Grab bag of ideas to reform government

Slides for a talk that roughly corresponds to this page can be found at:

Here I discuss a number of ideas for reform of government. My focus is on the U.S. government, although almost all of the talk is applicable to most other governments, as well. This document focuses on structural problems and solutions, by which I mean, I focus on the basic structure of institutions, and on the decision-making procedures used (I focus on these because I believe that meta-issues are the most important).


The ideas discussed below have a number of goals.

Currently, political decision-making is heavily influenced by money, and in my opinion political decisions are often made on the basis of what entities with money want, rather than on the basis of "the greater good". I aim to reduce corruption and the influence of money on the political process.

Currently, voters are mostly only allowed to directly express their opinions about politicans or political parties through voting. But voters also have opinions about individual issues. I aim to allow voters to express their opinions on issues, too.

Currently in the U.S., only large political parties can compete, and currently, only candidates that "energize their base" can win. I aim to allow small political parties, as well as centrist candidates, to compete.

Currently in the U.S., a lot of decisions are (in theory) made by the Senate and the House of Representatives, which have 100 and 435 normal voting members, respectively. In my view, such groups are much too large for effective discussion and deliberation. I aim to transfer decision-making power to small groups.

Currently, much of the time that ordinary voters spend discussing politics is spent discussing it with other people who agree with them, other people in their faction (their political party). I aim to get people to spend more time discussing politics with people in other factions.

Currently, there seems to be a bias in the political process that leads to governments growing larger, more powerful, and more secretive over time. I aim to counteract this by adding opposing mechanisms that favor smaller, less powerful, more transparent government.

Currently, the public seems to pay less attention to important political issues than it might. I aim to increase the ability of the political process to attract the public's attention.

The 3 biggest problems

First, here are what I believe are the three biggest problems with our political system, and how to fix them with fairly conventional means.

  1. Money. Money has too much influence on politics.
  2. Vote-counting method. Our vote-counting method causes two-party domination and divisive rather than cooperative politics.
  3. Uncompetitive districts. Out of the 435 U.S. House districts, only about 25-100 of them are competitive, depending on how you define "competitive" [1] [2]. In uncompetitive districts, the incumbent is unlikely to lose an election (or at least, to lose an election to someone not in their party). Therefore, in an uncompetitive district, the politician has little incentive to do what the voters want; they just have to stay within the good graces of their political party.

Fixing money

The way to reduce the influence of money in politics is simple. Ban money. Don't allow any politician to accept any campaign contributions or gifts, period. All campaigns are publically financed (that is, the government pays for political campaign ads, etc).

Fixing vote counting

There are at least two classes of situations in which representatives are elected by election; single seat and multiple seat. In the single seat case, there is only a single office to be filled (for instance, the election of the U.S. president); in the multiple seat case, there is a number of similar offices to be filled (for instance, the election of a legislature).

Single seat vote counting methods

Our current system is called plurality (sometimes also called first-past-the-post). The plurality system is: each voter picks a candidate. The number of ballots for each candidates is counted. Whoever gets the most ballots wins.

There are at least two big problems with plurality.

Problem: vote splitting

The first problem is the possibility of splitting the vote or spoilers. If only one candidate appears to be competitive for faction A, and there are multiple candidates for faction B, then faction B's votes are "split" among the different faction B candidates. Faction A has an unfair advantage because faction A's candidate will get as many votes as there are faction A voters, whereas none of faction B's candidates will get as many votes as there are faction B voters. For example, if faction A has 40% of the voters, and faction B has 60%, but there is only one candidate for faction A and there are two for faction B, both equally popular, then each faction B candidate will only get 30% of the votes, and faction A's candidate will win with 40% of the votes -- even though faction B has 60% of the votes. When one of the multiple candidates is less popular than the other, the less-popular candidate is said to be a "spoiler"; a real-life example may the 2000 presidential elections in which it is thought that Nader and Gore may have split the left vote, giving the victory to Bush.

Vote splitting is clearly a problem in the short term because it leads to a less popular faction winning elections. In the long term, it is an even bigger problem because it prevents anyone but the top two front-runners from gaining a large number of votes, because voters don't want to "waste their vote" on a spoiler. This greatly contributes to two-party dominion, because it prevents third parties from gaining an appreciable number of votes (see also Duverger's law).

Problem: no compromise candidates

The second problem with plurality is that "compromise candidates" cannot win. An example of a "compromise candidates" is one who is no one's favorite, but everyone's second choice.

For example, imagine that every voter is in one of two factions, call them Democrat and Republican, that each faction controls about 50% of the voters, and that there are three candidates for President; Bush, a Republican, Gore, a Democrat, and Mr. Second Choice, who is a centrist. Republicans like Bush, don't mind Mr. Second Choice, and hate Gore. Democrats like Gore, don't mind Mr. Second Choice, and hate Bush. With plurality, all the Republicans will vote for Bush, and all the Democrats will vote for Gore, and no one will vote for Mr. Second Choice. Which is a shame, because if Mr. Second Choice were to be elected, no one would hate him, whereas what will happen instead is that either Bush or Gore will win, and either way about half the country will hate the president.

As with vote splitting, in addition to the short-term effects, this leads to a long-term problem. The long term effect is that politicians have no incentive to appeal to voters in the "enemy" faction. You have to either be the most republican-y Republican or the most democratic-y Democrat; it does no good to be the second best Republican, even if the Democratic voters like you pretty good at the same time. This means there are incentives to demonize the "enemy" party, and no incentives not to. When one party is in power, there are no incentives not to do things that the other party hates. Therefore, plurality makes politics more divisive, and less cooperative.

Solution: either range voting or Condorcet

There are many alternative vote counting methods that don't have the problems of vote splitting or elimation of compromise candidates. My two favorites are range voting and Condorcet (actually these are each a small family of methods). I won't go into them much here, but you can read about them at and

Currently, I recommend range voting. A simple form of range voting is this: for each candidate, each voter gives them a score of either 0, 1/2, or 1. These scores are summed up, and the candidate with the highest score wins.

Multiple seat vote counting methods

Our current system for electing congresspeople combines plurality with districts. Each congressional seat is assigned to a district, and a single-seat plurality election is held in order to deterimine who wins that seat. In addition to the problems with plurality, this has the additional problem that it does not provide "proportional representation". What that means is that the factional proportions in congress might not match the proportions in the electorate. For example, imagine if 60% of the voters belonged to faction A, and 40% belonged to faction B, and faction loyalties were evenly distributed geographically. In this case, using our current system, faction A would win every seat, and congress would be made up of 100% faction A. A better outcome would be if congress would be 60% faction A and 40% faction B.

An example from real life is the green and libertarian parties, which get a few percentage points of the votes but don't have any representatives in congress (because there is no single district in which they have a majority). Most likely, many more people would vote for these parties if doing so was not "wasting your vote"; my guess is that on the order of 20% of congress would be green and libertarian if we used a vote counting system with proportional representation for congress.

Soution: either reweighted range voting or Loring

There are many alternative vote counting methods that provide proportional representation. My two favorites are reweighted range voting and Loring. Currently, I recommend reweighted range voting.

In both of these methods, either single-seat congressional districts are eliminated and replaced by larger districts with multiple seats per district, or congress is augmented by the addition of at-large members representing the whole nation, or both.

Fixing Districts

There are multiple reasons why districts are uncompetitive. Part of the problem is clearly that politicians get to draw the district boundaries during redistricting, and they often choose to draw them in such a way as to make them less competitive.


When an electorate is subdivided into districts, the districts can usually be drawn in such a way as to influence the results of elections. This phenomenon is called gerrymandering.

For example, in 2002, the Republican-controlled legislature in Florida proposed a map with 18 Republican-leaning seats and 7 Democratic ones, even though in 2000, the national presidential election indicated that Florida is about half Republican and half Democratic [3]. Another example; as of 2002, illinois's 4th congressional district looks like this:

Such weird shapes are a clear sign that the district shapes are being drawn with some specific objective, like gerrymandering.

Other factors

However, gerrymandering is not the only reason for uncompetitive districts (some commentators believe it is not even the main reason [4]). For example, another cause of uncompetitive districts is geographic clustering of people with similar political views.


One class of solution is to fix gerrymandering. There are various proposals for this, for example, (a) give the task of redistricting to disinterested parties, (b) automate the task of redistricting according to some specified algorithm (example, thanks to Dana Dahlstrom: distort the map in such a way as to make the population density on the distorted map uniform, then draw a grid upon the distorted map, and finally inversely distort the grid in order to draw districts upon the original map), and/or (c) place various constraints on redistricting, such as requiring most districts to be similarly sized rectangles.

Another class of solution is to augment district representatives with at-large representatives. With a counting method that has the property of proportional representation (that is, if the at-large representatives are chosen in such a way so as to mirror the factional proportions of the entire nation), some of these seats will always be "competitive" because, with a sufficiently large number of at-large representatives, even a small fluctuation in the factional makeup of the nation will lead to a change of how many seats go to each party (by the same token, however, many of these seats will be "uncompetitive" in the sense that, unless there is a huge change in the popular will, each faction will continue to hold most of its seats).

The fix using at-large representatives has the advantage of fixing the uncompetitive district problem even to the extent that it is caused by geographical clustering rather than intentional gerrymandering.

I recommend introducing at least a few at-large representatives, and in addition adopting constraints on redistricting, or even automating it.

This concludes the section on the three biggest problems with our government. The rest of this document contains a grabbag of less conventional ideas. I think that all of the ideas that follow are 'feasible', meaning I believe that they would work well in practice, not just "in theory". However, in the following, I won't be concerned with the question of whether they are 'winable', that is, the question of whether it is at all likely that the political establishment can be convinced to try them.

Collaborative budgeting in initiatives

Problem (?)

California (and presumably some other places) allow voter initiative measures in which the voters can directly pass statues and constitutional amendments. This mechanism has been used to pass many measures that constrain budgets. For example, California Proposition 98 (amended by Proposition 111) mandates a minimum amount of spending on education, according to a formula (the formula may be suspended for 1 year with the consent of 2/3s of the legislature and the governor). Some commentators think that the patchwork of constraints is responsible for California's budget deficit although this is disputed by others who believe that the most of the money mandated by constraints would spent that way anyway [5]. Some of the commentators who think that constraints are the problem think this is evidence that direct democracy doesn't work and that the initative system should be abolished.


I am not convinced that initative measures are causing California's budget deficit, but assuming for the sake of argument that they are, I do not think that direct democracy needs to be abolished. A better solution is to design a voting procedure that is more likely to return a balanced budget.

Instead of considering each spending measure in isolation, voters should consider the whole budget at once. A specialized budget voting procedure should be chosen which results not in a list of English-language constraints, but rather, a list of, for each spending item, how much money should go to that item. A well-defined procedure will allow voters (or the legislature) to separately consider the questions of (1) how much money should be spent, and (2) what it will be spent on.

For example, here is a simple "collaborative budgeting" procedure. Each voter could submit a ballot with a complete budget. From each ballot, two types of information could be extracted: (1) the __proportion__ given to each item, and (2) the total amount of spending. To determine the actual total amount of spending, the total amounts of spending from the ballots could be averaged. To determine the actual amount spent on each item, the proportions from the ballots could be averaged, and then the resulting proportion multiplied by the total amount of spending. The budget obtained in this fashion would serve as a starting point for deliberation in the legislature. The proportional allocations from the budget obtained in this fashion could be amended a few at a time by the legislature, but the total spending amount would be binding (that is, each legislative amendment that increased spending on one item would have to decrease spending on a handful of other items to compensate).

The procedure given above is simple, but other procedure have been proposed which allow voters to express budget allocations in absolute terms, rather than proportions, for example: Tax rates could be set in a similar fashion. Instead of considering the total amount of spending, the procedure could use the amount of surplus/deficit.

Liquid democracy


The 1-bit problem

In our current system, every issue that comes up is discussed, argued, and analyzed many times over in countless small groups of acquaintences across the country. However, almost none of this thought is put to use, because when it comes times for these citizens to cast their votes, they generally are only asked one question: Republican or Democrat? By which I mean, most of the ways in which a citizen's vote influences the political system is through electing representatives, and usually there are at most two representatives who have any chance of winning, one from each party. The political system is based on inputs (votes) from citizens, however, all of a citizen's opinions and knowledge must be filtered into that 1 bit of information (perhaps I am being too harsh; sometimes you also get to vote in a party primary, adding another two or three bits).

Besides being a waste of thought, the 1-bit problem also wastes an opportunity to get people to care about the political system, because many people care about issues, not about politicans.

The black hole effect

In addition to voting, you can contact your representatives and give your opinion. You can email or call their office, and sometimes you will get something back from an intern indicating that the intern has received and processed your communication. Sometimes you may even get a noncommittal reply from the representative. The problem is that there is almost never any way to know whether or not your communication had an influence. It's like putting a suggestion into a suggestion box at a fast-food establishment; maybe someone's mind was changed; probably not; but you'll never know either way. This is called the black hole effect, because from your point of view, it's as if your advice was thrown into a black hole.

Imperfect solution: tell everyone to become an activist

You can do more; you can become an activist; you can get involved. And then maybe you'll see the difference that you're making. But most people have other priorities. They are happy to think about and discuss politics for an hour every now and then; they are happy to sign a petition; but that's all the time they can spare. If you are one of these people, you can see the (negligable) effect that your vote had, but you are only allowed to vote on one question. You can give advice on any other question, but you won't know if it had any effect. These two combine to discourage people from thinking about, or at least caring about, politics, because they feel that they can't (verifiably) make a difference.

Now, you may say, tough. Of course you can't become an activist just by spending an hour here and there; getting involved in politics, just like most everything else, demands a minimal amount of time. Most other fields of endeavor are the same way. You can't, for example, learn to play a musical instrument and join a band without dedicating some serious time. Same with martial arts. Same with higher mathematics. What's the problem? The problem is that, it seems like in a democracy, it is kinda necessary for a large part of the electorate to know what's going on, and to care, in order for the system to work well. While it may be nice for more people to play a musical instrument, to know martial arts, or higher mathematics, it seems much more crucial for more people to understand and to interact with the political system. It's not just for the individual's sake that we want to make it easier to interact with politics; it's for the sake of the political system, as well.

Imperfect solution: every citizen votes on every bill

The most obvious form of direct democracy would be for every citizen to vote on every bill. This isn't a good solution because people don't have time to learn about all the bills.

Solution: Transitive proxy voting

The idea of "transitive proxy voting" is also sometimes called "liquid democracy" or "delegable proxy".

The idea is that you can vote on any individual bill that you want, but for the rest of them, you can delegate your vote to any other voter. You can delegate votes to anyone, including friends and family. The delegations may be per topical area; so you can delegate science policy to one person, and foreign policy to another. The people who receive your delegated votes can re-delegate them. So, for example, you can delegate your science policy vote to your sister, who likes science, who can delegate it to a friend of hers, who is a student, who can delegate it to a professor that she respects, who can delegate it to a science policy expert.

This provides a procedure for citizens to cast votes on every measure , but it doesn't provide a procedure for citizens to draft measures in the first place. Existing initiative/referendum processes do that in theory, but in fact, using these processes is so difficult and expensive that most ordinary citizens can't participate in the drafting stage; furthermore, these processes are geared towards drafting by small groups, not by a massive collaboration of the entire electorate -- for example, there is no way for the electorate to amend the initial proposal -- only to vote yes or no. Transitive proxy can be combined with a better initiative referendum process (here's my proposal for that) to produce a mass legislature.


In our current system, a bill only needs to win more than half of the votes (called a "simple majority") in order to become law. Here, I argue that it would be better to require significantly more than half of the votes (say, for example, 60%) -- this is called a "supermajority".

Consensus decision-making and simple majority rule are two different group decision-making methods, each with their own advantages and disadvantages. Consensus decision-making is said to produce better decisions and to create less enmity between factions. Simple majority is said to be much faster. A supermajority-based system can be thought of as a middle ground between these two extremes.

Under the current simple majority system, a faction with 51% of the votes gets 100% of the power. This seems unfair -- for example, the majority faction can pass laws that spend 100% of the people's money on something that 49% of the people don't agree with. A supermajority would at least lower the maximum propostion of people who disagree and yet who are forced to contribute anyway. With supermajorities, the government simply would not take action on issues for which there is not a broad agreement on which action to take.

Under the current system, the Republicans can one day have 50.1% of the votes, and use this to direct policy in a certain direction. Then after the next election, the Republicans may lose 0.2% of the votes to the Democrats, and now the Democrats have 50.1%, and they can take policy in the opposite direction. So, only 0.2% of the voters have switched sides, but policy completely reverses. This flip-flopping of policy seems undesirable. If, say, 60% were needed to pass a bill, then 20% of the voters would have to switch sides before a policy could be replaced by its opposite.

Under the current system, a large faction with a simple majority doesn't have to compromise with other factions (except perhaps as demanded by the cloture rules in the Senate -- which involves supermajorities). This is bad because it means that large factions have little incentive to work together -- and in addition, that a majority faction has no incentive not to do things that voters in other factions absolutely despise. In our current system of two-party domination, it is always the case that either the Democrats or the Republicans have a simple majority, so this stuff happens all the time. With a 60% supermajority, this would only happen when one faction controls 60% of the votes. The rest of the time, even a majority faction would have to work with at least a few of the other faction's members in order to get anything passed. This would lead to more inter-party cooperation, and in addition would force majority parties to pay a little bit of attention to the will of voters in other parties.

If a supermajority system is used, it can be tweaked in order to accomplish specific objectives. In my opinion, the current political process contains a bias that leads to governments growing larger, more powerful, and more secretive over time, and that leads to armed conflict between nations. With a supermajority system, the usual supermajorities would be required in order to pass bills that make the government larger, more powerful or more secretive, or to arm or mobilize for war. However, one could create a special exception in which only a simple majority would be required to make the government smaller, less powerful or more transparent, or to disarm or demobilize. This would introduce a countervailing bias.

Above, I said that with supermajorities, the government simply would not take action on issues for which there is not a broad agreement on which action to take. I think this is the correct decision in most circumstances; often the costs of a decision which alienates a substantial proportion of citizens is greater than the costs of doing nothing. However, there are undoubtably some issues on which some action is better than none. For example, some sort of budget must be passed every year. This could be dealt with by designating budget bills as a special case; in order to simply re-pass the same budget as was passed last year, only a simple majority would be needed. The supermajority would only be needed to modify a budget.


Problem: many procedural checks are not enforced

I am dismayed when I see our government violating our civil liberties, or otherwise ignoring procedural checks, and even more dismayed when nothing is done about it. The problem is that it is difficult for ordinary citizens to hold the government accountable for violating procedure. In order to take the government to court, a citizen must usually have "standing", a legal concept. One way in which a citizen can often fail to have standing is if a citizen can't show a specific harm to themselves resulting from the illegal government action. But often violations of procedure have vague, diffuse (yet very large) long-term harms, rather than a provable harm to a specific individual. Other times, in the case when the violation of procedure took the form of surveillance, the citizen does not have access to the information of who specifically was surveilled, and so even when s/he was specifically targetted, s/he cannot prove it. And other times, the citizen has standing, yet they are prohibited from taking the government to court for national security reasons.

The courts have the authority to force the government to obey procedure, but they are not permitted to act unless a case is brought by an entity with standing, and even then sometimes they do not act due to fears that the prosecution of the case would reveal secret information that would hurt national security.

In theory, congress has the power to keep the executive branch in check in these situations. But in fact, congress almost never takes action in such situations, because in order to do so would require months of wrangling and jumping over procedural hurdles, and because invariably a large proportion of the public will support the government's side in any case. If there's one thing the public doesn't like, it's watching congress fight itself for months at a time over some abtruse issue. Doing this would make the incumbents less likely to win future elections, therefore they don't do it. Therefore, political realities prevent the only entity that is empowered to rein in the government in these situations from doing so.

Problem: secrecy

A related problem is secrecy. Often we are told that the government is making some controversial decision on the basis of secret information, and assured that we would agree if only we could know what they know. Often, we learn that some information was kept secret ostensibly on the basis of national security, but actually because its disclosure would show that the government had done something wrong.

In addition to these obvious harms, secrecy contributes to the long-term problem of conspiracy theorizing; without sufficient information to prove that the "official story" is correct, various people begin to believe various alternative hypotheses, leading to diverging worldviews between different factions, views which fundamentally contradict each other. This makes it hard for people to even discuss things with other factions, much less agree on what should be done. In other words, even if the official story is correct, even if the government isn't hiding anything bad, even if we really would agree with the government if we knew what they knew, still secrecy has a large cost in terms of causing unnecessary dissent and conflict between factions.

Despite these costs, there are few entities in the current system with incentives to prevent unnecessary secrecy. Even congresspeople who are briefed on classified matters are not allowed to decide that what they just heard should be declassified. Only the president can do that. And the president rarely has any incentive to do so.

Solution: tribunes

I propose the creation of a new branch of government with a watchdog function. I call it, "the office of the tribune". A tribune would be a combination of special prosecutor and investigative journalist. The tribune would only have three powers:

(1) Investigate government. Like the president, the tribune automatically has the highest possible security clearance, and is empowered to demand any information from any government agency or official. Their requests for information cannot be denied for any reason.

(2) Declassify information. Like the president, the tribune has the power to unilaterally declassify information.

(3) Prosecute. The tribune has the standing to take the government to court over any issue relating to infringement of rights, corruption, or a violation of procedure.

There would only be one or a handful of tribunes, and it would be a high-profile, directly elected position. Like a prosecutor or investigative journalist, a tribune would acquire repute by exposing and reining in government; the office of tribune is fundamentally adversarial to the government. The tribune is NOT permitted to take individual citizens to court (unless they are bribing a government official or something similar), or to prosecute citizens for ordinary laws such as theft, murder, etc. Since the only thing that a tribune is empowered to do is to rein in government, I expect that often, people would run for the office of tribune on a campaign platform that the government needs to be reined in more. In this way, a position would be created that is both empowered to rein in government power and secrecy, and also has an incentive to do so.

Really small legislative bodies

The House of Representatives has 435 normal representatives, and the Senate has 100. I think that deliberation and decision-making should happen in small groups. Now, it is true that in the present system, actually most of the discussion and deliberative decision-making happens in smaller groups; in congressional committees, and among the senior members of each political party. But the way this is accomplished is no good, because the voters don't get to choose who gets on the committees or which politicians will be in charge of the parties. So you end up with situations like when some issue that you care about is being decided by a small senate committee, none of whose members are your representative (and who won't even read your emails because you are not their constituent). So, I should add that I think that decision-making power should reside in small groups who are directly chosen by the voters.

So, I think we should replace things like the House of Representatives and the Senate with small groups of size close to 11 people.

There are two main benefits to this. First, you'll get better deliberation. With only 11 representatives in the House, each representatives would have time to spend many hours each month talking to each other representative. They would be able to have discussions with one another about political philosophy; they would be able to go into great detail during discussions; they would get to know each other's points of view very well.

Second, the citizens will pay more attention to the legislature. Right now, there's only about 11 or so congresspeople who are consistently talked about in the national media anyway. So, it seems like the people's attentional focus only has "room" for a few congresspeople. In addition to making it easier to follow the legislature, small groups will make the citizens more interested in the legislature. Once the public gets to know the personalities of those 11 people, everything will seem more dramatic -- the doings of congress will become interesting in the same way that a soap opera is, that is, because it is interesting to hear about arguments and changing alliances between people whom you know a lot about. So, both because it is easier and because it is more interesting, with small groups, the people will pay more attention to the legislature.

Separate domestic and foreign representatives

It seems to me that the citizens care more about domestic issues than foreign ones. Consequently, they choose leaders who they will serve them well domestically, even if they are not very good on foreign issues. This explains why we elect presidents and congresspeople who don't know much about foreign policy before they take office.

One solution is to have separate representatives for domestic and foreign affairs. Then, voters can elect their top choice for domestic issues, even if s/he doesn't know much about international issues, without having to put the same person in charge of foreign affairs.

Optimistic concurrency

Currently, both the House and the Senate must pass each bill before it can go to the president and become law. This process could be streamlined, while still preserving checks and balances, by allowing a bill to become law after passed by only one chamber, __unless__ the other chamber objects.

Hierarchial voluntary constituencies

"Hierarchial voluntary constituencies" is a different way of electing representatives to a legislative body.

Voluntary constituencies

First, the "voluntary constituency" part. Instead of each voter being placed into a set district, each voter __chooses their own__ district. In this scheme, districts (which are referred to instead as __constituencies__) do not necessarily have anything to do with geographic boundaries; each constituency is simply a group of people who has decided to band together and elect a single representative. Any citizen may at any time leave their current constituency, and join some other constituency, or even create a new constituency along with a group of other citizens.

Constituencies have a certain minimum size; for example, one person cannot be their own constituency. Constituencies also have a maximum size; this ensures that a nationally recognized politican cannot create a constituency so large that its individual members effectively have no influence over hir.

It is expected that people who share political beliefs or interests will group together into constituencies. Many people may choose to group based on geographic proximity; therefore, a "voluntary constituency" system provides a way for local interests to be served, just as the current fixed-district system does; one might see, for example, a "Del Mar" constituency. Other people may choose to group together based on party affiliation, for example, a constituency named "Republican chapter #5341". Still other people may choose to group together based on groups of issues that cross traditional party lines; for instance, a "fiscal discipline" constituency that advocates reducing the deficit. Any of these may be combined; for example, one might see a "Del Mar Republicans" constituency, or a "Democrats for fiscal discipline constituency". Voluntary constituencies also provide a mechanism for groups of dissenters within a political party to have a voice; for instance, one might see a "Republicans for marijuana legalization" constituency.

In this scheme, the representatives are called __delegates__. Each constituency elects one delegate. The constituency's delegate is __recallable__, meaning that the constituency can vote to replace its delegate with another person at any time, not just at pre-specified election times.

All of the delegates from all of the constituencies together form the legislature.


In a medium-size organization, the structure just described may be all you need. However, in very large organizations, either the total number of constituencies would have to be very large (and since each constituency has one delegate, that means the number of representatives in the legislature grows unweildy; but as noted earlier, we'd rather have a very small number of people in the legislature), or the number of people in each constituency would have to be very large (causing problems such as the way that each U.S. Senator serves so many people that they don't have time to talk to you, and they don't really need your individual support anyway). This is where the "hierarchical" comes in.

Instead of putting all of the delegates into the legislature, let the delegates themselves play the part of the voters in a second level of voluntary constituencies. That is, the delegates group themselves together into 2nd-order constituencies, and each 2nd-order constituency chooses a 2nd-order delegate. For example, one might see both of the delegates from the "Del Mar Republicans" and the "Republicans for marijuana legalization" together join a 2nd-order constituency of "Southern California Republicans".

These 2nd-order delegates could themselves form the legislature, or the process could repeat again to produce 3rd-order delegates. The total structure forms a "delegate pyramid", with the voters on the broad bottom level of the pyramid and the legislators at the top. However, the delegates in the middle are not powerless. For example, the 1st-order delegates (who represent the base voters' constituencies) can always recall and replace the 2nd-order delegates representing them. Because of the constant threat of recall, each delegate will be expected to constantly consult with the members of the constituency that elected them.

The advantage of such a hierarchial system is that one can have both a small number of legislators at the top, and also a relatively small number of members in each constituency. The advantage of a small number of members in each constituency is, first, that delegate do not have to do mass campaigning; your delegate will be someone who attends group meetings with you, someone who has time to talk to you; someone who is a good friend of an acquaintence of yours; not someone who you see on T.V. and who has to be judged by soundbites. In other words, with small enough constituencies, there is no need for mass campaigning, and politicans can be chosen based on sounder criteria than who is good with television appearences. Second, because the delegate __can__ communicate with their individual constituents, they will be __expected__ to; if the delegate doesn't answer individuals' phone calls, and make time to explain to them why they voted a certain way, and let their constituents try to convince them of a point of view, then the constituency will most replace them with someone else who does.

It seems to me that it would be best to have the lower constituency sizes larger, allowing one to make the higher constituency sizes smaller. For a nation approximately the size of the U.S., with about 200,000,000 people eligible to vote [6], one could have a delegate pyramid with 4 levels, with approximately 1285 people in each voter constituency, 262 delegates in each 2nd-order constituency, 53 delegates in each 3rd-order constituency, and 11 people in the legislature.

What it would be like

In other words, you would get together with about 1,300 other people with similar political views (they can live near you, or they can be geographically dispersed), and elect a delegate. All the delegates would form themselves into groups of about 260 delegates, and each of those groups would elect a 2nd-order delegate. All of THOSE delegates would arrange into groups of 53 2nd-order delegates, and each of those groups would control one spot in the legislature.

This means that your delegate would only have about 1,300 constituents to answer to. Compare to the current average size of districts of U.S. House Representatives; 700,000 people (not all of these people are elgible to vote; but if only half of them are, that's 1,300 vs. 350,000). You won't be having lunch with your delegate every week, but s/he will be worlds more accessible than your current House Representative. Furthermore, since you got to choose your constituency, you probably chose one with people who share your concerns; so a lot of those 1,300 have very similar interests and beliefs as you, making the impact of that number not as big as it seems. The numbers get even smaller as you go higher up; each legislator directly answers to a constituency of only 53 people. This small number will enforce frequent communication and serious accountability, and will ensure that legislators are chosen by people who have personally evaluated them, not just seen them on T.V..

One criticism of a system with very small legislative bodies is that you would like to have at least a few legislators who are specialists on every issue. In a delegate pyramid, this specialization function can be fulfilled by the 53 2rd-order delegates behind each of the 11 legislators.


If you are interested, I've created some formulas that I believe give reasonable answers for how many levels there should be and large the constituencies at each level should be for any size group. They can be found as part of a larger document, describing some proposed rules of operation for a delegate pyramid (that document is part of a larger set of bylaws that aren't necessarily relevant here; but most of that document is just about the delegate pyramid).



The group polarization effect is that, if people who already share similar views discuss something together in a group, then afterwards the individuals' viewpoints tend to me more extreme in the direction of whatever the group already agreed upon. It follows that if you are trying to acheive consensus in a large population, the worst thing you can do is to segment people into groups according to viewpoint and then have people discuss the issues with others who already share their views. Yet, this is exactly what happens in the current system (and in the voluntary constituency system proposed above) -- people tend to group into political factions, and most discussions about political issues are between people in the same faction.

To counteract this, it is desirable to get individuals from different factions talking to each other. In the current political system, the elected politicians (are forced to) talk to their counterparts in other factions, but ordinary citizens do not.

Solution: cross-councils

Create deliberative groups that include a representative sampling of citizens from different factions. Give these groups an incentive for reaching near-consensus on issues. The groups proposed here will be called __cross-councils__.

If a delegate pyramid (hierarchical voluntary constituencies) is used, then the following method may be used for creating small groups with a representative sampling of citizens from different factions. Let's say that the number of delegates in the legislature (the top layer of the delegate pyramid) is 11. We will create small groups, each of size 11. For each citizen, trace their representatives up through the delegate pyramid; that is, first find that citizen's constituency, then find that constituency's delegate, then find that delegate's 2nd-order constituency, then find that constituency's 2nd-order delegate, etc, until the trail stops (if some delegate is not a member of any constituency), or until you reach the top of the delgate pyramid, in which case you have associated that citizen with one of the 11 delegate legislators.

After doing this for every citizen, almost every citizen is associated with one of the 11 delegate legislators. This is like dividing the population of citizens into 11 factions. Now, all you have to do is assign the citizens into groups such that each group contains one citizen from each of these 11 factions.

You would only make cross-councils out of those citizens who are interested in serving on one. It's possible that some citizens would want to participate, but not be able to be assigned, because each group must have one citizen from each faction, and there could be a shortage of volunteers from one of the other factions.

Now, how to motivate the citizens in these groups to actually talk to and compromise with each other, and come to some sort of rough consensus? One way is to give them political power, conditional upon reaching near-consensus. So, give each 11-person cross-council one vote within some legislative chamber. Specify that a cross-council cannot cast their vote without the consent of at least 10 of the 11 people.

Putting it all together

Each of these ideas can be implemented in isolation; however, if you are interested in seeing an idea for a decision-making system which implements all of these ideas, see Fluid Democracy. Fluid Democracy is not just for governments (although it could be used for a government), but is rather a generic group structure and decision-making procedure that could be used by any group.