Let's Toss for It

A Surprising Curb on Political Greed


But anyone who deliberately tries to get himself elected to a public office [in Utopia] is permanently disqualified from holding one.

Thomas More, Utopia

Nowadays elections are almost universally regarded as the keystone of political affairs. Besides paying taxes and perhaps serving in the military, average citizens participate in political life mainly by voting. Although people disagree about election procedures and often feel disgust with election outcomes, hardly anyone today doubts that elections provide the only way to establish, legitimize, and control a government. Historically, however, general elections have been the exception rather than the rule for selecting and guiding governments. Alternatives include various autocratic or despotic systems and processes and, in more democratic systems, methods that supplement or substitute for elections. Among the latter is lot-drawing, also known as sortition (from the Latin root sort, meaning "lot"). This procedure has intriguing characteristics and effects as well as potential for present-day utility. In this article I discuss its history and nature and consider some possible applications in the American political system.

Instances of Lot-Drawing

In the fifth century B.C., the Athenians filled their civic offices in two ways, either by the random operation of the lot (kleros) or by election. Most officeholders were selected by lot. Aristotle, among others, viewed lot-drawing as the more democratic procedure and election as the more aristocratic (Stockton 1991). Scholars are not sure about the exact procedure used: Was the selection made from all those eligible or, at least in some instances, from those eligible and willing to serve? Lacking enough willing citizens, was compulsion used? Did the Athenians sometimes select from a short list of citizens eligible, willing, and qualified to serve? Although the evidence is inconclusive, it seems likely that "sortition from among volunteers" was the rule, at least in a number of cases (Stockton 1991, 115-16). Holders of the most important offices, the archonships, were selected from a short list, direct election having been abolished in 487-86 B.C. and replaced by sortition from (probably one hundred) preselected candidates of the two highest classes of citizens. After about 460 B.C., all four classes of citizens were eligible for archonship. Later the appointment process became one of straightforward double sortition — two rounds of lot-drawing (Stockton 1991, 108-9). Before taking office, the selected candidate underwent a scrutiny (dokimasia) in which citizens could object to his character or record (Stockton 1991, 110-11). Athens had many civic offices, most with only modest power and many requiring only part-time service. Except for membership in the Great Council (the boulê), offices typically could be held only once in a citizen's lifetime. This system fostered participation, rotation, and amateurism: Many citizens participated in wielding power; rotation in office prevailed; and the outlook of the ordinary citizen pervaded the civil service and the judiciary. Including the boulê, "about one thousand posts had to be filled year in and year out from among citizens aged at least thirty" in a citizenry of "between thirty or forty thousand or so above the age of eighteen." In addition, some six thousand served as dicasts (jurors) in the courts (Stockton 1991, 112).

In the heyday of their republic, the Venetians selected their lifetime leader, the Doge, by a complex system involving lot-drawing. The system had developed through the Middle Ages, becoming ever more complex to avoid manipulation, before being codified in 1268. The procedure consisted of a series of ten ballots that alternated between sortition and election. All participants had to belong to the Great Council, which included several hundred members of the most prominent families. The steps were as follows (Dahl 1994, 14-16):

1. The ballottino, a boy chosen at random, draws thirty names by plucking balls out of an urn, thus setting the process in motion with a blind draw.

2. Those thirty are reduced to nine by a blind draw.

3. Those nine put forward forty names, each of which needs at least seven of the nine possible votes.

4. Those forty are reduced to twelve by a blind draw.

5. Those twelve put forward twenty-five names.

6. Those twenty-five are reduced to nine by a blind draw.

7. Those nine choose forty-five new names, each of which needs at least seven of the nine possible votes.

8. Those forty-five are reduced to eleven by a blind draw.

9. Those eleven choose forty-one, who must not have been included in any of the reduced groups that named candidates in earlier steps.

10. Those forty-one then choose the Doge.

The Venetian system seems devised to make it impossible for any individual, family, or coterie to plant candidates or exercise undue influence. However convoluted the procedure, it supported a republican government that lasted a thousand years, until 1797.

Selection of a new Dalai Lama in Tibet involves a baroque procedure. By traditional esoteric divination, a committee of priests identifies a boy destined by the gods to become the lifetime head of government and the high priest of the country's major Buddhist sect. The boy is treated as a prince and educated for his office, which he eventually assumes. This procedure differs from lot-drawing because it gives considerable influence to the priests. But it includes an element of chance because no one can predict what sort of man the boy will become. It also precludes self-promotion and power-seeking by candidates: no candidates, no campaigning.

Historically many positions of leadership have been filled by hereditary succession. This involves randomness because heredity itself is an arbitrary yet objective criterion, and no one can know what sort of person the new monarch or prince will be. In this case the randomness does not entail participation, rotation, or amateurism. The king represents a dynastic interest and a political class. Still, an old-fashioned monarch is not beholden to any particular interest, which allows him to take a broad and long view if he cares to do so. Needless to say, the monarchical principle of legitimacy has long since fallen out of favor among opinion leaders.

In countries with a jury system, the jury is formed by lot-drawing. Names typically come from local census or voting rolls and are subject to elimination on grounds of insanity, criminal record, or other bona fide reasons. Counsels may also have the right to exclude some jurors, though they may not put forward any person for inclusion in the jury.

Amateurism and Representativeness

A system of universal lot-drawing for public office would create proportionality in the sense of giving public officials roughly the same composition as the general citizenry without giving disproportionate weight to people with any narrow set of characteristics. Thus, if x percent of the citizens are one-legged, red-haired, nearsighted, nonsmoking Methodist cabinetmakers, one would expect to find that about x percent of the officeholders fit that description. Such proportionality differs from that usually discussed by electoral reformers. Conventional proportionality aims at the proportional representation of political parties or, more broadly, groups defined along occupational or geographic lines. By implication, every economic interest should have a correctly weighted influence in dividing the benefits dispensed by the government. Accordingly, determination of the type of characteristic to be represented tends to be highly selective — race and sex are currently fashionable. Universal lot-drawing leads to amateurism and individualism, whereas proportional representation leads to professionalism and corporatism.

A large political entity, such as a nation-state, cannot fill every public office by random selection, nor has any national legislature been composed in this way. On a local or regional level, however, it would be possible to form a legislative council by drawing the name of, say, one out of every hundred eligible citizens, thereby creating not an elected assembly but a "stand-in folkmoot" — a diminished version of the real thing.[1] As a universal system, sortition belongs to the small-scale participatory democracy of the parish, the municipality, and the historic city-state, not to the large republic. But the modern world does present some interesting examples of amateur officeholding and broad participation.

The Amish people in North America furnish one such example (Hostetler 1993, 105-11). They belong to a German-Swiss Protestant religious brotherhood, conspicuous by its members' old-fashioned dress and manners. The Amish communities are agrarian, largely autonomous, and self-sufficient. The group's main offices — bishops, preachers, and ministers of the poor — are held by unpaid amateurs (men only) selected by the leaders of the congregation according to traditional rules and by general consent. The humility ideal of the religious doctrine prohibits putting oneself forward as a candidate or campaigning for others. Some appointees find their selection to office troubling, but they cope with it. Many Amish women volunteer to train briefly as teachers or nurses and then serve without pay in those capacities.

As already mentioned, jury systems exemplify random selection as well as broad participation and rotation.

Many rural communities, though less now than previously, practice rotation of communal duties, usually unpaid, among local citizens.

Many associations also expect members to volunteer to serve as officers or functionaries or to accept such duties when asked to do so.

Even modern governments retain unpaid honorary officials, such as justices of the peace in Britain and the United States, honorary consuls, and holders of various local offices of a legal nature. In these cases we find amateurism though not randomness or broad participation.

In a large polity such as a state or federal union, the use of random selection more often applies to particular officials or to cases that cannot be dealt with otherwise.

In a representative system, the most obvious means of achieving amateurism is not sortition but term limits or part-time officeholding by people who retain their ordinary jobs. The Swiss, who use the latter even in their federal parliament, call it the "militia system" of politics (Fenner and Junker 1989, 75, 78-79).

The Nature of Lot-Drawing

Allowing decisions to hinge on chance has a long and fascinating background. Historically, especially when information has been scarce or both sides of an issue have had considerable merit, people have often trusted in chance — perhaps in combination with a bit of human interpretation or intuition — employing such means as augury (divination from omens), haruspicy (inspection of the entrails of sacrificial animals), Delphic oracles, Chinese fortune bones, the throwing of dice, and the tossing of sticks.

Because random selection by its very nature seems irrational, some may reject it out of hand, but the unfathomability of sortition also makes it a tool for creating legitimacy and efficiency. Accepting the principle of random selection and knowing that chance has given rise to a public decision, one has little to complain about if one dislikes the outcome. On the other hand, because of the claim that electoral democracy epitomizes perfect rationality, an election may engender vociferous complaints by those disappointed in its outcome. Many people may also find the long debate before an election exhausting and frustrating. If postelection disappointment were frequent and acute, the very legitimacy of the electoral system would be endangered. Popular wisdom understands the utility of the luck of the draw. When a quarrel has continued for a long time without anyone's changing his mind, someone may suggest "tossing for it." The result of the toss settles the matter, the loser swallows his dismay, and life goes on. As Guglielmo Ferrero (1942) remarks, no principle of legitimacy is entirely rational, but that deficiency does not prevent a given principle from working well in practice if it is adhered to consistently and faithfully.

Powerful as they are, reason and analysis generate reliable answers only where the relevant facts, well defined and accepted, are at hand; where the objectives are clear and agreed upon; and where the alternatives are few. But such conditions rarely exist. When reason has done what it can, something else must complete the job: intuition, faith, or chance. Of these, only chance is objective and has no bias.

With random selection of officeholders from the citizenry at large, there are no candidates and no campaigning. This is a quiet procedure. Only when lot-drawing is combined with election or appointment do self-promotion and campaigning appear.

By its objectivity, sortition avoids the engagement of passions and interests. Discussion can continue forever, but once Fate has spoken, one must be silent. One may doubt human counsel but not the finger of God. The quietness of the process gives it dignity; the absence of manipulation confers legitimacy on it. Thus, sortition is especially apposite for selecting one among equals for the distinctly unequal position of holding the right to command others. (To exclude unfit persons before the final draw, other procedures may be necessary.)

Politicization: The U.S. Presidential Race

The U.S. presidential election system illustrates a maximally politicized election process. By politicized, I mean simply that considerations of merit and utility as understood by the typical citizen take second place to the desires of political insiders and organized special interests. We cannot, of course, "take the politics out of politics." Politics will be partisan, and politics will never be snow-white. Still, if a democracy is not to be a sham, it must strive to serve the general rather than any particular interest.

In the present perspective, the main elements of the American presidential election process are the following. First, the various subgroups in each party put forward candidates regarded both as friendly to the subgroup and as having a reasonable chance of winning; of course, the candidates must be fairly well known and willing to run. Then, in each party, the various subgroups compete to get the party to accept their candidates as the party's candidate; again, the successful candidate must make himself known and acceptable to all and make a case that he can win. Then, in the final campaign, each party spends a vast amount of money and effort to make sure that its candidate is well known, well liked, and a possible winner. Throughout the process, launchers, candidates, and supporters eagerly search for sponsors outside the political system — individuals, groups, and organizations willing to supply voters or funds. The ultimate winner gets monarchic powers for four years, during which he must work in various ways to repay his launchers and sponsors, especially if he hopes to secure reelection.

Clearly, the system turns on self-promotion by candidates, eager maneuvering by initiative groups inside the political system, deal-making with outside special-interest groups, and calculations of candidates' salability. Moreover, the campaigning to publicize the name and policy positions of candidates requires much money, partly because the United States is an immense country. So money is the major consideration. Traits of the American character, such as brashness and the worship of financial success, compound the insidious predominance of money in campaigning. A president's great capacity to affect the material conditions of individuals, groups, and firms by wielding his statutory and discretionary powers colors the entire process.

Because so much of the selection procedure occurs before the choice is at last put before the public, the system as such does not ensure that the man elected is the one regarded as best qualified by the public (Vile 1984, 86). One recalls Gaetano Mosca's acid dictum that a representative is someone whose friends have arranged for him to be elected. Further, the system practically ensures that the two (or more) final contenders will be men reared by the political establishment, beholden to it, and constituting no threat to it — in short, men such as Bill Clinton and Bob Dole. The system also ensures a great deal of sound and fury as the contenders attempt to convince voters that the choice is one between night and day. The earnest voter might be pardoned for feeling, during the throes of a presidential campaign, that the best outcome would be for the two candidates to stop their electioneering, engage in a manly duel, and shoot each other dead.

The all-importance of money in American presidential campaigning means that if a party or a party subgroup has confidence in the winning potential of its candidate, the rest is regarded as a matter of money. Even an independent candidate may perceive a fighting chance if he has money or celebrity status. Personal wealth and a gambler's self-confidence can buy the aspirant a place in the public eye, at least for a (perhaps considerable) time, as illustrated by the recent campaigns of Ross Perot and Steve Forbes. Whoever seeks the candidacy must promote and publicize himself; in this endeavor, modesty is not a virtue but a liability. Nor does anyone view the president as "above politics." The indignities of the campaign make it hard for the bruised survivor to establish his authority.

One may view a presidential campaign, not unreasonably, as a gamble with rather favorable odds and astounding gains — provided one has, or is, a candidate with good communication skills, an entrenched position in a political party, and access to ample funds, which in turn implies good connections to special-interest groups for whom one's victory has high value. From an insider's perspective, one sees campaign spending, not unrealistically, as having the power to better one's odds considerably or even decisively. For the outside interest group, campaign contributions are simply business investments with a calculated chance of paying off. So, from all sides come cries of Take me! No, me! No, me! and the suggestive crackling of dollar bills. Surely one must wonder whether the sort of person likely to engage in this huckstering and enjoy it — the endlessly flexible, ever smiling, eagerly self-advertising type — is likely to be the person best suited to lead the country.

Disciplining U.S. Presidential Elections

The Founding Fathers had a plan, though unfortunately not a very good one, for imposing discipline on the presidential election process (see paragraphs 1-4 of article 2, section 1, of the U.S. Constitution, partially superseded in 1804 by the Twelfth Amendment). They prescribed a two-step procedure, whereby the people in the several states would elect the members of an Electoral College, who in their greater wisdom would then vote on the presidential candidates. The president of the U.S. Senate would announce and confirm the winner or, in the extraordinary case that no candidate achieved an absolute majority, leave it to the House of Representatives to choose a winner from among the three top vote-getters. The Founders feared voter immaturity and mob politics, and paternalistically they trusted in the principle of leaving the actual selection to elected representatives. Whatever the wisdom of such a principle, history has rendered it hollow: The pressures of party politics have ensured that, although the formal mechanism remains the same, today's electors are bound by their party mandate to vote for the party's candidates; hence, the outcome of the election turns entirely on the party composition of the Electoral College. In other words, indirect election works as if it were direct. The safeguard, such as it was, has collapsed. Only the pointless complexity of the procedure remains.

Later reforms added further complications but failed to dispel the curse of oligarchy. The system of presidential primary elections, adopted by many states early in the twentieth century to ensure direct popular control of the nomination of party candidates, has not lived up to its promise. Nor are the party conventions for selecting a presidential candidate all they seem. "Conventions could be managed by party leaders so that they became a façade for oligarchic control by professional politicians" (Vile 1984, 87). A good recent example was the Republican national convention in San Diego in August 1996: the party's presidential candidate, Bob Dole, had already been selected before the convention began.

Including an element of lot-drawing in the selection of a given official would make the investment of money and effort by an individual or his campaign backers unnervingly risky. If, for example, the final selection were by blind draw and the final pool contained nine or twelve names instead of the usual two, the odds of any particular candidate's winning would be far different from the fifty-fifty of a conventional, closely contested election. The salient point is that no additional spending of money or effort or compromises could ensure victory or appreciably improve one's chance of winning. Inscrutable, immovable, unbuyable chance would be decisive, and the vicious cycle of individual self-promotion, cynical compromise, and blatant investment of money by pressure groups would be stopped. The dynamics of selection for office would be altogether different. Insurers talk about "uninsurable risks," which are incalculable and therefore beyond the bounds of actuarial science. Something similar (an "unfinanceable bet") would obtain in a presidential race in which chance removed crucial choices from human influence. By including elements of random selection in the choosing of a president, such a system would drastically reduce the importance of money.

A reformed system of presidential election could employ sortition in several ways: (1) to compose an electoral college from a larger body; (2) to eliminate names from an elected body of candidates; (3) to pick the final winner from a pool of candidates. It is not necessary that a blind draw make the final selection; alternatively, one could require a strong majority in an election, with random selection as a disciplining fallback option. The objective to bear in mind is that the procedure somehow blight the natural inclination of political insiders or special-interest outsiders to further their own interests by manipulating the workings of the selection process. Fundamentally that objective can be achieved only by reducing their expectation of a profit from manipulation. Including an element of chance at one or more suitable points in the procedure is the only expedient sure to have that effect. Recall the system devised by the Venetians; they knew what they were up against.

Curbing Particularism, Centralization, and Partiality

Selecting a president in a federation such as the United States combines two kinds of problem. First is the general problem of ensuring competence or impartiality in an elected leader, which we have just considered. Second is a federalist requirement: maintaining the legitimacy of the union despite its size and diversity. This raises the perennial issue of centralization versus states' rights. The federalist problem pertains to every sort of higher federal institution. Concerns about federalism often become entangled with concerns about impartiality.

A process involving lot-drawing could diminish conflicts grounded in particularism or regionalism in a federal system. To select a federal official, one might proceed by drawing from a pool of names put forward by the constituent states. Thus, the states would propose, but luck would dispose; no state could impose its candidate, and no state could complain that the result was unfair. By removing bones of contention, random selection would cement the union, bolster its legitimacy, and conserve political energy. By offering an alternative to appointment by the central government, such a process would check centralization.

Processes involving lot-drawing can also help to ensure both the competence and the impartiality of officeholders. For an office requiring specific professional skill or political experience, a two-step procedure could be adopted: first, a conventional selection of a pool of qualified candidates, to ensure competence; then a final selection at random, to thwart politicking. Selection of judges, for example, might proceed according to this procedure. It might also be used to select police officers, election officials, and members of various local boards.

Appointing Supreme Court Justices

Concerns about competence and federalism surely arise with regard to the selection of justices of the U.S. Supreme Court. One would hope that the professional soundness and political impartiality of the justices would not be twisted by political partisanship or expediency. To stand above politics is part of the Supreme Court's very raison d'être. Alas, the present system of presidential appointment of justices subject to the approval of the Senate fails badly, producing a maximum of politicization in both the selection process and the composition of the Court. Federalism suffers, too. The union is not supposed to be a unitary state. According to the Tenth Amendment, "The powers not delegated to the United States by the Constitution, nor prohibited by it to the States, are reserved to the States respectively, or to the people." Federalism touches the Supreme Court directly because one of its functions is to adjudicate disputes arising between the federal government and the states. Appointment by the federal government predisposes the justices to side with the federal government. This bias generates a dangerous drift toward a unitary system because a subservient Supreme Court is unlikely to check centralizing actions by the federal government. To counteract such drift as well as the politicking that endangers political impartiality and professional soundness, reform of the judicial appointment process deserves consideration.

I suggest a two-step system.[2] First, create a pool of qualified candidates by having each state nominate candidates from among its own judges. Second, fill the actual Supreme Court vacancy by a blind draw from that pool.

Such a procedure would bring about many changes. (1) It would eliminate the rewards of justice-selection politicking at the federal level; (2) the justices, no longer the creatures of the federal government, would have less inclination to expand the power of the central government; (3) political vetting of judicial candidates would occur only in the states, where genuine support for federalism would carry more weight than it does in national politics; (4) the rewards of justice-selection politicking in the states would have small expected value, because no state's candidate in the federal pool would be assured of actual appointment; (5) reaching the Supreme Court via state selection, the justices would give greater support to states' rights, but their possible particularisms would tend to balance each other out and, in any event, to matter less than they do now. The likelihood that the Court would consistently favor a particular state would be small, whereas under the present system the likelihood that the Court will favor the federal government over the states is great.

Cutting Gordian Knots

Sortition could serve in some cases as a fallback measure: If a political body were unable to agree on a decision by deliberation and voting, it could resolve the matter by casting lots. Agree or else. Such a threat would have a disciplinary effect, helping to end unreasonable or dishonest holdouts. Thus the mere threat of decision making by sortition could become the sword that cuts Gordian knots.

Consider a case in which a parliamentary body agrees in principle but disagrees on details, and as a result the relatively trivial disagreement produces "gridlock," preventing the implementation of the principle. For example, in a budget debate a majority of the legislature favors reducing the budget, but each party has pet projects it will not give up, thereby blocking overall reduction of spending. Such a deadlock could be resolved by dividing the problem into two parts and applying lot-drawing as a goad to agreement. First, the legislature decides on the budget total; then it considers which elements to cut, under the threat that if agreement is not reached, the overall excess of spending will be eliminated by random deletion of individual projects. Under this constraint, the members must choose between chosen cuts and randomly selected cuts, and the relative undesirability of the latter may spur them to reach agreement quicker.

Another possibility is to use random selection in the appointment of an official from a limited list of candidates by an electoral college subject to two constraints: first, the winner must gain unanimous approval or, perhaps, a strong majority such as three-fourths; second, if the college does not make a selection in step one, the winner will be selected by blind draw from the list of candidates. The potential for proceeding to step two would greatly diminish the expected profit of any elector from holding out against a good candidate for manipulative motives, such as extracting concessions from other electors. In this system of selection, either the winner would have a strong majority of votes, which would make him more than the representative of one party, or he would be simply the lucky candidate, which would attest that he does not owe his office to manipulation by any particular interest. Either way the winner would have secured legitimacy lacking under present methods of selection.


However difficult it might be to adopt the use of sortition in political decision making, the device deserves serious consideration by constitutional reformers. Our present methods of decision making are not so fine that they cannot be improved, and including an element of sortition is not nearly so daft as it may sound when first proposed.


* Sigmund Knag is an independent scholar and author living in Bergen, Norway.

1. James Fishkin, a professor of government at the University of Texas, has carried out experiments with "deliberative polling," using a representative group of citizens similar to what I call a "stand-in folkmoot." His tentative conclusion, reported in the Economist of May 16, 1998, is that such a group of laymen is capable of dealing rationally with complex political issues, given some counselling by outside experts and debate within the group.

2. John F. Knutsen of Oslo suggested this constitutional idea to me.


Dahl, Birger. 1994. Venezia, et kulturhistorisk eventyr. Oslo: Tell Forlag.

Fenner, Martin, and Beat Junker. 1989. Bürger, Stat und Politik in der Schweiz. 5th ed., reprint. Basel: Lehrmittelverlag des Kantons Basel-Stadt.

Ferrero, Guglielmo. 1942. Principles of Power. New York: Putnam.

Hosteller, John A. 1993. Amish Society. 4th ed. Baltimore: Johns Hopkins University Press.

Stockton, David. 1991. The Classical Athenian Democracy. 2d corrected impression. Oxford: Oxford University Press.

Vile, M. J. C. 1984. Politics in the USA. 3d ed., reprint. London: Hutchinson.

Sortition Contents | Text Version