, A. B. (1995), University of Michigan PressCondorcet’s Paradox Voters rank candidates. Each candidate competes one-on-one with every other candidate. Whoever wins the most number of matchups, is the Condorcet winner. While it did reflect majority preference, it had one clear circular dilemma. ©) OS S co 4o% 35% 25% Alice > Bob > Charlie Bob > Charlie > Alice Charlie > Alice > Bob Head: to-head comparison results: @ 6 @ co Alice > Bob Bob > Charlie Charlie > Alice YO% + 25% = 65% YO% + 35% = 75% 35% + 25% = 60% (Alice wins (Bob wins against (Charlie wins against Bob) Charlie} against Alice) And there you go, no one is a clear winner! Every time there are more than 2 candidates, the Condorcet paradox can kick in! Source: Condorcet, M. (1785)yA ht) Voters approve multiple candidates, each getting one vote. The candidate with the most approvals wins. For instance “) Final Votes Alice Charlie 4O%® YO% + i = 75% 35% + 25% = 60% — - S = Organizations like the UN and the IEEE use this as it is simple and reduces spoiler effect. However, this method doesn't reflect the strength of preference between approved candidates. Also, voters may rig the results by voting ‘only’ for their top choice. Source: Brams, S. J, & Fishburn, P. C. (1983)am Tyke Te LAY Pa A question that papers on statistics, probability and behavioral studies long pondered over, until in 1951 when Kenneth Arrow outlined 5 requirements for a fair voting system: co 2. Non- (.Unanimity dictatorship f everyone prefers : 7 Alice en Bob, the No single voter's group should too choice should v7 fe iit we A al'| Fle 4. Consistency 5. Irrelevance of New Options I the group prefers Adding a new Alice ae B25, and ccneth, David, Bob over Charlie, they shouldn't change the should prefer Alice group's ranking of over Charlie Alice vs Bob vs Charlie However, satisfying all these 5 conditions is nearly impossible. Source: Arrow, K. J. (1951)Arrow’s Theorem: Why N Voting System is Perfect 4 Arrow showed it's impossible for any ranked voting system with 3 or more candidates to a meet alll five rules. Imagine this scenario where 7 34 voters rank 33 voters rank 33 voters rank A>BoC BoCoA C>A>B Most voters agree on A > B and B > C. However, another majority Group preference Unanimity forms a cycle: prefers C > A, forming a loop, eo Cc . violating Consistency fon One fixed order A > B > C wins due to majority. Unanimity + (A>B>C)is But this violates Universal Consistency | chosen, ignoring 66 Participation, ignoring the voters preferences of 66 voters Also known as “Arrow’s Impossibility Theorem’, this won him a Nobel Prize in 1972. Source: Arrow, K. J. (1951)When Arrows landmark Cc» D Gs) discovery rejected ranked voting, researchers found an alternative: > Co HD Gm? SSO Voters rate candidates from 0 to 5, and the highest score wins. Reality TV shows and online polls apparently found this system fascinating, and adopted it. However, most nations didn't! Probably because there is still a scope of rigging the system. Today, while 108 out of 167 nations are ‘democracies’ of some kind, whether most of these are pseudo-democracies is still a question.