Governance the Petri Dish
Blockchains have governance. Even instances of ‘ungovernance’ such as that displayed in Bitcoin or Monero are technically forms of governance. However, what having a multiple-blockchain ecosystem provides is a means to test implementations of other governance systems without as much at stake. What also often angers many individuals is the idea of “governance theater,” where governance is used to promote an asset by being a ‘cutting edge’ feature. Meanwhile, most of these new systems will likely end up proving to be inferior due to being overcomplicated. However, it’s important to simultaneously reject the act of “governance theater” and instead treat governance experiments how they should be: as petri dishes.
Petri dishes are traditionally used to culture cells under controlled conditions. If we treat blockchains as petri dishes, we can test how these governance systems play out with lower stakes and observe outcomes. On more nascent chains, tweaking these systems and inserting variables becomes an easier task (due to the low financial risk and malleability) while getting the right parameters to their respective Goldilocks zone. If the system pans out well enough, a larger chain might see its success and choose to adopt such a system. This isn’t to say that a more seasoned chain would just simply adopt something (as there are plenty of other ideological variables to attend to), but it might make it more palatable in a proposal if there’s a functional implementation floating around.
Futarchy as a governance component for a blockchain is a great example of a petri dish in action. It’s relatively new, untested on a massive scale, and could be implemented in a number of ways. Decentralized predictions markets and oracle solutions are on the rise, which aid in the betting markets component to pass particular resolutions based on a voted-in metric. To first understand the implementation of futarchy in blockchain-based settings, one must first understand how futarchy actually works.
What is Futarchy?
Vote values, but bet beliefs.
Robin Hanson, an associate professor of economics at George Mason University, originally posted his “manifesto” on futarchy back in 2000 on his George Mason site, and a longer paper in July of 2000. The system’s intent was to address the shortcomings of traditional democracy by using betting markets in order to determine policy implementation. “Values” as measurable metrics (“x to be y by z”: where x is a measure — either monetary or otherwise, y is the desired result amount, and z is a time threshold) would be voted on in traditional ways, and “beliefs” or policy that’s enacted in order to reach that goal would be left to speculators. If that betting market favored implementation, the policy would be implemented and assessed at the z threshold specified in the value. Updated forms of Hanson’s work were published in 2003, 2007, and 2013.
Enacted policies determined through predictions markets (beliefs) are used to achieve the desired metrics (values) that are voted on.
Futarchy begins with a voting phase. Although the system is heavily dependent on predictions markets determining the path toward a particular metric, the metric itself must be determined. Rather than simply vote on a means to an end as we traditionally do, we’re voting on what the ‘end’ is.
In a simple case, this could be something such as “unemployment rate to be 2.5% by 2020 in the United States,” (see: x to be y, by z) where a threshold, a duration, and a sector are all chosen. After the metric has been selected through voting, predictions markets take over to determine what policies will be chosen to hopefully achieve that metric (value).
Hanson’s idea that predictions markets can be a superior way of determining policy choices to reach voted metrics comes from his critique of democracy. In his paper, Shall We Vote on Values, But Bet on Beliefsfrom the Journal of Political Philosophy in 2013, Hanson attributes the failure of our current democracies and voting systems to the weak incentives for being a well-informed voter. This in turn shapes the case for speculative markets as a means to drive “info successes,” as there is a direct financial asset on the line and everyone is incentivized to drive as close to an accepted or acceptable price.
The importance of futarchy in decision making comes from the abstraction of bias that predictions markets provide. We may vote for what makes us feel good because of subjective bias, but the outcome might not create an overall long-term positive outcome. Leveraging a predictions market in favor of traditional voting leaves participants typically caring more about a bottom line rather than their feelings. According to Hanson, “those who think they know more tend to trade more, and specialists are paid to eliminate any biases they can find” which ends up creating a desire for accuracies for anyone daring to enter the market itself if it were aimed at policymaking.
Once the metric has been determined by a traditional vote, predictions markets then shape policy implementation. In the above example, the metric for success was a 2.5% unemployment rate by 2020. The next step in a futarchy is for a proposal to be published and markets to be opened on choices. For the sake of the example, we can use proposal “x” as someone’s means of reaching the metric where markets y¹ and y² represent the pricing of the options determining the fate of the proposal.
If y¹ represents the price of “yes,” and exceeds that of y² representing the price of “no” by the time the market closes, the proposal is implemented, and all trades on the “no” market are reverted. After the success of the proposal is tested over time, and if it truly makes an impact toward the success of the initial value (2.5% unemployment), then those that were on the winning side receive rewards based on their holdings on the winning side of the proposal. The actual rewards and losses all depend on the implementation.
Where Does Futarchy Go Wrong?
Market manipulation, value subjectivity, low participation, measurement of implemented policies (human arbitration), and volatility. All of which have been summed up in an excellent post by the Ethereum blog (reiterated here), with reasoning collected by Mencius Moldbug and Paul Hewitt.
Volatility: The “informed” participant narrative gets crushed under buyers simply following a market if there’s a distinct trend due to volatility. One individual manipulating the market can simply swing one way and have the less-informed traders simply follow the market trend rather than understand what type of proposal they’re actually voting on.
Low Participation: Much like traditional voting, turnout may prove to be quite the issue for predictions markets in a futarchy system. Traders might simply be the only ones playing the market while your standard voter might continue to veer away from the still-implicit communication costs.
Market Manipulation: Market manipulation in a futarchy can come in the form of someone with enough capital continuing to “buy yes tokens on the market and short-selling no tokens” and vice versa to dominate the entire market. Considering that turnout may already prove to be an issue with predictions markets in a futarchy, the books may in fact already be weak through the lens of a larger entity.
Value Subjectivity: “Values” aren’t always easy to measure considering that they can still be subject to emotional biases, which hurts the root of a futarchy. Individuals may end up with values that they don’t even align with if the voting phase is already tainted or is subject to any form of manipulation.
Measuring Impact: Measuring whether or not a policy makes an impact is subject to human error and may prove difficult. Other factors may have affected the originally established value, and simply reducing the effect of the proposal enacted to a binary isn’t easy, as it isn’t necessarily automated.
Blockchain Usage and Implementations
The importance of distributed futarchy implementations comes from the voting and betting system being open and left to a mass of participants rather than a dictator. Eliminating this single point of failure also adds a safety component to the futarchy system by allowing participants to understand the pre-programmed outcomes of their bets (depending on the market sway) as well as what forms of oracles are being potentially formed in conjunction with the markets. There could even be easier ways of determining what the “collective welfare” is (our values) through ideas such as Merkle’s “Democratic Collective Welfare” where metrics are determined by all citizens (users) rather than delegations.
As mentioned earlier, nascent chains and smaller projects make excellent petri dishes to observe futarchy in action. The following section is of a number of projects seeking to or those that have already implemented forms of futarchy:
The futarchy-based governance is in control of four distinct categories:
- Blockchain variables such as the block reward and block periods
- VM variables such as how many functions and variables a contract can define
- Oracle variables such as the cost of launching one, and its associated time constraints and questions
- Transaction fees including all 15 types.
The built-in oracles of Amoveo have results recorded based on predictions markets that are opened based on the question outcomes of either “True,” “False,” or “Bad Question.” Markets are resolved based on how long one side of the market simply has open orders. The reasoning for switching oracles in Amoveo to a market-based approach rather than simply voting on whether or not something has occurred was to ‘incentivize players to reveal true facts about the world to the chain’.
Betting markets are imperfect, as someone with enough money could simply “buy” their own version of the truth. Amoveo defaults to forking if an adversary decides to replace the truth with a lie, as the entire blockchain loses value from a false oracle. Miners that prefer the honest chain will simply move to it, while the oracles continue to provide truthful information in the background. Since miners are incentivized to stick to the “honest” chain, users are incentivized to bet against lies. The markets prevent spam by requiring attackers to make large losing bets in their markets which will cover the cost of individuals having to manually answer the question truthfully.
For more information on the intricacies of Amoveo’s predictions markets, Tallak Tveide has put together an excellent post.
Amoveo is unique in that it is the first blockchain to have its governance either conducted or enabled by a form of futarchy. Amoveo finished its first futarchy market back in August of 2018 to determine whether or not the blockchain should hard fork and alter its difficulty calculation algorithm. Hess’ previous work on Truthcoin, Augur, and Aeternity all display his interest in predictions markets as a means to both conduct betting and governance, and it all came to an apex with Amoveo.
In August, Tezos founder Arthur Breitman released an article called Toward Futarchy in Tezos which described the ways in which futarchy can be used as a mechanism by which protocol decisions can be made. For those unfamiliar of the connection of Tezos to futarchy, the original position paper contained a subsection regarding the possibility of its implementation:
To Arthur, Tezos currently takes the “belt and suspenders governance” approach of simply voting on and implementing proposals through voting and subsequent confirmation voting. The actual mechanism as it stands consists of four distinct periods:
- A Proposal Period in which bakers can submit proposals and vote on up to 20 proposals
- An Exploration Vote Period in which the top-ranked proposal is voted on with the threshold of passing being 80%
- A Testing Period where a testnet fork runs parallel to the main chain for 48 hours, followed by an off-chain run for the remaining cycles of the period
- A Promotion Vote Period where the proposal is activated if both a minimum quorum is met along with 80% of non-abstaining bakers voting in favor of it
Some changes have been on the table, such as lengthening the period, introducing fees, increasing the length of the testing period, and changing quorum minimums. However, what Arthur does is take this original approach, and adds futarchy as an appetizer.
In his vision, the exploration vote period (2) and the confirmation (4) would remain the same, but the approval vote (1) would be modified to fit a futarchy-based framework. To Arthur, predictions markets provide a check against voting periods, as a proposal that might slip through said markets may not make it through a vote and vice versa. Since Tezos’ governance process is one of layers, testing, and an additional layer, inserting futarchy as a tool could yield positive results.
The values associated with the Tezos futarchy in this pondered system would fall to an ex post vote, or a satisfaction survey in which policies are scrutinized based on how the future would be with them enacted. This is because “Participants may not be good at deciding ex ante what is a good policy, but deciding out ex post if a decision was a good call is much easier” or an individual can’t gauge a policy’s effect on all stakeholders, as they are typically concerned with their own well-being. To test this, Arthur starts with a decentralized oracle style market in which users place hidden bids on values. The rewards are then distributed in that market based on their proximity to the median, as it typically represents a natural Schelling point if participants are honest. The arbitration process then becomes an assumption of a “do-over” mechanism in the case of a wealthy dishonest actor. We now have our set “value.”
In the proposal futarchy, two markets are proposed: the probability of P (the selected proposal) being adopted, and the effect of P on x — our ‘values’. Included in the futarchy system is an automated market maker, and thoughts on potentially sticking to a single auction in which orders are submitted to a smart contract to have a price for “predictive purposes.” However, Arthur does point out that single auctions are weaker than continuous auctions in terms of the ability to react to mispricings. Errors in single auctions can completely throw off figures, and also don’t allow voters to gauge sentiment throughout the duration of the market period.
The importance of futarchy as a tool, in this case, is as a check on simple voting schemes. Outlandish proposals that may get passed in votes may not be able to receive the same ease in a futarchy-based predictions market and vice versa.
Ethereum — Gnosis
The Ethereum foundation has always taken an interest to futarchy, and Gnosis, in particular, has been at the forefront of tooling for the unique system. Gnosis is at its core a predictions market platform but has developed plenty of tools alongside their original vision including a widely used multisig, a non-custodial dutch auction exchange, and even a robust wallet. Back in March of 2016, Gnosis received a grant from the Ethereum Foundation to research blockchain-based futarchy systems, and as a part of their endeavors, a section titled “Futarchy — Experiment to safeguard against malicious attacks” was added to their whitepaper.
The simulation experiments assume the value of having a “better CEO for a company,” where the markets for CEO A and B are represented by the Ethereum difficulty and a modified difficulty. The tests were aimed at finding ways in which the markets could be manipulated. In August 2017 Gnosis announced their intention to carry out five related experiments, along with a detailed blog post, to test both feasibility and cryptoeconomic constraints. Their goal is to essentially add depth to the argument that even when “noisy traders” are in predictions markets, they are still among the most manipulation-resistant means of decision making.
For a more detailed version of this experiment in slides, check out their presentation here.
Gnosis has even built smart contracts to implement futarchy systems on top of Ethereum, and have even created a mockup for an application that would leverage it.
Back in 2016, Gnosis even put forth a proposal for implementing futarchy in The DAO. The idea was around interfacing The DAO and other DAOs with Gnosis, allowing individuals to speculate on probability estimates on proposals being implemented, The DAO’s token value based on opening predictions markets, and on token holders selling their DAO tokens upon meeting certain conditions (approval of a proposal).
Opening a Market on Feasibility
With the launch of the first market on Amoveo, the thought of implementation in Tezos, and all the research being done on futarchy by Gnosis, it seems as though the governance method has a future (pun intended). Regarding the values portion of a futarchy system, it might be tricky to model it for more subjective DAOs, but there can in fact be natural points with blockchains. In this case, this would be futarchy as a means to govern a blockchain rather than a smaller scale organization. For example, hashrate in PoW becomes a natural Schelling point, and even the size of a validator set in a PoS system could both serve as “values.” Even the price of the associated asset can be a particular “value” for those speculating on a currency associated with a chain.
Personally, I see value in Arthur Breitman’s thoughts on seeing futarchy as a check against voting in a dual approach system. Rather than convert an entire system to a futarchy-based approach, it may sit in an established stack as just one means to a particular end rather than the only one. However, it’s important to remember that futarchy is simply one of the many governance-based experiments that have been made accessible due to the sandbox, or petri dish-like qualities that blockchains and their respective communities have to offer. Will futarchy be the best way to govern blockchains in the future? Let’s open a market on it.
Nothing in this article should be taken as legal or investment advice.