Charlie Douglass: Guaranteeing gender equality

A post by Charlie Douglass.

Democracy we are told is government by the people. What if many of the people aren’t involved? What if some have a much stronger voice than others?

If a government is to act on behalf of people of every colour, gender, age, and wealth status, then we must either rely on politicians ignoring self-interest completely, or find a way in which a house of parliament can have its members chosen very differently to today. Bias matters because today our parliaments under-represent many groups. Women are under-represented and although some change is happening, the change is slow. Why should the men who are already there be the ones who decide how fast the change is? Why is there so little effort to represent people from below the poverty line?

If you could draw equal representation from across society into a house of parliament, then self-interest matters much less. If a parliament was half female, half male, and contained people from every type of wealth status, then many really important issues would see quick and effective change. Poverty has the attention of many of us, but without power not much is changing. When there is a vote which gives voice to all, then the bias of media barons and political donations can be put in the trash-can where it belongs.

In many countries politics has two houses of government, and the proposal here is to replace the one is which the nation’s leader does not stand. In the USA, France and Australia this is the Senate. In the UK this is the House of Lords, and in Germany it is The Bundesrat. After two terms perhaps this new senate could replace both houses, or the second house could be replaced by sortition also – forming two houses.

The new Senate would have 200 members (the reason for 200 is explained later). Replacing a government level is much better than forming a new one. Existing levels of government are sure to have more power than the new one. The new boy will be bullied. Many monkey wrenches are sure to be thrown to unfairly undermine sortition in an attempt to ensure the idea does not take hold. Replacing the Senate means the new boy has presence and power, and this is vital. Some countries don’t have a popular vote for the senate at all, in the few that do voting results are leaving everyone including the politicians confused. This general distaste of the current system should be used to motivate people towards a new system which offers more inclusive and objective representation.

Initially when I discovered sortition I had no idea what the word meant or even that it even existed. The thought I had was to replace voting with a lottery. An internet search showed me my idea was not new. I did study Greek history many years ago so I guess the idea was simply hiding somewhere quietly in the back, rather than being completely new.

Sortition is an old system most known from the Ancient Greeks, but also from the Venetians and the Florentines. Using random chance rather than a vote to choose the members of a parliament offers a better chance at truly representative government. Sortition’s most important need is integrity in the draw, and that everyone can see this integrity for themselves. In using a random choice our first concern is skewed outcomes. One vote can be the balance of power when the house of government votes. Completely random assignment could mean that there are 105 men and 95 women members in our new senate. The chance of such skewed outcomes is quite likely when probability outcomes are examined. Such outcomes are perhaps an even bigger problem when we look at wealth inequality. Some randomly chosen senates would have more rich members than poor, and some senates more poor than rich.

The next concern is who conducts the draw. To use current government officials, or to allow a televised draw like lotto are both problematic. A single rigged result is not so bad because this person becomes simply one vote within one hundred and ninety-nine others. Still there is the need for transparency and integrity in the draw. A live draw with all the people affected by it in the audience, which is conducted by officials selected by the draw last time the sortition lottery ran is the best way. The very first round of lotteries should use sports umpires because of course there was no previous lottery. They should first attend a draw of their own for selection and random assignment to different positions. It is important that the new officials have as little as possible to do with the old electoral system.

Sortition can be modelled such that the individual person chosen is a random choice while the distribution of wealth and gender are held perfectly equal. The entire population without any need for anyone to register is split into groups. Two hundred groups are formed to each have their own separate lottery and thus avoiding skewed outcomes, because every group has one eventual winner. Each wealth group represents one step of one hundred up the wealth ladder from poor to rich. Given the wealth differences between men and women it is far more accurate to split the population by both wage and gender, rather than just wages alone. Having a live lottery also means electoral rolls are not needed, as only the winners need be checked, and of course this means that no pre-registration is required. This would help include many people currently not on the electoral roll, and gets past problems of registering address changes, those newly turning 18, new citizens, and ensuring registration weeks before an election. The first nation to have sortition lotteries like this would arguably be the first nation in which homeless people have the same chance and say as university professors or yacht club members.

With huge populations there is a need to split the population into manageable groups so it would be possible to fit them all somewhere. A live lottery is very desirable to help scrutiny. The integrity of the draw is best if the maximum number of people can directly observe and scrutinise proceedings. So there are multiple good reasons to split the population into groups for their own lotteries. The first and most obvious split is gender. Next comes wealth distribution. Current democratic voting often leaves out way too many poor people. If they were entering a lottery for a solid desirable wage for three years then surely this is good incentive to turn up. Currently most senators are paid at above the 95th percentile wage, which means 95% of the population earns less than them annually. I think a more reasonable and fair wage would be the 90% percentile wage. This across many countries equates to a little above $100,000. This is more than double the average wage in most countries. Politicians approving their own pay rises has gotten out of hand, ( don’t even start me on the Italians ) and I think it is far better to set the wage using a percentile rank. This means that pay rises only happen as everyone overall gets higher wages. Such a wage will obviously be very attractive to nearly everyone.

Gender and wealth discrimination have the most important needs to be countered by equal representation. Drawing exact lines between different races and religions is too complicated and not likely to be accurate. Wealth discrimination can be given a solid number. First the population is split in half by gender, then each half is split into one hundred sequential wealth groups. Other needs for equality are taken care of by the randomness of selection which means everyone over 18 has the same chance of selection.

Again, each wealth group represents one step of one hundred up the wealth ladder from poor to rich. First the wealth assessment is by wages, then this is also checked against overall wealth. Wage limits are set. When your wage passes the limit you pass into the higher group. Wealth limits are set for each group. Asset wealth is important, not all houses and cars should be considered the same. It does matter whether you live in The Hamptons or Detroit. The wealth limits should be adjusted over time. Once every three years, the period of government, demographers reset the limits using tax returns and census data. Only upon wining selection by sortition is your wage and assets checked. This saves a lot of effort compared to keeping individual records, and preserves privacy much better. Governments have proven over and over again that personal details are not safe. The best way to guard against such danger is to have as few details as is needed. Also less details is less cost and less confusion. Details of the wealth limits are posted well before the sortition lotteries, then each individual works out for themselves to which group they should attend to win the lottery. If they work it out wrong they lose selection, and the next runner-up steps up.

The system used for random choice of candidates must be robust. It must assume people will cheat in every way possible. The way to success is to have everyone able to watch proceedings. To hold a game of chance with several rounds so that cheating is more difficult. To use new officials for each lottery and choose them at random. We have 200 groups, therefore many lotteries will be performed each day. The officials should be a assigned a lottery game to preside over only on the day of the lottery itself. For the sake of integrity it could be useful to switch-off all mobile phone towers within range of lottery venues on the lottery days.

The lottery itself takes the form of a bingo game. A standard round bingo cage is used with ninety-nine balls inside. The bingo cage relies on gravity and so is easier to check than the more modern perspex globe and blower. Gravity is constant and the blower could be varied. Players enter the venue with no assigned seating. The lottery cards are printed beforehand and have six random numbers (one to ninety-nine) on them. Consulting leaders in anti-counter fitting technology is important to ensure that fake cards are not able to be manufactured as the numbers are read out. Each player tries to match all six numbers to get their bingo and win the round. The cards are shuffled in the view of the participants. Each pile is randomly given to a distributor. The distributors are then assigned at random to particular areas. When they reach their area they hand the cards to the participants. No lines are to be used. Rather people gently crowd and receive cards in a more random fashion this way. There is no hurry as the game only begins when everyone is ready, so the order of lines is not necessary. It is very useful if all participants write their name on their card at this time, also writing a favourite lyric or quotation would also help ensure that no card-swapping takes place. Fifty winners are chosen in the first round, who make their way to the seating on the centre stage. Then another round between these fifty is performed with new cards to pick one winner. This final round keeps running until the order of the entire fifty is reached. The second place-getter will be needed if the winner has lied about their wage level. This is much easier than holding the lottery again. Also these place-getters can now be offered their choice of a number of positions though they are not obliged to take up any as they are different to becoming a member of the senate which was the original objective. The place-getters can be ombudsmen, legal jurors, overseers of police integrity, overseers for complaints within the military, animal welfare inspectors, food health inspectors, and officials for the next round of lotteries in three years time. Any job can be added that benefits from the integrity provided by random assignment. This is just a basic summary of the lottery workings. The maths has been omitted as it is probably boring to most of you. To be technical for a moment, there actually is a need for an intermediate round between the first and second round as described here. I have written a whole book in which the biggest chapter is sortition. The full detail is too long for this article but hopefully the book will be available soon.

Most senate members will only serve one term. It is possible under this model to allow returning candidates. Firstly they can win the sortition lottery again. They participate as normal and so have the same slim chance of winning. Also a lottery winner can nominate a previous candidate of their choice instead of serving in the new parliament themselves. Only three terms are allowed for each candidate in total.

In time as people get used to this new senate it may be able to replace both houses of government with the one house of two hundred members. This leaves us with a leader to choose. As the candidates are truly representative and not driven by party politics then they can effectively represent a popular vote. They will have a closer knowledge of who the candidates are and how they work, and so can arguably make a better choice than the public. They know the person through their own experience rather than through a media lens or a political campaign, which of course often represents a distorted view. A popular vote as currently used has the bias from who bothers to turn up to vote, or who is allowed to vote. There are also the effects from people missing off the electoral roll.

A leader chosen like this should only serve one term. The current practice of long serving leaders increases the chances of corruption. With longer to learn how integrity is enforced, and longer to build a public profile which presumes no-one may ask questions, then it becomes easier to beat measures meant to ensure honesty with deception. Friendships with business leaders grow over time. This can make better more efficient leaders, but it also gives the chance of more corrupt ones.

The transition between these newly chosen leaders can be much smoother and more congenial than current politics allow because party alliances are removed. The new leader will form the two hundred and first vote and so makes a nicer number to ensure all votes have a result.

This new senate has much greater freedom to get the job of government done. Party politics is no more. More diverse views can be freely represented because there is no need to tow the party line. Members will be less stressed by party pressures and so more relaxed, attentive and intelligent. Without party meetings and back-room manoeuvres politicians will have considerably more time for the real business of representing their people. Most members will only serve one term and so there is no career to protect. We the public really can expect much more honourable behaviour is possible from our politicians.

In our new senate no party loyalty is required. Members need not be nursing leadership ambitions, and there is no need to climb a ladder through the party to achieve position. Currently many honourable people are put off serving by the way politics is conducted. The only way to change the conduct is by changing how people get there. Selecting candidates by lottery means the that bias of winners characteristics compared to the general population is only that they bothered to turn up.

We deserve better than the current outcomes of politics. We must stop this current crap. If we can change then we are better.

We are all very frustrated with our current system which is called democratic. The need for political campaigns means corruption through bribery. Campaigns mean that the voters get ignored because the important voice is that of the sponsors and donors. Yes behaving as we do now has gotten us into quite a mess. Politics is biased rather representative because of the power of money.

Poverty and unemployment really shouldn’t exist. Most nations and perhaps every nation has the wealth to prevent these problems. If you have thousands of dollars to bribe politicians – then poverty, unemployment, poor education, poor health care and the problem of expensive lawyers are all well below you. Why would you care for the problems of commoners? You wouldn’t bribe a politician to look after the poor.

12 Responses

  1. Charlie:> First the population is split in half by gender, then each half is split into one hundred sequential wealth groups. Other needs for equality are taken care of by the randomness of selection which means everyone over 18 has the same chance of selection.

    I don’t understand the need for such a carefully stratified sampling process. If your goal is to increase the representation of those further down the income distribution this will be performed automatically by the law of large numbers (as with the “other needs for equality”). Unfortunately the only equality that is guaranteed is the equal chance of being selected, sortition will certainly not promote the (numerical) equality of minorities, that would require detailed quotas.


  2. Stratification can take place along many characteristics, but I am not sure why it should be – either along gender and income or along any other characteristic.

    Also, the lottery procedure proposed seems very complicated. I think it should be possible to devise a method that is not too complex and at the same time hard to manipulate.


  3. Even a mandatory participation sampling benefits from stratification and it’s easy to illustrate why. Imagine 100% of people above 50 years of age support a proposal and 100% of people below 50 oppose it. Stratification along age (with a cut-off between groups at age 50) would always result in a perfect reflection of the public’s preferences with exactly zero chance of an outlier. Now imagine the opposite extreme. Imagine there’s no correlation whatsoever between age and policy support. Then absolutely nothing is gained or lost by the stratification. It’s exactly the same as if the stratification did not exist at all. In practice we will always be between the two extremes. As long as there is *some* correlation between policy preferences and the factors being stratified for then something is gained and outliers are made less likely. It’s effectively using non-random data (like the census) to shore up the random sampling against fluctuations as much as the existing correlations allow.

    So I would absolutely stratify as strictly as the size of the sampling allows. However, 100 wealth groups is probably unnecessary. In fact, it’s probably harmful, as it would limit our ability to more effectively stratify along more significant lines. I’d cut the number of wealth groups down to three or four, and also add age, region, religion, ethnicity, and likely other dimensions as well, assuming we can do so comfortably given the size of the sampling.


  4. I agree with Naomi that stratification can do no harm (as long as the process remains adequately transparent). Any segment of society that wishes and constitutes at least 1% of the population in a hypothetical 100-seat mini-public, should be allowed to demand stratification for their trait. This doesn’t “benefit” them, since without stratification they would be just as likely to have been over-represented as under-represented (except if service were non-mandatory and certain groups were less likely to voluntarily serve).


  5. Stratification can do harm. First, it requires classification of people along the stratified characteristic which can be controversial – whether this is done by “objective” criteria or by self-identification. Second, it creates the a-priori expectation that people in each stratum have a distinct identity and distinct interests and thus reduces the tendency toward cooperation and solidarity.


  6. Agree with Yoram on this. The problem with stratification is, that as Ollie Dowlen and Peter Stone argue, it introduces reasons for selection and these can be subjective/arbitrary/controversial. It’s not that long ago that the idea of 50/50 gender balance would have been seen as absurd. If the presence of certain minority interests has to be guaranteed (for some reason) then it should be done through quotas, not stratification. It’s worth bearing in mind that the BC constitutional convention was deemed to be descriptively representative on account of the stratification criteria employed, thereby ignoring the elephant in the room (as participation was voluntary the sample only represented 4% of the target population — a worse statistic than Britain in the age of rotten and pocket boroughs.)


  7. Keith,
    I’m surprised. I figured you’d favor more consistent policy outcomes at a given sample size. I don’t frankly care what the dimensions of stratification are. If self-proclaimed chess players were strongly correlated with policy preferences that would be a fine question to put on the census and stratify on. The dimensions need only be specified in ordinary law so they can be adjusted as needed.


  8. My preference would be to increase the sample size, rather than rely on subjective and arbitrary stratification choices (chess playing or whatever). If the sample is large enough to deliver 50/50 male/female at (say) a 5% confidence level and the initial decision threshold was set at a 60% pass rate, I think that would provide sufficient consistency between samples. I don’t have any strong views on this, just think Charlie’s proposal is too complex and deterministic (and has nothing to do with equality).


  9. As we discussed before, a supermajority threshold does not decrease randomness, it only shifts it from one side to the other (passage versus defeat). If your concern is defeating legislation that does not deserve to pass (as your previous comments would seem to indicate) your participation doesn’t likely matter. The odds of a <50% proposal passing are negligible. If your goal is passage (as it will be for other people), the randomness is more severe, things which should pass will be defeated and your exclusion when a proposal is at the edge of passage becomes more offensive and the arrangement is delegitimized in their eyes.

    This, on the other hand, is a genuine and completely free decrease in actual randomness. Maybe it's not so elegant, but then again it's not so elegant to have the population divided up along party lines for elections either. Practical issues should be our first concern.


  10. OK, as long as we don’t end up camouflaging any elephants.


  11. Charlie the author here.
    Thank you for all your comments and feedback.

    The model is not elegant. The thing is that we want representation by everyone. The poorest groups are entirely left out of most if not all democracies. Most parliaments have 70-300 or so members. Therefore I chose 200 as it is a normalish size, and can give voice to the groups that need it. Also life is very different at the top of wealth as well. Being in the top 5 is nothing like the top 1%. Small groupings are needed to get the view of everyone. Also everyone can feel their identity is represented.

    This model acknowledges human are lying cheating scum. We are not always so bad, but if we want a robust system then it will need to be reasonably complex. Expecting the best of people is not a good idea in politics.


  12. Charlie:> Small groupings are needed to get the view of everyone.

    But, as there are so few people in (say) the bottom 1%, you might end up with one randomly-chosen person from this stratum, who may well not be an effective spokesperson for the interests she is supposed to represent. Electing an advocate for the underprivileged may well be a better option than descriptive representation. Would they be better off with Bernie Sanders or Vicky from Little Britain?

    >Expecting the best of people is not a good idea in politics.

    Yes indeed, hope for the best but expect (and plan for) the worst.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: