Given the chaos of the current political system, especially over the past decade, the temptation of this slow descent into authoritarianism across the West is not hard to understand - or at least to sympathise with. The success of states such as Singapore, which I would argue enact democracy only in the most superficial sense, or the economic miracle of China over the past half-century, which has rejected the ideology entirely, all seem to point at a dirty truth: democracy, at least as we generally understand it, is not a particularly efficient form of government.
The problem though with efficiency through totalitarianism, unfortunately, is that it relies on a core assumption: that those with absolute power have motives aligned with the interest of the people - and, on even shakier ground, that this alignment will persist in perpetuity. So the question remains: is there a system that removes the inefficiencies of elective democracies while ensuring its goals remain aligned with the populace?
In the early 2000s, economist Robin Hanson proposed a new system he named “Futarchy,” along with the, (I would argue rather catchy), slogan: “Vote Values, But Bet Beliefs.” I would recommend anyone to read his 2013 paper on the subject[1], for the sake of this article I will include a brief explanation here that differs slightly from Hanson’s method but I think is easier to understand (economically they are roughly equivalent anyway I believe). The core idea is simple. Instead of people voting on policy, even indirectly through elected representatives, the population votes through some democratic mechanism on a metric of success. For the sake of example, let’s say the metric chosen is a nation’s GDP - although in reality this would be a terrible measure on its own.
This metric then determines the outcome of a betting market open to anyone. Given some proposed policy action, each participant places a monetary bet and states what they believe the GDP will be after a set period if the policy is enacted and if it is not, (two separate predictions).
Once bets are placed, the policy option with the highest predicted GDP, weighted by the amount each participant has wagered, is enacted.
Finally, once the time period has elapsed (say, a year), the actual GDP is measured and the market is settled. Participants receive a share of the total pool based on how much they bet and how accurate their prediction was (on the enacted decision); those who were closer profit, while others take a loss.
Crucially, the market is open to anyone with no minimum or maximum bet, potentially including participants who are not themselves part of the democracy.
The first time I read about this system, I was honestly stunned by its elegance - in particular, that it solves a problem for which all previous solutions I had heard were unethical at best. People often float the idea that the average voter is not sufficiently informed (I would certainly count myself well within that average), and that perhaps it would be better for those who can demonstrate more academic or professional credentials to receive a higher weighting in their democratic vote. The problem is that this idea is as slippery a slope as a vaseline-lined bin chute. Immediately we encounter questions about who defines “qualified,” and so on.
Futarchy, on the other hand, actually provides a solution to this problem. The more knowledge someone has about the likely outcome of a policy decision, the more, on average, they would be willing to bet - essentially giving higher weight to the voices of experts while also making use of insider knowledge. At the same time we ensure the “motives” of these experts are aligned with the desired outcomes of the population through financial incentive.
I think another really nice potential outcome is the creation of think tanks whose funding is also economically aligned with the “mean interests” of the country. Current think tanks, however they may be funded - be it by special interest groups, private donors, or corporations - have no real incentive to align their core ideologies with those of the average citizen. At its simplest, this is because even in the fairest societies, value is not evenly distributed. Think tanks (assuming they aren’t funded by the state) should, on average, have their ideology aligned with those possessing more than the median amount of wealth.
A futarchy-based system sort of flips this on its head. Suddenly we have created a motive for private think tanks with a direct market opportunity to work out which policies will best benefit the median voter - assuming the metrics defined by the voters have been well chosen. The size of this market opportunity is of course debatable. The state would likely have to seed the pot with some money such that the average payout at the end of the prediction market is a net positive rather than zero, just to encourage participation in the first place.
Additionally, there will be some who, due to ideological pressure, try to bet in order to shift policy towards their preference without proper analysis of the outcome. This is actually beneficial, as it gives more incentive to groups doing deeper analysis to participate - they have a good chance of winning money from the less thorough participants. Essentially the more bad actors participate the more money honest participants can potentially make.
The potential improvement over current decision making systems to me seems quite hard to over state. The efficiency of the market in creating complex system is really the great miracle of capatilism. Famously orange juice futures can often predict weather outcomes better than traditional weather forecasting services, and this is due to one simple point, there’s a lot of money to be made in predicting the future.
The nature of the metric used to define and settle the market is also extremely important, and this really leads us to the difficulty of implementation. You could define some combination of several numbers such as purchasing power and so on; however, it may be better to define the metric relative to other countries to smooth out the effects of random events such as geopolitical crises or global pandemics. This definition starts to become far more complex when you imagine more than one bet happening in parallel, as participants betting on one policy action could interfere with another.
Given this exploding complexity, we might first experiment with futarchy in simpler systems with better-defined metrics of success.
It was actually in looking for the solution to a very different problem that I first encountered the topic. Much of my day job working in decentralised finance is centred around the concept of a DAO, or Decentralised Autonomous Organisation. For the uninitiated, this is similar to a public company with shares, but instead of shares, a token is created on the blockchain with voting power. This token has absolute control over the company, both through access to the company treasury (which is only possible with a successful vote) and through control of other parts of the company that exist on the blockchain. An important thing to note here is that the control of the token is not defined through legal systems, but rather through the programmatic design of the system - it’s literally impossible for the company to access any funds without first being given access to a set amount by the DAO.
This reliance on well-designed systems rather than the law creates many attack vectors that are often mirrored in democratic systems at the national level. It was one particular attack that I was trying to find a solution to.
Imagine the token holders of some company have complete control over the company and all assets through the voting system. Someone could create a new token which any original token holder can mint at a one-to-one ratio, until the total amount of the new token matches 51% of the old. Then a vote is created to transfer ownership of the entire company to the new token. If someone were to create and start such a vote and every token holder acted in their best interests, the most sensible thing to do would be to try and get hold of some of the new token as fast as possible - and if you make it in time, to vote to pass the transfer of power. Essentially, this would remove all ownership from 49% of holders. Once this is a proven strategy, there is then the risk of it spiralling, as someone realises they can do the same, and so on, until eventually the entire company is owned by a single entity.
We could very easily say that when the company and the voting system is set up, we create a constitution that can never be changed - one that bans such votes that remove voting power from any original holder. Unfortunately, and I won’t go into the technical details of why, adding a system enacted after the vote to enforce this constitution in code is extremely difficult. For a while I thought it might even be impossible.
This is where futarchy may potentially provide a solution. The core problem is that it’s very hard to create a program that can detect if the outcome of a vote would break the constitution. Instead, we could define a veto system that any vote must pass through before being enacted. This veto system could provide a betting market similar to the one I described earlier, where anyone can bet on whether they think the value of the token will be higher or lower a week after the vote. The outcome with the higher predicted price is then enforced.
In simple terms, it’s easy to see that if a group tried to transfer all power to a new token owned by 51% of the original holders, the old token’s price would drop to zero. Therefore, if a vote was raised on this topic, it’s an extremely risk-free bet for anyone in the world to bet on the price going down in the case of the vote passing. Of course, they still need to predict the outcome if it doesn’t pass, which is the prediction actually taken - creating some betting risk. However, this price change should be comparatively small, and over many votes any entity participating should on average make slightly more than they lose.
Although rather convoluted, I think this example nicely shows how futarchy can solve some problems that are so far ignored in our traditional democratic systems. The DAO attack I’ve described is a textbook case of “tyranny of the majority” - and this isn’t just a problem in decentralised finance. It occurs in national democracies all the time.
Consider the UK’s Brexit referendum, in which a small majority was able to enforce a very large constitutional change on the entire population - in this case by only a couple of percent. The complex preferences of society were flattened into a winner-takes-all vote. A futarchy-style mechanism wouldn’t have prevented the vote itself, but it might have introduced a check: would the predicted economic outcome of leaving have survived a betting market?
Perhaps experimentation in the use of markets to generate policy at smaller scales could eventually provide some kind of pathway to its use on a national stage. Until then, it seems that elective democracies remain the best of a shoddy bunch.
The following is an explainer for those interested on the method described by Hanson in his paper on Futarchy
Given some measurable metric (say, GDP per capita, normalised to a scale of 0 to 1) and some proposed policy - let’s say raising the minimum wage to £15/hour - two separate markets are opened.
First, a “welfare asset” is created; an asset that will pay out in proportion to the measured GDP one year from now. If GDP per capita ends up at 0.7 on our normalised scale, the asset pays £0.70. These assets are created in pairs: one that pays £W, and one that pays £(1−W). Together they always sum to exactly £1, meaning whoever creates them takes no risk - they’re just putting £1 in and getting £1 back, split across two assets. Each of the two pairs for a given asset can be freely bought and sold on the open market.
If the current trading price of £W is 0.7 (so £0.3 for £1-W) and a trader believes that the final value of £W will be 0.75 they would buy £W, if they believe it will be 0.65 they should buy £1-W.
Now, with the minimum wage proposal, two conditional markets are created. In the first, people trade the welfare asset under the condition that the wage increase is enacted. In the second, people trade under the condition that it is not. Importantly, all trades in a market are cancelled - “called off” - if that market’s condition isn’t met. So for instance, if you have bought £W or £(1−W) in the market based on the condition that minimum wage is increased, but the final decision is to not increase, your trade buying that asset is cancelled and you receive a refund for whatever you paid.
Anyone can trade in either or both markets. Through the process of buying and selling, a market price emerges in each. These prices represent the collective estimate of what GDP will be in each scenario. Say the “enacted” market settles at £0.72 and the “not enacted” market at £0.68 - speculators are saying they expect higher GDP if the minimum wage is raised.
The decision to increase minimum wage or not is then enacted based on which market is trading its £W at a higher price (as this is the predicted GDP).
All trades in the “not enacted” market are now cancelled. A year later, GDP is measured and comes in at 0.71. The welfare assets pay out £0.71 each (or £0.29 for £(1−W)). If you bought at £0.72, you lose £0.01 per unit. If you bought at £0.65, you gain £0.06.
Original Version https://maxwickham.substack.com/p/futarchy-and-tyranny-of-the-minority
Given the chaos of the current political system, especially over the past decade, the temptation of this slow descent into authoritarianism across the West is not hard to understand - or at least to sympathise with. The success of states such as Singapore, which I would argue enact democracy only in the most superficial sense, or the economic miracle of China over the past half-century, which has rejected the ideology entirely, all seem to point at a dirty truth: democracy, at least as we generally understand it, is not a particularly efficient form of government.
The problem though with efficiency through totalitarianism, unfortunately, is that it relies on a core assumption: that those with absolute power have motives aligned with the interest of the people - and, on even shakier ground, that this alignment will persist in perpetuity. So the question remains: is there a system that removes the inefficiencies of elective democracies while ensuring its goals remain aligned with the populace?
In the early 2000s, economist Robin Hanson proposed a new system he named “Futarchy,” along with the, (I would argue rather catchy), slogan: “Vote Values, But Bet Beliefs.” I would recommend anyone to read his 2013 paper on the subject[1], for the sake of this article I will include a brief explanation here that differs slightly from Hanson’s method but I think is easier to understand (economically they are roughly equivalent anyway I believe). The core idea is simple. Instead of people voting on policy, even indirectly through elected representatives, the population votes through some democratic mechanism on a metric of success. For the sake of example, let’s say the metric chosen is a nation’s GDP - although in reality this would be a terrible measure on its own.
This metric then determines the outcome of a betting market open to anyone. Given some proposed policy action, each participant places a monetary bet and states what they believe the GDP will be after a set period if the policy is enacted and if it is not, (two separate predictions).
Once bets are placed, the policy option with the highest predicted GDP, weighted by the amount each participant has wagered, is enacted.
Finally, once the time period has elapsed (say, a year), the actual GDP is measured and the market is settled. Participants receive a share of the total pool based on how much they bet and how accurate their prediction was (on the enacted decision); those who were closer profit, while others take a loss.
Crucially, the market is open to anyone with no minimum or maximum bet, potentially including participants who are not themselves part of the democracy.
The first time I read about this system, I was honestly stunned by its elegance - in particular, that it solves a problem for which all previous solutions I had heard were unethical at best. People often float the idea that the average voter is not sufficiently informed (I would certainly count myself well within that average), and that perhaps it would be better for those who can demonstrate more academic or professional credentials to receive a higher weighting in their democratic vote. The problem is that this idea is as slippery a slope as a vaseline-lined bin chute. Immediately we encounter questions about who defines “qualified,” and so on.
Futarchy, on the other hand, actually provides a solution to this problem. The more knowledge someone has about the likely outcome of a policy decision, the more, on average, they would be willing to bet - essentially giving higher weight to the voices of experts while also making use of insider knowledge. At the same time we ensure the “motives” of these experts are aligned with the desired outcomes of the population through financial incentive.
I think another really nice potential outcome is the creation of think tanks whose funding is also economically aligned with the “mean interests” of the country. Current think tanks, however they may be funded - be it by special interest groups, private donors, or corporations - have no real incentive to align their core ideologies with those of the average citizen. At its simplest, this is because even in the fairest societies, value is not evenly distributed. Think tanks (assuming they aren’t funded by the state) should, on average, have their ideology aligned with those possessing more than the median amount of wealth.
A futarchy-based system sort of flips this on its head. Suddenly we have created a motive for private think tanks with a direct market opportunity to work out which policies will best benefit the median voter - assuming the metrics defined by the voters have been well chosen. The size of this market opportunity is of course debatable. The state would likely have to seed the pot with some money such that the average payout at the end of the prediction market is a net positive rather than zero, just to encourage participation in the first place.
Additionally, there will be some who, due to ideological pressure, try to bet in order to shift policy towards their preference without proper analysis of the outcome. This is actually beneficial, as it gives more incentive to groups doing deeper analysis to participate - they have a good chance of winning money from the less thorough participants. Essentially the more bad actors participate the more money honest participants can potentially make.
The potential improvement over current decision making systems to me seems quite hard to over state. The efficiency of the market in creating complex system is really the great miracle of capatilism. Famously orange juice futures can often predict weather outcomes better than traditional weather forecasting services, and this is due to one simple point, there’s a lot of money to be made in predicting the future.
The nature of the metric used to define and settle the market is also extremely important, and this really leads us to the difficulty of implementation. You could define some combination of several numbers such as purchasing power and so on; however, it may be better to define the metric relative to other countries to smooth out the effects of random events such as geopolitical crises or global pandemics. This definition starts to become far more complex when you imagine more than one bet happening in parallel, as participants betting on one policy action could interfere with another.
Given this exploding complexity, we might first experiment with futarchy in simpler systems with better-defined metrics of success.
It was actually in looking for the solution to a very different problem that I first encountered the topic. Much of my day job working in decentralised finance is centred around the concept of a DAO, or Decentralised Autonomous Organisation. For the uninitiated, this is similar to a public company with shares, but instead of shares, a token is created on the blockchain with voting power. This token has absolute control over the company, both through access to the company treasury (which is only possible with a successful vote) and through control of other parts of the company that exist on the blockchain. An important thing to note here is that the control of the token is not defined through legal systems, but rather through the programmatic design of the system - it’s literally impossible for the company to access any funds without first being given access to a set amount by the DAO.
This reliance on well-designed systems rather than the law creates many attack vectors that are often mirrored in democratic systems at the national level. It was one particular attack that I was trying to find a solution to.
Imagine the token holders of some company have complete control over the company and all assets through the voting system. Someone could create a new token which any original token holder can mint at a one-to-one ratio, until the total amount of the new token matches 51% of the old. Then a vote is created to transfer ownership of the entire company to the new token. If someone were to create and start such a vote and every token holder acted in their best interests, the most sensible thing to do would be to try and get hold of some of the new token as fast as possible - and if you make it in time, to vote to pass the transfer of power. Essentially, this would remove all ownership from 49% of holders. Once this is a proven strategy, there is then the risk of it spiralling, as someone realises they can do the same, and so on, until eventually the entire company is owned by a single entity.
We could very easily say that when the company and the voting system is set up, we create a constitution that can never be changed - one that bans such votes that remove voting power from any original holder. Unfortunately, and I won’t go into the technical details of why, adding a system enacted after the vote to enforce this constitution in code is extremely difficult. For a while I thought it might even be impossible.
This is where futarchy may potentially provide a solution. The core problem is that it’s very hard to create a program that can detect if the outcome of a vote would break the constitution. Instead, we could define a veto system that any vote must pass through before being enacted. This veto system could provide a betting market similar to the one I described earlier, where anyone can bet on whether they think the value of the token will be higher or lower a week after the vote. The outcome with the higher predicted price is then enforced.
In simple terms, it’s easy to see that if a group tried to transfer all power to a new token owned by 51% of the original holders, the old token’s price would drop to zero. Therefore, if a vote was raised on this topic, it’s an extremely risk-free bet for anyone in the world to bet on the price going down in the case of the vote passing. Of course, they still need to predict the outcome if it doesn’t pass, which is the prediction actually taken - creating some betting risk. However, this price change should be comparatively small, and over many votes any entity participating should on average make slightly more than they lose.
Although rather convoluted, I think this example nicely shows how futarchy can solve some problems that are so far ignored in our traditional democratic systems. The DAO attack I’ve described is a textbook case of “tyranny of the majority” - and this isn’t just a problem in decentralised finance. It occurs in national democracies all the time.
Consider the UK’s Brexit referendum, in which a small majority was able to enforce a very large constitutional change on the entire population - in this case by only a couple of percent. The complex preferences of society were flattened into a winner-takes-all vote. A futarchy-style mechanism wouldn’t have prevented the vote itself, but it might have introduced a check: would the predicted economic outcome of leaving have survived a betting market?
Perhaps experimentation in the use of markets to generate policy at smaller scales could eventually provide some kind of pathway to its use on a national stage. Until then, it seems that elective democracies remain the best of a shoddy bunch.
[1] https://mason.gmu.edu/~rhanson/futarchy2013.pdf
The following is an explainer for those interested on the method described by Hanson in his paper on Futarchy
Given some measurable metric (say, GDP per capita, normalised to a scale of 0 to 1) and some proposed policy - let’s say raising the minimum wage to £15/hour - two separate markets are opened.
First, a “welfare asset” is created; an asset that will pay out in proportion to the measured GDP one year from now. If GDP per capita ends up at 0.7 on our normalised scale, the asset pays £0.70. These assets are created in pairs: one that pays £W, and one that pays £(1−W). Together they always sum to exactly £1, meaning whoever creates them takes no risk - they’re just putting £1 in and getting £1 back, split across two assets. Each of the two pairs for a given asset can be freely bought and sold on the open market.
If the current trading price of £W is 0.7 (so £0.3 for £1-W) and a trader believes that the final value of £W will be 0.75 they would buy £W, if they believe it will be 0.65 they should buy £1-W.
Now, with the minimum wage proposal, two conditional markets are created. In the first, people trade the welfare asset under the condition that the wage increase is enacted. In the second, people trade under the condition that it is not. Importantly, all trades in a market are cancelled - “called off” - if that market’s condition isn’t met. So for instance, if you have bought £W or £(1−W) in the market based on the condition that minimum wage is increased, but the final decision is to not increase, your trade buying that asset is cancelled and you receive a refund for whatever you paid.
Anyone can trade in either or both markets. Through the process of buying and selling, a market price emerges in each. These prices represent the collective estimate of what GDP will be in each scenario. Say the “enacted” market settles at £0.72 and the “not enacted” market at £0.68 - speculators are saying they expect higher GDP if the minimum wage is raised.
The decision to increase minimum wage or not is then enacted based on which market is trading its £W at a higher price (as this is the predicted GDP).
All trades in the “not enacted” market are now cancelled. A year later, GDP is measured and comes in at 0.71. The welfare assets pay out £0.71 each (or £0.29 for £(1−W)). If you bought at £0.72, you lose £0.01 per unit. If you bought at £0.65, you gain £0.06.