Every year, the faculty of my college elect a vice chair from among the tenured professors serving on the Faculty Council. In most years, we choose from among at least three candidates, and the voting has always been by plurality. That is, the candidate with the most votes wins. This process can result in the selection of a candidate who is the least favorite of the majority of the faculty. For example, in a case with three candidates, if just over a third of the faculty vote for Candidate A, and the remaining are split evenly between B and C but would all rank A last, A will win, leaving almost two thirds of the faculty with their third choice as vice chair.
Every year after the election, one of my mathematics colleagues fantasizes about trying to persuade the faculty to adopt an alternative to the plurality system, but he always concludes that too many of them will be suspicious of something unfamiliar or intimidated by anything more mathematically complicated than simple majority rule. What he needs is a concise, straightforward account of the alternatives that outlines their strengths and weaknesses and provides some historical context for their use. After reading such an account, his colleagues might be willing to consider a new way of voting.
George G. Szpiro offers Numbers Rule as just such an account. For a general audience, Szpiro sets out to explain the standard alternatives to plurality voting, as well as the problems inherent in allocating seats in a parliament. He aims to provide “an elucidation and a historical account of the problems and dangers that are inherent in the most cherished instruments of democracies” (x).
Szpiro focuses primarily on alternatives to plurality voting that are variations on what are commonly called the “Borda count” and the “Condorcet system.” Both determine a winner by taking into account the electors' rankings of the candidates. For the Borda count, each candidate receives points according to his or her position in each elector's ranking. The winner is the candidate with the most points.
The Condorcet system uses the voters' rankings to evaluate each candidate in head-to-head contests with every other candidate, requiring that the overall winner beat each of the others. The Borda count and the Condorcet system do not always produce the same winner. Moreover, the Condorcet system does not always produce a winner at all, but may result in a cycle, with A beating B, B beating C, and C beating A.
Szpiro describes these voting systems and the inconsistencies and paradoxes they produce by describing the texts and their authors in which the systems have appeared. Although these two alternatives to plurality voting bear the names of eighteenth-century mathematicians who discussed them at length (see below), the systems appeared for the first time (as far as is known) in the work of two earlier authors. Ramon Llull (ca. 1235–1315) proposed a Condorcet method and Nicolas Cusanus (1401–1464), who was familiar with Llull's writings, described a Borda count. Both of these authors write about their ideas in the context of electing monastery or church leaders, though it seems the methods were never put into practice. Szpiro discusses Llull and Cusanus in chapters 3 and 4.
Jean-Charles, chevalier de Borda (1733–1799), outlined the method that bears his name in a paper published in the 1781 volume of Histoire de l'Académie Royale des Sciences. The Condorcet system derives its name from the mathematician and philosopher Marie-Jean-Antoine-Nicolas de Caritat, marquis de Condorcet (1743–1794), whose work on voting theory began appearing in print in the 1780s as well. In fact, Condorcet most likely authored the introduction to Borda's paper as the beginning of a mostly one-sided debate between himself and Borda over the advantages of the two systems. Neither Borda nor Condorcet seems to have known of their predecessors' ideas. Szpiro begins this narrative in chapter 5, focusing on Borda, and turning to Condorcet in chapter 6.
The story continues with the introduction of the mathematician Pierre Simon de Laplace (1749–1827) in chapter 7. Laplace lectured on the relative merits of the two systems in 1795 and concluded that since each had its flaws, the best system would require the winner to earn a majority (not just a plurality) of votes, leaving open the possibility that an election would produce no winner. For a few years at the turn of the nineteenth century, the French Académie des Sciences used the Borda count in its elections. Apparently unfamiliar with any of these writers, Charles Dodgson (Lewis Carroll) presented a series of pamphlets in the 1870s, recommending variations on the Condorcet and Borda systems to his colleagues at Oxford's Christ Church, who were trying to find fair ways to elect students and faculty to college office. Szpiro tells Dodgson's story in chapter 8.
Chapters 9 and 10 discuss the problem of apportioning representatives to the states or other regions they represent. Since an exact allotment of representatives to states of different populations requires fractional representatives, some method must be chosen to round those fractions to whole numbers. All methods have the potential to favor large states over small, or small over large, or to produce allotments that seem in some way paradoxical. In the United States, debates about the best method go back to Thomas Jefferson and Alexander Hamilton, and continued into the twentieth century. Szpiro outlines the early history and explains the various paradoxes in chapter 9, focusing in chapter 10 on the controversy of the 1920s through the 1940s between sociologist Walter Willcox and mathematician Edward V. Huntington that made its way from Congress to the pages of Science to a special committee of the National Academy of Sciences.
Chapters 11 and 12 explain why neither a perfect election system nor a perfect apportionment method can exist. Kenneth Arrow's Impossibility Theorem is the subject of chapter 11, while chapter 12 explains the work of Michel Balinski and Peyton Young. In chapter 13, Szpiro describes the apportionment systems in Switzerland and Israel and a proposed election method for the French president.
In this book, Szpiro has brought together information from a wide range of the secondary literature and covers much ground. He explains the various voting and apportionment systems thoroughly and in an accessible way, and he gives the reader some sense of the people who produced the texts. It would be difficult to find a single book that combines all these topics, although plenty of other introductions to the mathematical ideas provide less rambling explanations of the systems and their paradoxes. The biographical and other historical details given here rarely go beyond what students could find in basic encyclopedias, and at times are so oversimplified as to be almost misleading. The bibliography contains a fine list of sources for further reading, but since Szpiro almost never cites his sources, it is not always easy to tell which will provide information about particular topics. One does not expect detailed footnotes in a book for a popular audience, but an author has some obligation at least to cite sources for direct quotations, which Szipro rarely does. Finally, I found the author's frequent use of colloquialisms and his tendency to make flippant remarks distracting. Thus, while this book might provide an entertaining overview for students getting a more thorough treatment of the subject in a mathematics or political science course, I would be reluctant to include Numbers Rule in the syllabus.
