This article explains how rational democracy and futarchy, summarized as “voting on values, betting on beliefs”, is a promising solution to the central problem in politics and ethics: moral and empirical uncertainty.
Rational ethics can be summarized with the slogan: “accurate in beliefs, effective in means, coherent in ends”. We start with the ends: our moral values. These values can be expressed in ethical rules or principles, such as the principle to maximize well-being. Our ends are coherent if they do not contain unwanted arbitrariness. Unwanted arbitrariness means making a choice whereby the consequences are unwanted by at least one individual (i.e. they cannot be consistently preferred by everyone) and the justification of that choice is not based on a rule. The latter condition means the choice is arbitrary. Examples of unwanted arbitrariness are discrimination between individuals, inconsistencies between ethical principles and ambiguities in moral values.
After determining the ends, we need effective means to reach those ends, and in order to find those means, we need accurate beliefs about the world. Now we face the central problem in ethics and politics: how to deal with moral and empirical uncertainty? Moral uncertainty is uncertainty about moral values: which ends or values are the correct ones? Empirical uncertainty is uncertainty about empirical facts: which beliefs about the world are the correct ones? The solution to this problem of moral and empirical uncertainty in politics is rational democracy.
In our current democratic system, political parties are characterized by political ideologies which contain a mixture of moral values and beliefs about empirical facts. This makes the choice or vote for most preferred moral values, most effective means and most reliable beliefs almost impossible, and it increases the risk of politicians being irrational and biased due to their identification with ideologies that distort their judgments about policies. So we first have to disentangle the moral values from the empirical facts.
Voting on values (the parliamentary model for moral uncertainty)
There are many possible coherent ethical systems, such as a deontological rights ethic, a consequentialist utilitarian welfare ethic, a libertarian ethic or pluralist ethics that combine several ethical principles (see this example). We have a moral uncertainty about which ethical system is correct. In a sense, all coherent systems are equally valid: I cannot give reasons why my coherent ethical system would be better than yours.
Nick Bostrom proposed a Parliamentary Model to deal with this kind of moral uncertainty. Here I take this idea literally: political parties should be primarily defined by their ethical systems. A political party corresponds with a cluster of similar ethical principles. For example, a consequentialist utilitarian party clusters consequentialist utilitarian ethical systems. A party member or eligible candidate can have his or her own preferred ethical system that is broadly in line with the position of the party.
An ethical committee, consisting of experts in moral philosophy, controls the coherence of the ethical systems held by all parties and party members. People who have incoherent ethical systems (i.e. with inconsistencies or unwanted arbitrariness) are not allowed to participate during election. For example: a party with a racist ideology is not allowed, because racism is a kind of unwanted arbitrariness.
Each voting citizen has 10 demivotes to vote on parties or eligible candidates. For example: if you have 90% confidence in a prioritarian welfare ethic and 10% confidence in a libertarian rights ethic, you can give 9 demivotes for the candidate who is closest to your prioritarian ethic and 1 demivote for the candidate of the libertarian party.
All the elected officials debate in the parliament about the most preferred mixture of moral values. The resulting consensus view is a kind of weighted average of all ethical systems, weighted according to preference or credence. The most important task of the members of parliament is the determination of measurable indicators for the following legislature. Analogous to life cycle impact assessments, we can make a distinction between midpoint and endpoint indicators. Possible example of midpoint indicators are: GDP, the Gini index for income inequality, lifespan, life satisfaction, depression rates, crime rates, the level of greenhouse gas emissions, measures of progress in scientific research,… These midpoint indicators can be aggregated into a limited number of endpoint indicators such as the human development index. These endpoints measure for example economic prosperity, environmental sustainability or general happiness. During the election, each eligible candidate can present his or her preferred midpoint and endpoint indicators, so all voters can have a clear picture of what the candidates find important.
The ethical committee checks if the resulting parliamentary consensus and chosen indicators do not contain unwanted arbitrariness. An independent bureau of statistics has the task to collect all the data to calculate the chosen indicators.
Betting on beliefs (the prediction market model for empirical uncertainty)
After determining the moral values or ends measured by the chosen midpoint and endpoint indicators, we now have to find the most effective means to reach those ends. These means are the policies and laws. To find those means, we need accurate beliefs. However, most people are biased and a lot of politicians have hidden agendas or personal (e.g. financial) interests. These biases generate inaccurate beliefs, resulting in ineffective means.
The question becomes: what is the best institution to find the most effective means? No institution is perfect: we do not have a completely unbiased, impartial institution with perfect scientific knowledge about economics and other relevant disciplines, that can determine the most effective policies. But this does not mean we cannot look for the least bad institution.
One interesting proposal that deserves more research, is Robin Hanson’s futarchy, which he describes with the slogan: “voting on values, betting on beliefs.” Voting on values was a least bad solution to moral uncertainty, and perhaps betting on beliefs is the least bad solution to empirical uncertainty. In a futarchy, prediction markets (speculative markets trading in idea futures) are used to determine the most effective policies. A prediction market is probably one of the most reliable and effective institutions to gain crucial information about e.g. the likelihood that a certain policy has a positive effect (measured as increases in the chosen indicators).
In a prediction market, people can trade in conditional bets. For example, if the chosen indicator is GDP and the proposed policy is a certain trade agreement, I can sell you a conditional bet that pays you $1 if the policy is adopted and GDP increases after a certain amount of time and $0 if the policy is adopted and GDP decreases. The bet is annulled if the policy is not adopted. If you have 70% confidence that GDP will increase if the policy is adopted, you (as a rational agent) are willing to pay at most $0.7. The maximum price you are willing to pay corresponds with your subjective degree of confidence (your subjective probability). Similarly, the minimum price I am willing to sell this bet corresponds with my degree of confidence. According to some research mentioned by Hanson, if there are many speculators, chances increase that the market prices of those conditional bets become reliable estimates of the probabilities of the effectiveness of the policies.
Prediction markets cannot only give probability estimates for the effectiveness of policies, but also probability estimates for future indicators chosen by future governments. This is important, because taking our moral uncertainty into consideration means taking into account that our future moral values might be different. So we have to be aware that in the future people might have other preferences for their moral values, or that new insights and technologies allow for the adoption of other, better indicators in the future.
In futarchy, a policy is adopted (and competing policies are rejected) if two conditions are met: 1) the price of the conditional bet for that policy and for the currently chosen indicator (i.e. conditional on that policy being adopted and the indicator being chosen) is clearly higher than the prices of the conditional bets for the competing policies (i.e. conditional on the other policies being adopted), and 2) the price of the conditional bet for that policy and for the most likely future indicator (i.e. conditional on that policy being adopted and the future indicator chosen) is not clearly lower than the prices of the conditional bets for the competing policies with that future indicator. The second condition guarantees that we will not regret our policy decision when our moral values (and the corresponding indicators) change in the future. For example, if chances are high that in the future another indicator than GDP will be chosen, and if the prediction market clearly predicts that the proposed trade agreement worsens that future indicator, the proposed agreement does not become law.
For a lot of policy choices, reliable information about empirical facts is highly important, and prediction markets are a good source of information (they can effectively aggregate information). Such information has a lot of value, because a lot is at stake. Therefore, prediction markets that predict the effectiveness of promising policies should be subsidized in order to attract enough speculators. This subsidy reflects the financial value of the information. The details of futarchy, as well as the rebuttals of some criticism, are discussed in Hanson’s paper.
Futarchy is just one promising proposal. There are many other possible solutions to the problem of moral and empirical uncertainty in politics. We cannot tell in advance whether futarchy or another proposal works well, but these promising proposals deserve more research and experimentation.