This is a revised and expanded version of an article which originally appeared on as Tournament Elections: A Revolution in Democracy.


This paper proposes an alternative method to the traditional "one person, one vote" method of conducting democratic elections. Instead of the pool of voters electing from a small set of candidates, each person is a participant that both votes and runs for office. Large populations are repeatedly subdivided geographically into eight groups of roughly equal size, until each group has no more than eight participants. The participants in each group discuss issues during a meeting, at the end of which a representative is selected by and from the members of the group to move on to the next round. Eventually the winner of the tournament will be elected into office as the final representative chosen from the final group. The goals of this method are to eliminate corruption through thorough examination of all candidates and elimination of the influence of money and special interests, while encouraging more active participation in politics by normal citizens.

An article previously posted on Kuro5hin discussed the potential advantages of electoral reform through the approval voting method. Though it seems likely that voting reform could greatly improve the state of democratic politics, there are other issues such as the influence of money and special interests that cause the political process to yield results that oppose the wishes of the citizenry. These issues can only be resolved by greatly restructuring the means to political advancement.

Introduction: The Problems With the Electoral Process

Most people are fairly cynical when it comes to politics. Even those who vote often feel they have no real impact on the outcomes of elections, lawmaking or matters of policy. Politicians are seen as corrupt, dishonest and beholden to special interest groups and powerful corporate allies able to supply the funding necessary to achieve political success. In the United States a large part of the problem is seen as the virtual Democratic-Republican duopoly in all major elections. Smaller third parties offer more diverse philosophies than these largely centrist parties, but are seen by the majority of voters (including those who may identify strongly with their philosophies) as fringe groups incapable of real political impact. Voting for candidates from such parties is considered throwing your vote away and even worse, aiding the least-liked major party candidate by not supporting the other, the lesser of the two evils. Alternative election procedures such as approval or Condorcet voting may help break this two party stranglehold on the election process by allowing voters to follow their conscience without sacrificing pragmatism, but this only solves part of the problem.

The most perfect election process that requires voting for candidates running in a single winner-take-all election cannot prevent special interests from influencing the process. Running for any major office requires serious resources. Candidates need to advertise, travel, and make public appearances, and that requires money. Some money is raised through contributions from individual citizens but most comes from corporations, often in the form of soft money donated to political parties rather than directly to candidates. Large corporations will often donate heavily to both major parties to ensure their influence regardless of an election's outcome. Campaign finance reform law recently passed in the US is supposed to deal with these issues but many people feel that it is not a viable answer. The law may help reduce the blanket effect of corporate influence by restricting soft money, but this may only force special interests to choose individual candidates to support more carefully. In addition there is much debate over whether campaign finance reform will affect outside interests or even candidate's own political parties from offering support implicitly through issue based advertising due to the First Amendment protections for Free Speech disallowing restrictions on such ads. Likewise many individuals feel that governmental restrictions on the ability of a citizen to give money to any recipient they choose, for almost any reason, is in itself an infringement upon Freedom of Speech.

I believe that to solve the fundamental problem of inequitable representation in politics due to the requirements of funding and the influence of special interests requires fundamental changes to the process used to elect representative public servants. In a perfect system an individual would be elected to office based on integrity, intelligence, knowledge and experience relevant to the particular office. Such a representative would be the best individual capable of representing the interests of her constituents accurately and efficiently. All current systems are virtually guaranteed to not select this individual, if for no reason other than the incredibly small pool of viable candidates. For the office of President of the United States there are typically just two viable candidates on the actual election day, and only a few more at the beginning of state primaries. These candidates have typically been within the political system for years before running for President, are members of one of the two major parties, and have the backing of other major politicians and corporate sponsorship. These huge de facto barriers of entry to running for major political office eliminate a large pool of potentially superior candidates, some of whom would like to become involved in politics but lack the resources, others who have never given politics serious consideration but may posses qualities that would make for a fine representative.

It is these individuals that hold the key to true political reform. By replacing the current electoral system with one that gives every individual possessing intelligence, integrity and determination significant influence on the selection of a representative, or even a credible chance to achieve political office themselves, a whole new pool of viable candidates would open up. It could simultaneously replace voter apathy in the current process with interest and motivation to get involved in the political process at a truly grassroots level.

The electoral process I propose is a radical departure from the traditional implementations of representative democracy. It would eliminate the massive scale elections that happen today in the US where every adult in a district, state or the entire nation may vote amongst a small group of elite candidates to choose their representative. Some people may not even consider my proposal to be a form of democracy at all, though I see it as democracy in a new form better designed to scale to massive populations, and based on mathematical principles designed to eliminate all deeply flawed candidates through persistent examination. It also eliminates the separation between voters and candidates, making everyone an equal participant in the election process.

Tournaments and the Pursuit of Excellence

One method use in sports competitions to determine the "best" member in a fairly large group is a tournament. In a traditional tournament competition the complete set of entrants is divided into equal sized groups, which are themselves divided again, until each group contains two members. In the first round the members of each of these pairs face-off against each other in competition. The loser is eliminated, the winner moves on to the next round. The pairings for the second round are predetermined by the original divisions, e.g. if A-B-C-D are a group of four that is split into A-B and C-D, and A beats B and C beats D in the first round, then B and D are eliminated and A will face C in the second round. Eventually only two competitors are left in the final round, and the winner is the tournament champion. There are variations on this setup, but this is the simplest and most common method.

Excellence is required to win a large tournament. A competitor that falters at any step is eliminated, and the level of ability is raised each round as weaker competitors are removed from the pool. Weaknesses that may be hidden in a single face-off will be exposed given enough rounds. The winner of a tournament may not be the absolute best competitor in the field, but is almost certainly one of the best. Elections are like tournaments in that there is a single winner whom we hope will excel in their field. Unfortunately, the similarities end there. It is unlikely that most elected politicians are really the best individuals for the job due to the large barriers to entry in modern politics. It is likely that more than 99% of eligible citizens will not even consider running for most major offices, what are the chances that the person most fit for service is in the remaining one percent? I believe that an open tournament format which considers all participants as candidates, and in which competitors win by directly convincing their fellow citizens that they are the most capable of representing their interests, is much more likely than the current system of electing truly excellent individuals to office.

The Proposed Solution: Tournament Elections

I call the election process I have devised Tournament Elections. For any elected office the individuals eligible for participation in the election tournament are the same citizens who would be eligible voters in the current system. The district's pool of participants is subdivided geographically, not into paired groups as in a traditional tournament, but into a fixed number of groups of roughly equal size. Each of these groups is further subdivided into the same fixed number of groups until the remaining groups are too small to be further divided. The last few sets of group divisions at the neighborhood level will be done randomly rather than geographically, to reduce the immediate effect of any personal biases within particular combinations of people.

Each round of the tournament follows a common format. Each group meets in a designated area for a single discussion. During the meeting participants will discuss issues relevant to the office at stake as well as each individual's qualifications and platform. At the end of the meeting the participants will vote amongst themselves for the member they feel is best suited to represent the constituents of the group's region (i.e. the current participants as well as members in all previous groups each participant was chosen from). The chosen member moves on to the next round of the tournament, all other group members are eliminated. The member chosen from the final group, the tournament champion, is elected to office.

If for some reason a single group representative cannot be chosen in any round except the last, all members are eliminated and that group will not be represented in the next round. Likewise, in case a chosen participant decides to drop out of the tournament at any time the district she was chosen to represent will simply go without representation. If the final victor declines the office the job goes to the next highest selected participant according to the chosen voting method. In cases of unbreakable deadlock in the final round, a popular vote of all eligible citizens will be held to decide the winner.

Tournament Elections are obviously very different than traditional electoral processes. By completely eliminating the single vote per person principle, it may seem like this method is no longer democracy at all. However, the fundamental principle of a democracy, or more specifically a representative democracy, is that every citizen has equal say in selection of government representatives. Tournament Elections meets this criteria because everyone starts of on equal footing. How far an individual advances in the tournament depends on their political fitness in the eyes of their peers.

Some people might feel disenfranchised because they are eliminated in the first round, while others move on. However, I believe that a person eliminated in the first round of the tournament holds just as much influence as the average voter in a plurality election. In that first round every individual who wants to advance must sell their abilities to their own neighbors. In a group of just 5 to 10 people, every vote counts. While it is impossible to guarantee that a chosen representative will eventually attain office (even winning another round is unlikely) it is much easier to eliminate undesirable participants than in the current systems. Participants with no desire to advance further themselves will simply use their positions to determine the most appropriate member for group representation. Current elections make voters feel powerless, because except in the tightest possible races, one vote really does not matter. In Tournament Elections, every vote does matter, so every voter has real power.

Implementing Tournament Elections

The previous section gives an overview of my vision for a tournament election process, highlighting only essential points. Left out are numerous features which must be considered. In this section I discuss possible implementation details.

Meeting Format

Meetings are held at geographically central locations within each region. In the initial rounds there may be a meeting hall designated for the entire neighborhood, and the meeting times would be staggered for practical purposes. Employers and universities would be required to make accommodations for every participant to attend their scheduled meeting as is currently required in the US on election days. The time required for meetings would be fixed for each round and would likely start small and increase at each stage as the level of importance in the outcome increases. For neighborhood meetings four hours should suffice, in the later rounds for the most important political offices meetings may take place over two or three days. Likewise the time between meetings may grow, though not as greatly. Each round would likely take about one week. Early round neighborhood meetings would be spread out over the course of a week while later meetings would require several days in between to allow for travel time and preparation. In later rounds it is more likely that participants would communicate before the actual meeting takes place to learn about each other's basic platforms, agendas and philosophies so that the meeting time can be concentrated on more in-depth discussion.

Each meeting is held privately, but recorded for public review, available to anyone immediately afterwards. Voting record would also be made immediately public. Participants who move beyond the first round would be given priority access to ensure that they are able to review previous rounds of later group members. Local live broadcasting will be practical in later rounds, for early rounds (which may consist of hundreds or thousands of meetings within a short span of time) local public viewings may be arranged on closed circuits. The final rounds of national elections would be broadcast regionally or nationally. Because the electoral process is beneficial to the public good, travel and accommodations necessary for participation in meetings beyond the neighborhood rounds will be arranged by the government and payed for with public funds.

Voting Method

The voting method used to select representatives to move past each round is probably the most crucial implementation detail. The method must work well when the number of voters is equal to the number of candidates, and potentially every voter wants to win themselves. In addition, the chosen method should discourage strategic voting. A strategic vote is one which favors a weakly favored candidate with strong chances of success over a strongly favored candidate with lower chance of success. Voting methods which encourage strategic voting benefits established parties and discourages outsiders who may take votes away from candidates who actually share similar platforms. Strategic voting plagues the US system, as discussed in the introduction, making it difficult for third party candidates to contend.

Methods such as traditional plurality voting are right out. Many rounds could deadlock at one vote apiece. Maybe worse, in such situations a single member who doesn't care to move on personally automatically gets the deciding vote! Approval voting is not much better, though modified versions, such as each ballot requiring at least one non-self approval and at least one disapproval would make deadlocks less likely. After researching the most popular voting methods I have concluded that the Condorcet method would likely be superior. It is the most mathematically correct method known for choosing the ideal representative while discouraging strategic voting. For the Tournament process the Condorcet method may be modified to require that each participant rank all group members to prevent stand-offs from each participant ranking everyone else in the group last, although that would be such a risky strategy (the case where one member deviating from this approach could end up selecting the winner with a single vote) that the modification may not be necessary.

Group Size and Tournament Length

The standard group size is an extremely important factor is designing a Tournament Election process because it directly influences the number of rounds necessary and the manner of discussion possible in every meeting. The chosen size must meet two key criteria: the group must be large enough to sufficiently limit the number of rounds for major elections, and small enough to facilitate worthwhile discussion and reasonable representative selection. Fairly recently (in historical terms) the populations of China and India each surpassed one billion people. This forms a reasonable basis for the maximum expected number of eligible voters in a standard democratic election, or participants in a tournament election. The number of rounds in a tournament equals the logarithm of the number of participants with base equal to the size of each group. Looked at another way, the group size taken to a power equal to the number of rounds yields the maximum supported participating population. For one billion participants, groups of 10 requires 9 rounds (109 = 1,000,000,000), groups of 5 require 13 rounds (513 = 1,220,703,125).

If we assume one week per round, then 13 rounds would just fit within a 3 month span. This is quite a long time for a single electoral process, though doubling the group size (from 5 to 10) would only reduce the length by one month. I believe that 5 members is a reasonable lower limit to group size, and that 10 members or 13 rounds are reasonable upper limits to group size and tournament length, respectively. I personally favor groups of 8 as nearly ideal for round table discussion, while limiting the tournament to 10 rounds maximum with today's populations. With this size, the first 3 rounds would take place at the neighborhood level, with up to 512 participants per neighborhood.

One potential flaw with a fixed maximum group size is that depending on total population, the first round groups may not break down into appropriate sizes. For example, with maximum group size of 8 and neighborhoods of up to 512 eligible participants, it would be possible that neighborhoods may contain as few as 65 participants. Breaking up this neighborhood into 8 groups, and then breaking those groups into 8 more groups each would leave 63 groups of one and one group of two. One simple solution in this situation would be to reduce group size at the neighborhood level. In this case dividing by four yields groups of 16 or 17, and dividing by four again gives groups of four or five. Another solution would be to simply reduce the number of rounds by one and slightly increase the size of first round groups, because chances are that not all eligible citizens will participate in the first round (especially because participation in the tournament is a greater commitment of time than traditional voting), while selected representatives are much more likely to show up in later rounds.

Potential Drawbacks

Though I believe that Tournament Elections offer a superior alternative to current electoral systems, they are not without drawbacks. The first is the large commitment required by individual participants. People who see waiting at the polling place for an hour to vote as a chore may be less likely to want to spend several hours discussing politics instead. I believe this tradeoff is reasonable, given that those who do choose to participate have more real power in the system. Absentee participants would be more difficult to manage, though it should be possible to utilize modern telecommunications technology to setup absentee kiosks where citizens could participate in the designated meetings remotely. Unfortunately, this option may not be viable in some circumstances, such as citizens traveling abroad, and members of the armed forces stationed outside of the country. The potential benefits of Tournament Elections must be weighed against these flaws.

The organization and infrastructure for Tournament Elections would also be significantly greater than traditional elections, though not as much greater as one might suspect. Because the vast majority of eligible participants are eliminated after the first round, the first round would actually require more resources than all subsequent rounds combined. Therefore most of the increases in resources are directly related to the number of meeting locations required compared to the number of polling places, the span of several days compared with a single day election, and the equipment and operators to record, broadcast, and archive each meeting. These resources are significant, but not at all unreasonable for a potentially superior political system.

An End To Political Corruption?

I believe that Tournament Elections may hold the potential to end the influence of money and special interests in politics, and return influence directly to the people. Currently, winning the US Presidency is nearly equivalent to winning a single best-of-two election, where only those already with money, power and influence have any hope of becoming one of the two finalists. With Tournament Elections a candidate would have to win as many as ten best-of-eight elections. Back-door politicking would be extremely difficult, with discussions and voting a matter of immediate public record.

Campaign donations today are in effect bribery. Corporations and interest groups donate money to candidates to fund expensive campaigns in exchange for preferential legislation in the near future. Most politicians are always looking towards the next election, so they are compelled to pass laws desirable to these interests in order to keep the money flowing. In a tournament format, participants have to sell their platform to a small group of peers who are also their competitors, rather than the public at large, so campaign funds lose all power. The unpredictability of a large tournament compared with today's elections would make it nearly impossible for interest groups to donate money efficiently until later rounds.

Some may say that with every vote counting as heavily as it does within each group that votes will be for sale and bribery could run rampant. There are two obvious schemes in which bribery could be used to attempt to cheat the system: bribing a majority of the members of a single group in each round of the tournament, and seeding the first round of the tournament with enough bribed individuals to ensure that they control at least a majority vote in the majority of groups in every round. In short, both methods are extremely impractical at best. The first method will not work because it requires that even the last rounds of the tournament have groups with a majority of participants willing to sell their votes, despite these individuals being the best qualified and most determined to attain the office for themselves. The second method cannot work because of the massive amount of money and organization required to bribe a sufficient number of participants. Detailed analyses of both methods are available in Appendix B.

One problem Tournament Elections does not solve is the issue of corruption while in office. I believe that by eliminating the power of campaign financing, and with the inherent unlikelihood that an incumbent would be reelected, that these problems would be diminished. If the system lives up to its potential, those elected to office would be of higher moral character than today's politicians, and would stay above corrupt influences. In any case, the situation would be no worse than it is today. Additionally, political parties would be reduced in power, soft money would likely disappear, and private donations could only affect single terms, not long political careers. Indeed, the days of the career politician would likely end.


No political system is without flaws, and this would certainly be no exception. Though I believe it could root out corruption and influence it is possible that corruption would find other means, such as rigging the system to break down regions and randomly assign neighborhood groups. Strict controls and oversights would be necessary for these mechanisms. The number of elections that occur for different offices may cause issues, though by staggering election years the problem would not be too great. Other problems may occur with respect to citizen participation. For every person for which the new system brings interest and a sense of excitement to politics another will likely be turned off by the greater commitment required compared to voting every few years, even if that individual has no desire to participate past the first round. Others will dislike the fact that they cannot vote on the final representative directly. While this is true I believe we need a system that encourages active participation. Real participation, not just taking an hour to stand in line and vote but discussion and advocacy with people who have to listen because every vote counts at every stage in the game. This embodies the most cherished ideals of democracy. Those who care and want to have their voices heard would have them heard, and no one could possess the power to suppress a single voice.

This system is not likely to be implemented for a major election soon, if ever. For many people, it may just seem too different to consider seriously. however, I believe that reform is possible in incremental stages. First steps may be to reform the current voting methods. Some governments use Instant Runoff Voting in preference to plurality, I believe that Approval and Condorcet methods are even better choices. Smaller communities may be willing to experiment with alternative voting methods for mayoral or city council elections. Non-government organization may choose their own methods for electing officers. Regardless of whether Tournament Elections are the best answer for election reform, I believe it is important to always look at new possibilities, because existing systems can always be improved upon.


Appendix A: Implementation Variants

A number of alternatives to fundamental components of Tournament Elections have been suggested after review of the original version of this article. The most prominent are discussed in this appendix.

If a deadlock occurs in the final round, the winner is selected by popular vote rather than the intra-group used in all other rounds. An alternative method would be to always use a popular vote for the final round. The tournament serves only to select the eight finalists, and the final round discussion serves as a final debate. The popular vote may use any existing voting method, though the relatively large number of candidates suggests that Condorcet or Approval voting would be preferred over Plurality or Instant Runoff. I personally do not believe that a popular vote will yield better results; if the tournament process is successful in previous rounds in selecting the best representatives then all final candidates should be fit for office, and the final round selection should be successful as well. If the tournament method is not successful, then the entire system is a failure, and a popular vote at the end can only hope to select the least unfit candidate. Nevertheless, this alternative may ease the minds of citizens who feel disenfranchised by the Tournament Election system, by giving everyone a say in the final selection.

Taking the popular vote idea even further, some may like the idea of being able to vote in every round for their representative group. This would maximize possible participation by everyone but would require enormous resources and time commitments. Voters would have to learn eight new faces every week and watch hours of video to review each discussion. Most people will simply not be able to do this. The ability to distribute video and perform voting over the Internet could make this option technically possible in the future, but today would be virtually impossible.

Appendix B: Practicality of Bribery in a Tournament Election

The most straightforward way for a candidate to bribe their way into office would be to buy the votes of at least half of their group for each round. This may work easily enough in the early rounds, but would certainly fail in the later rounds for any major office. In each round throughout the tournament, participants will become successively more dedicated and ambitious. Whether they want the office for themselves or merely want to ensure that the best person is chosen for the job, the more rounds a person has won, the higher their price would get (if they are corruptible), or the more likely they would be to expose the briber (if they are not corruptible, or just want the office for themselves more).

A more sophisticated approach would be to seed the first round of the tournament with enough bribed individuals to ensure that they control at least a majority vote in the majority of groups in every round. Whether for small municipalities or large, national elections it would be impossible. For example, let us take a mayoral election for a small town, with 4,096 eligible participants and groupings of 8. The tournament would require 4 rounds (84 = 4,096). In the fourth and final round we would need 5 bribed individuals to guarantee the office is won. To choose those 5, we would need 5 x 5 = 25 bribed for the third round, requiring 25 x 5 = 125 for the second round, and 125 x 5 = 625 for the first round. That would mean successfully bribing 15% of the eligible voting population to guarantee victory.

Now let us look at a much larger election. Using the same groups of 8, we take an eligible population of 134,217,728 in a 9 round tournament (89 = 134,217,728), which would be sufficient for a moderately large nation. The majority needed to guarantee winning a round is still 5, so the total number of bribes will be 59 = 1,953,125. This is only 1.46% of the eligible participating population, proportionally much smaller than for the small town example, but a huge number in absolute terms. Few people have the financial resources, and no one has the organizational resources, to pull off a scheme of this magnitude, much less get away with it.

One more problem with both approaches is that you simply cannot trust someone who takes a bribe. Because the meetings take place under public scrutiny, bribes would have to be arranged beforehand, and money exchanged before or after. If money is payed beforehand, the person bribed can simply change their mind when casting the actual vote, and the briber is left with no recourse, at least legally. If the money is agreed to be payed afterwards, the bribed individual has no guarantees that the briber will pay up, because the vote cannot be rescinded. With these observations taken into account, it would be extremely difficult to significantly corrupt the Tournament Election process.

Matt Heinzen -- 2003.06.04

Speak to me about this essay.

Last Updated 2003.06.16