2015/03/24

Darwinian government

A better way to govern ourselves than simple elections might be called a “Darwinian” (or "Darwinoid") process, similar to what is done using genetic (or evolutionary) algorithms. Perhaps the best historical example of that was the Venetian system. It could be implemented in many ways. Consider one design:

1. At the precinct level (using the U.S.model of equipopulous precincts) two equal sized panels are selected by lot, or sortition. Then they select a pool of candidates to the next level (ward) by each panel voting for the best ten percent of the members of the other panel, and together for an equal number of individuals from outside either panel.

2. Candidates from the precinct pool are drawn at random to form two equal sized ward panels. The process is repeated to select two equal sized district panels.

3. The process is repeated to form two equal sized state panels (unless there are political subdivisions in between), and again to form two equal sized national panels.

4. The process is repeated to select a small number, say nine, candidates for the one national pool, from which a single official is selected at random.

A similar multi-step process would be used to select legislators, judges, administrators, etc.

In the judicial track, pairs of grand juries would select members of the next grand or any trial juries, and pose the questions they are to decide, after consulting with witnesses.

So random selection alternates with fitness election in a way that should enable the best and brightest (who don’t necessarily want the job) to bubble up to the top. Each participant in the process has an incentive to vote for the best rather than just a fellow partisan, because an obvious partisan would be less likely to survive to reach a higher level.

Voting rules within panels would use super majority votes, approval voting, or some other alternative to first-past-the-post.



A well-designed constitution should be able to combine many processes with many advantages, including both impartiality and prevention of the domination of the political process by factions, and also a statistically-representative microcosm of the entire citizen body, in order to discover what everyone would think under good conditions. There is no fundamental reason why those elected under some “principle of distinction” must be or become unrepresentative, if it is done in the right way. Having two legislative houses, one selected by sortition and the other by election, might be one way to do that, but we also need processes for executive and judicial questions. I have proposed such solutions at http://www.constitution.org/reform/us/con_amend.htm .

Concerning the Republic of Venice see
http://www.rangevoting.org/VenHist.html
http://www.uni-mannheim.de/mateo/camenaref/cmh/cmh108.html#275

This is not arguing there is a uniquely “best” choice, but there are certainly some that are better than others. The most important decisions that a polity must make are not just matters of taste or fair distribution, but of survival, that is , fitness. There can be more than one way to survive, perhaps giving rise to a fork in evolution, into two or more different lines, but most of the possible decisions lead to extinction. To build on the Platonic analogy, a captain may take his ship safely to any of several favorable ports, if more than one exists, but only one may be within range, and unless he is not very, very good at sailing, he may also take it to the bottom of the sea. Navigation is more a matter of skill, not of luck, taste, or fairness.

A reference can be made to the utilitarian rule of “the greatest good for the greatest number”, but that is too simple. It is the greatest good for the greatest number that is not too inequitable and that does not risk survival. (There is always some risk with any choice, but an optimal strategy improves the odds.)

Representivity is a value, for decisions about taste or distribution (justice), but not the uniquely highest values, which are honor, liberty, and aggregate prosperity, which are always at risk, even if it seems only taste or distribution are at stake.


The Universe is not organized for our comfort or convenience. It allows us to survive, sometimes, for a while, but survival is always nmarginal, and most decisions are a course along the edge of survival, trying to hang on as long as we can. We live in a unique period of easy prosperity when it seems good times will go on forever, however foolish our decisions might be. But that time is not forever, and we can easily bring it to an end sooner than necessary.

Most proposals for some kind of diarchy between advocates and judges neglect to consider that if elites control who may advocate what and how, who may judge, and how the deliberations may be structured or conducted, and what procedures are to be used, then those elites may still steer the “judges” to get any outcome they want.

This proposal does not conflate the functions, but divides them into balanced bodies, and allows darwinian-selected bodies to decide not only judgments but also the questions, structure, and procedures, for themselves and other bodies. There is no one single decisionmaking body, but a complex system of bodies that check and balance one another, while moving decisions to a final stage and to implementation.

So, for example, grand juries can supervise the selection of trial juries and successor grand juries, rather than leaving that to professionals like judges or prosecutors. Grand juries would not just decide whether to indict, but appoint the prosecutor (who could be a private citizen). Jurors would hear all the arguments made to the judge in a trial or hearing, so that they could condition their verdict on what they observed.


Several terms might be used for a system that alternates between random selection and merit screening:

  1. darwinition
  2. eduction
  3. winnowing
  4. winition
  5. fetura — Latin for breeding

How it might be realized

Nothing is going to happen unless or until we have a lot of successful exercises of it at the local and private levels. The Morena party in Mexico provides an example of that for nominating candidates. Election itself was not adopted until it had been successfully tried in many exercises over decades or centuries.

A good place to start would be in public and private schools. That is where most of us first learned at least a simplified version of Robert’s Rules of Order, but that started with the election of class officers, so the indoctrination began early. I can imagine that using a kind of Venetian system might be popular among students.


So it might take a generation or two, and we may not have that long, but we do what we can.

There are several ways sortition could be done in the U.S. without amending the Constitution, although after it was working for a while we might want to amend the Constitution to entrench it:

(1) Selecting the electors of the Electoral College. The method we use now is prescribed by statue in every state. No reason those have to be selected by a popular vote for slates of electors, but might need to begin by changing the law to require electors be elected individually and separately rather than by slate. That might prepare the public for a better way.

(2) Selecting the nominees for U.S. Senate. The 17th amendment now requires a popular election, but says nothing about nomination. Sortition (or more precisely, fetura/eduction/winition — a Venetian-like method) could be used to narrow the alternatives to two, although write-ins would have to be allowed.

(3) Selecting the members, or at least the nominees, for U.S. House of representatives. Even if the statute requiring single-member districts were not repealed, this could be done in each district. Perhaps just for nomination, with election for the last two.

(4) Appointing federal judges not to a particular court, but to a general pool from which they would be randomly assigned to courts for one term, and to cases until the case is decided, including the Supreme Court.

(5) Use eduction for appointing and assigning civil servants, leaving them in one assignment long enough to learn their jobs, but not long enough to “build an empire” there. (This is now done for military personnel.)

(6) Use fetura to appoint and assign most congressional staffers to members, to avoid having them becoming the real power in each house.

(7) Use winition to appoint members of all kinds of boards and commissions.

(8) Require all states to use sortition to select members of grand juries.

(9) Use sortition to select a successor to the president and vice-president in case they both leave office prematurely.


Every one of these appointments is now done according to statute and could be done differently without constitutional amendment.

Adoption of any such reform may be unlikely unless or until we undergo a traumatic event that is blamed on the electoral process. (Although one might think we are already in the midst of such.)


Donate Now!

7 comments:

Jon Roland said...

I advocate what might be called a “Darwinoid” process, similar to what is done using genetic (or evolutionary) algorithms. Perhaps the best historical example of that was the Venetian system http://www.constitution.org/elec/sortition_knag.htm It could be implemented in many ways. Consider one design:

1. At the precinct level (using the U.S.model of equipopulous precincts) two equal sized panels are selected by lot. Then they select a pool of candidates to the next level (ward) by each panel voting for the best ten percent of the members of the other panel, and together for an equal number of individuals from outside either panel.
2. Candidates from the precinct pool are drawn at random to form two equal sized ward panels. The process is repeated to select two equal sized district panels.
3. The process is repeated to form two equal sized state panels (unless there are political subdivisions in between), and again to form two equal sized national panels.
4. The process is repeated to select a small number, say nine, candidates for the one national pool, from which a single official is selected at random.

A similar multi-step process would be used to select legislators, judges, administrators, etc.

In the judicial track, pairs of grand juries would select members of the next grand or any trial juries, and pose the questions they are to decide, after consulting with witnesses.

So random selection alternates with fitness election in a way that should enable the best and brightest (who don’t necessarily want the job) to bubble up to the top. Each participant in the process has an incentive to vote for the best rather than just a fellow partisan, because an obvious partisan would be less likely to survive to reach a higher level.

Voting rules within panels would use supermajority votes, approval voting, or some other alternative to first-past-the-post.

Jon Roland said...

Trial jurors don’t know the judge, the lawyers, or the parties, when a trial begins. They would not be allowed to serve if they did. The job of a panel would be to get to know the opposite panelists in a similar way, through investigation and interrogation, including into their reputations. Any merit screening depends on people having reputations of some kind. No way to avoid that. But it is a solvable problem, given enough time and focus.

Jon Roland said...

Jurisprudence, with a body of predictable custom, policy, and practice, is a needed institutional memory, and one of the objectives of sound constitutional design is to enable needed changes in that system, but not too fast or too carelessly. That mostly means incremental changes rather than sweeping ones, although we are discussing a sweeping one to get out of the cul-de-sac in which we seem to find ourselves.

One of my motives in making such proposals for Darwinian decisionmaking is that in most countries, and especially the U.S., our jurisprudence has become corrupt and dysfunctional. The present design has not been working very well, and elections are a large part of that. Perhaps we can find a better one, using evolutionary processes as models. Obviously, biological evolution is very different in many ways, but there is more commonality than most students of the subject now seem to realize.

Jon Roland said...

Members of a trial jury definitely have to get to know the parties before them, ar at least certain attributes of them: the credibility of witnesses, the honesty and competence of the attorneys, and the fairness and competence of the judge. All these assessments weigh into their verdict. In most criminal cases they may have to only decide whether the accused is guilty, but there are often multiple charges, and they may have to also determine the sentence. In civil cases they have to decide who gets how much of what relief is being sought, and that can involve very subtle assessments of the parties before them.

The Darwinian process I suggested is to select policy deciders, not policy positions, except insofar as the deciders represent certain policy preferences. Once a decisionmaking body is selected, they can’t be prevented from debating and negotiating among one another if there is enough time to do so, and they will tend to take as much time as they need to settle on their decision. Of course, the decision rule they use need not be a simple majority. They can have rules that require supermajorities, at least for more important decisions.

I have not seen a process as elaborate as the one in my proposal, but I have observed or been involved in simpler but somewhat similar versions of such a process, especially large private organizations, such as the two major U.S. political parties nominating candidates and developing a platform at national conventions. My experience with those was in Texas, for which a primary election is the starting point of a delegate selection process.The rules for those two parties are a lot more complicated than my proposal, although the only role in them for randomness is a coin toss to decide ties. They get to know one another by conversations and speeches at the conventions at each level. Some such process inevitably seems to arise when a large body of members have to arrive at a small, roughly representative, number of those that make a final decision. But there is deliberation at every level, albeit not always very thorough.

How about examining the merits of the details of my proposal, and perhaps coming up with other proposals with comparable detail. We need to descend from the heights of abstraction down to specific procedures that might be tested in some willing organization.

Jon Roland said...

I submit that a well-designed constitution should be able to combine many processes with many advantages, including both impartiality and prevention of the domination of the political process by factions, and also a statistically-representative microcosm of the entire citizen body, in order to discover what everyone would think under good conditions. There is no fundamental reason why those elected under some “principle of distinction” must be or become unrepresentative, if it is done in the right way. Having two legislative houses, one selected by sortition and the other by election, might be one way to do that, but we also need processes for executive and judicial questions. I have proposed such solutions at http://www.constitution.org/reform/us/con_amend.htm .

Concerning the Republic of Venice see
http://www.rangevoting.org/VenHist.html
http://www.uni-mannheim.de/mateo/camenaref/cmh/cmh108.html#275

Jon Roland said...

I am not arguing there is a uniquely “best” choice, but there are certainly some that are better than others. The most important decisions that a polity must make are not just matters of taste or fair distribution, but of survival, that is, fitness. There can be more than one way to survive, perhaps giving rise to a fork in evolution, into two or more different lines, but most of the possible decisions lead to extinction. To build on the Platonic analogy, a captain may take his ship safely to any of several favorable ports, if more than one exists, but only one may be within range, and unless he is very, very good at sailing, he may also take it to the bottom of the sea. Navigation is more a matter of skill, not of luck, taste, or fairness.

An earlier reference was made to the utilitarian rule of “the greatest good for the greatest number”, but that is too simple. It is the greatest good for the greatest number that is not too inequitable and that does not risk survival. (There is always some risk with any choice, but an optimal strategy improves the odds.)

Representivity is important for the values of taste or distribution (justice), but those are not the highest values, which are honor, liberty, and aggregate prosperity, which are always at risk, even if it seems only taste or distribution are at stake.

The Universe is not organized for our comfort or convenience. It allows us to survive, sometimes, for a while, but survival is always marginal, and most decisions are a course along the edge of a cliff, trying to hang on as long as we can. We live in a unique period of easy prosperity when it seems good times will go on forever, however foolish our decisions might be. But that time is not forever, and we can easily bring it to an end sooner than necessary.

Jon Roland said...

Need a name for what I am proposing, to include both random selection (sortition) and merit screening (election), in a multistage process that involves multiple bodies that check and balance one another.. I am considering the following:

“darwinition” -- By analogy to breeding or natural selection.
“eduction” -- To draw out.
“winnowing” -- but this seems too tied to agricultural methods.
“winition” -- dropping the “dar”, suggested by “winnowing”.
"fetura" -- Latin for breeding

Translate

Search this and affiliated sites

Blog Archive