Introduction

Crypto and the dweb hold the promise of dragging governance and decision making processes out of boardrooms and out from behind closed doors, thrusting them into the open for all to see, comment on and participate in. The crypto-deam is sometimes portrayed as an egalitarian future in which everyone has a voice and is able to express their concerns and needs to the fullest extent possible. In this hypothetical future, society has been transformed into a kind of digital labor union, with mobs of the common man replacing the committees, boards, bureaus and c-suites that currently run our everyday lives. However in the trajectory of history up until the invention of the internet, there are few instances of true direct democracy being granted to citizens at scales beyond a few dozen people. Handing over control of everything (not just government, but also infrastructure and public goods, private companies, land, etc) to an interconnected web of individual voters could be disastrous without proper precautions being taken beforehand.

Governance is, however, required for certain phases of a protocol’s lifecycle. Without the ability to pivot quickly to respond to unforeseen events in its early days, a protocol sits at an increased risk of failure. Even Ethereum at one point needed to coordinate a hard fork of itself in order to secure its future. I do not believe that governance should never exist, rather that governance should only exist within a predefined timespan, after which the protocol is largely ossified; becoming just a another money lego to be shaped alongside others to achieve the desired outcome. In this paper I will review where governance is today, cover some examples of protocols in which governance strategy is clearly progressing, and what an endstate “ungoverned” system could look like.

Governance Today

Governance today has largely converged on a handful of similar systems. Typically projects are launched under the authoritarian control of the founders / dev team, alongside input from early community members and investors. As the project prepares to go to market, a token is typically distributed to preselected recipients in order to kick start both the accrual of liquidity to the platform and the beginning of community-led governance. CoinGecko currently lists roughly 200 hundred tokens under the ‘governance’ category, although this number is likely to only capture <10% of the true number of governance tokens out there. Prominent examples of governance tokens include $CRV $UNI $OP and $BAL, all of which were airdropped to users at the beginning of the transition to community stewardship. What is noticeably lacking from a good number of tokenized-governance-enabled protocols is a plan for the future that retires this functionality. Additionally the combination of purposes for governance tokens are widespread. The same token may be used for liquidity mining incentives, fee sharing, voting and other purposes, creating further potential for misalignment between token holders and the governance discourse.

Positives of DeFi Governance

In the bootstrap phase of life, which essentially everything in crypto is experiencing right now, governance is necessary. Protocols are all in a race to achieve some degree of credible neutrality in the eyes of other protocols, retail users/investors and institutions. Larger protocols have already achieved credible neutrality with other protocols, evidenced by their mutual integrations and increasing interdependence. Some DeFi projects have managed to attract a regular crowd of users, but this still is negligible compared to centralized businesses. In June 2022 just 10% of volume originated from decentralized exchange platforms. Whether its because of lack of awareness or fear/distrust, were most of DeFi to ossify now, it would probably never be used at anything near global scale. There is still a need to allow protocols to grow and evolve while continuing to appear as decentralized as possible. Some form of token based voting is required to achieve this for now.

Negatives of DeFi Governance

An open, token-based voting system, especially one where 1) token supply in voting locks > token supply elsewhere (CEXes, LP pools, Lending platforms, etc) and 2) anyone can submit a proposal are vulnerable to a number of direct attacks. I will cover some examples of these in the following subsection. In this section I will cover the more diffuse issue with tokenized governance: inefficiency.

Rule by the masses has never been a completely ideal system. Most people probably don’t know enough about the issues being debated and others may be knowledgable but apathetic towards the governmental process. As such, direct democracy has not been used at scale since antiquity, and even in those times had its detractors. Many philosophers including Plato in his Republic advocated for an aristocratic rule as opposed to tyranny of the masses. The relatively out-of-context reference to Plato is in fact relevant because his works, as well as the multi-generational philosophical tradition he helped to set in motion, has largely concluded, for thousands of years, that direct control given to ‘the people’ (with no forms of delegation/representation) is not a desirable nor sustainable system.

Despite this, one could argue that communities today are still small enough that they can function under what is certainly an inferior system of direct participation. The second chance afforded by having a ‘community-driven-reset’ button can sometimes outweigh the consequences, which can include occasional stalls and periods of inaction, as well as trolling and spam.

Trolling

Trolling has lead some protocols to either limit who can post official proposals (to the snapshot page, forum, etc) or increasingly rely on moderators (censors) to direct conversation. This, of course, hurts credible neutrality, as any rational person must assume that community managers and moderators, who likely have financial stake in whichever system is being discussed, may well have incentives and motives of their own. Although many, if not most, are acting benevolently one cannot assume this to be true always. Not to mention being a forum janitor is the perfect pretext to quietly steer group discussions towards an intended outcome. The best summation of this flaw recently has been in the chaotic Optimism token launch last month. Optimism’s airdrop of governance tokens to its users went less than smoothly, when traffic from the ~250,000 wallets that received tokens overloaded the publicly provided RPC endpoints, causing widespread outages for users. Many were sidelined for much of the day while other users who weren’t began dumping on them, causing the price of the $OP token to plummet in the first day of mainnet deployment. In the aftermath, many community members felt angry about lost (free) money, as well as the UX problems regarding the RPC. One user was so mad he drafted an Optimism proposal to ban all sellers of $OP on day one from all future airdrops. Soon after, Cobie, seeing the humor in this bagholder’s anger, drafted an ironic post mocking the original. His post was unlisted by mods while the original was not, due to 1) mods being unsure if this was ‘really’ Cobie, and 2) their suspicion (which was right) that Cobie was trolling. This begs the question, however, who gets to decide what constitutes a troll post? Although one was a joke and another wasn’t, both were clearly untenable. TrollPressure in this case exposed the arbitrary power certain individuals are allowed to wield to keep the ‘system’ appearing legitimate. A major CT influencer making jokes in the forum has far worse optics than an angry bagholder. Granting this responsibility to someone, however, puts a lot of power in that someone’s hands.

Fatigue

There are a series of assumptions about ‘active governance participants’ that I find to be generally true (anecdotally):

• Those involved deeply with governance discourse tend to be the biggest bagholders
• Most active participants participate across multiple communities simultaneously
• The only upside these token holder participants receive comes from token appreciation, and occasionally protocol revenue. There is no direct compensation for voting
• Small bagholders can easily make enough noise to derail productive discourse

The result of these factors, which I assert (baselessly) to be true 75-80% of the time, is a degree of fatigue among the most value-additive voters. Lewis, Pope and Lotti recently found, through interviewing some of the largest governance participants on Uniswap, that fatigue is widespread. This was in spite of a Uniswap Governance structure they described as being considerably minimized. Among these large participants, just 25% of them are actively either submitting proposals or engaging in discussion regarding the proposals of others. The remaining 3/4 simply vote. Their view of the current state of Uniswap governance was:

“the governance process itself, especially when it comes to on-chain issues. Other highly discussed themes were compensation for contributors (including delegates), and power asymmetry, indicating the need to address these important blockers before turning to ecosystemic issues in web3”

Even in a system that has worked relatively well since the token airdrop in 2020, there are still complaints about the speed and agility of the DAO to get anything done:

“Governance moves so slow. There’s just so much coordination we have to do.”

This sentiment is echoed by Fred Ehrsan in Governance Minimization: “What has been demonstrated is the opposite: that rapid evolution can occur without on-chain governance. For example, early Ethereum and Uniswap are protocols that evolved quickly through hard forking alone. This agility can be attributed to a combination of tight knit communities, lower systemic importance, and high levels of trust in core development teams”. A potential path forward for Uniswap and other maturing protocols is to emulate the early days of agility by first minimizing the things that need to change, then slowly automating governance out of the picture. Part of this plan also necessitates skillfully using compartmentalization, via the spinout of subDAOs, where needed. For Uniswap, delegates have discussed this in the context of the ‘fee-switch’ (the mechanism which enables revenue generation for token holders of $UNI from fees): “For example, every pool can be voted upon for the V3 fee switch, i.e. USDC/ETH pool, and also every pool can be voted on for trading basis fee %'s like the stablecoin pools proposal. These two things in themselves are a lot of votes to put through a traditional forum proposal voting process” (source: quote from Lewis, Pope and Lotti). Even the most efficient, well-run protocol DAOs seem to be facing the same issues of disinterest and fatigue; lacking a plan forward in these situations will likely cause problems in the future.

Recent Failures of Governance Systems

In addition to the slow rot that governance can sometimes incubate within a DAO, there are also short-term attacks that can decimate an entire protocol within one block confirmation. These events are relatively infrequent, but remain a potent threat, with many attacks causing losses of millions of dollars. Access to too much external liquidity and misaligned incentives are typically to blame.

MakerDAO / BProtocol Conflict

The Maker / BProtocol incident of late 2020 was a masterful use of flash loans, then a relatively new technology, to attack a token weighted governance system. Growing frustrated waiting to be whitelisted for MakerDAO pricing oracles, BProtocol decided to borrow $7 million worth of MKR via DyDx and Aave and use those funds to push through a proposal before returning the MKR at the end of the transaction. The ‘attack’ (which did not break any rules of the MKR ecosystem or exploit vulnerabilities in the smart contracts) exposed a major flaw of multirole governance tokens: large amounts of liquidity not locked for governance allow external actors to obtain sufficient voting power over short timeframes to overwhelm all the other actors in a system.

Less than one week later a proposal passed on the MakerDAO forum that implemented a waiting period before vote execution, so that the community would have sufficient time to undo these types of actions in the future. Since then, Maker has not been subject to these kinds of attacks, relying on an activist community to intervene as a court of appeals on governance matters. The voting escrow system used by Curve Finance achieves a slightly different workaround to the same problem, by requiring active voting tokens to be locked up before they can vote. This defeats flashloan based attacks but still remains open to attackers using the same strategy on a longer timeframe. MakerDAO is still, however, the farthest along in the path to degovernance in the DeFi space. This attack only highlights the vulnerability of even the ‘safest’ governance systems and thus the urgency we should have in trying to leave them behind.

Juno Hard Fork

Juno is a cosmos tendermint based smart contract chain that recently encountered a brief governance crisis during its token airdrop. Within the tendermint ecosystem ‘stakedrops’ are relatively standard. Stakedrops target stakers of ATOM on the Cosmos hub for distribution of new tokens. Juno were keen to not over-distribute tokens to large whales and mandated a one entity, one claim rule via community decision. Following the token generation event it was discovered that a particular large actor gamed the system and managed to claim from multiple wallets. Multiple proposals were put forward to reorg the chain in order to burn tokens from the wallets of this user, with the final proposal that passed mandating that all but 50,000 of the JUNO the entity received be burned. This proposal passed but with a small margin. The political fallout of this decision calls into question one of the fundamental principles of crypto. Blockchains should be immutable whenever possible; on-chain governance enforcing seizure of property goes against this. Juno is a reminder that the bleed of social, subjective decisions slipping into mass-token governance leads to outcomes which may be morally grey. Vitalik has mentioned in his blog (Governance, Part 2: Plutocracy is Still Bad) that he does not favor on-chain governance for base layers due to the murky blend of on-chain consensus enforced decisions that inherently occur outside of the state of the chain, such as hardfork upgrades. In this case the Juno DAO does not have a direct means of ensuring the hardfork goes as planned (what if other parameters are changed as well) other than the likely presence of core team members as DAO members.

Compound / Justin Sun Conflict

A minimally governed system allows governance to change numerical parameters but discourages or prevents subjective, prose-based proposals that make changes that require upgrades or human intervention. Compound governance is largely concerned with parameters and the addition of new assets, but showed in early February 2022 that token-weighted voting is still vulnerable to ‘attacks’ even on updates to parameters. Justin Sun, the creator of Tron intended to push through a proposal to change the collateral factor of TUSD from 0% to 80%. Because Sun’s wallets are tracked closely by external actors, the $13 million worth of COMP borrowed triggered alarm bells and the tokens were returned before vote execution. This event shows how subjective decision making on complex topics, such as whether or not a particular token is suitable for integration does not mix with token weighted voting (although this is a parameter change decision, it is necessarily informed by subjective risk-factors). This is because 1) votes can be bought, which defeats any due diligence core voters may have conducted, and 2) even without bribes or borrowed votes the average voter may not be capable of making an informed choice. Semi-subjective decisions like whitelisting must be automated as soon as possible to avoid these attacks.

EOS

In 2018, EOS was the hottest chain on the block. Block.one, the company responsible for much of the development of EOS and the initial push to catalyze an ecosystem around it raised $4.4 billion in one of the largest ICOs to ever happen. EOS runs on DPOS (delegated proof-of-stake), which implements a form of on-chain token weighted governance. At the time the rewards for being a validator were so high that it began to foment east-west tension over representation on the chain. Zhuan Lan described the politicization of EOS early on as such:

“Any super node candidate can canvass votes in any region, and as an EOS holder, you can also vote for any super node candidate regardless of country or region. But nonetheless, from the very beginning of the campaign, country wars began. According to data from Feixiao on March 24, the actual controllers of the top 10 exchanges with the largest EOS trading volume are Koreans, Americans and Chinese. Among them, South Korean exchanges Bithumb and Upbit occupied the top spot in the trading volume rankings. “

Chinese participants felt excluded as cartels from Korea and the United States took over. With only 21 ‘super nodes’ forming the backbone of the network, and hundreds of millions worth of EOS up for grabs at the time, securing a fair stake became a serious issue for Chinese validators. Geopolitical economic conflict, including backdoor dealmaking and the formation of cartels, goes against the global nature of crypto. Vitalik commented on this issue at the time:

“blockchains and cryptocurrency, originally founded in a vision of using technology to escape from the failures of human politics, have essentially all but replicated it”

This incident did not directly contribute to the eventual failure of EOS but did expose the pitfalls of using token governance at scale. Base layers are the only token markets with sizable liquidity and in most cases token based governance at this stage of maturity has hindered rather than improved progress, leading Vitalik and Vlad Zamifir to advocate against the widespread use of on-chain governance.

Steem / Justin Sun Conflict

Steem was a media focused blockchain project that revolved around its primary app Steemit. Steemit is/was a reddit-style forum app that rewards its users for posts and engagement with content with its native token. Steemit was an early success story for decentralized applications, attracting thousands of daily users well before the rise of DeFi or Jpeg NFTs.

But Steem had a problem that went all the way back to its genesis block. The founders stealth launched the chain in order to mine additional tokens for themselves. This lead to a significant stake being controlled by these actors, referred to as the ‘ninjamined coins’. Following the departure of one of the founders from the project, the stake was transferred to the foundation entity.

In 2019-2020, Justin Sun (pattern?) saw an opportunity to grow the Tron ecosystem by integrating Steemit, which at the time was still a popular dapp. He approached the remaining cofounder and offered to buy the entity and the ninjamine stake for $8 million in order to use the tokens, which represented ~30% of the outstanding supply, in order to manipulate the on-chain governance and pass proposals to complete his takeover plan. But the Steem community decided to strike first to retain their sovereignty: they quickly voted to lock the ninjamine stake to make Justin’s tokens worthless. Justin, being wealthy and connected, struck back by bribing major exchanges, including Binance and Huobi, to use customer STEEM funds to vote on his behalf. The exchanges obliged, locking up customer funds for 13 weeks without consent and loading the validator set (called witnesses on STEEM) with Justin’s appointees and allowing him to unfreeze his stake. Once again, multipurpose tokens with significant unlocked liquidity posed a major to the health of governance.

With no options left, the STEEM community decided to execute a scorched-earth strategy. Being so well capitalized, Justin Sun was able to outlast Steemians in what had devolved into a war of bribes between two validator factions. So, they began mass-exodus to a new chain, called Hive that replicated STEEM and Steemit and excluded all of Sun’s cronies from the token distribution. Some users, particularly from the Korean community, according to Decrypt, were largely left behind in the new token distribution. Meanwhile, the governance of the now occupied STEEM chain began freezing their accounts and banning posts regarding HIVE on Steemit.

The STEEM incident was truly a digital war. Millions of dollars were wasted on tit-for-tat bribes that only hoped to maintain the status quo. Millions more were destroyed from the disruptions to the regular growth and development of the ecosystem. It showed above all that on-chain governance at scale is terribly vulnerable to hostile attack, especially in cases where governance blends with network security, and the securing set of nodes is quite small. Furthermore, a protocol that can evolve in a positive direction can also move in the opposite way. The promise of on-chain governance, which was supposed to guide a project into its future, ultimately killed it before it could mature.

Lessons

I view blockchain politicking as a source of entertainment. Watching a political system blend different properties of a royal court, committee-based rule and democratic assemblies, backdoor deals and coups, bribes and debates, is endlessly fascinating. Although in many cases, such as those described above, these systems have destroyed millions of dollars in value, some ‘conflicts’ have resulted in positive outcomes, such as the Curve Wars. Clearly, however, governance in its current form appears more like an attempt to feign credible neutrality for a platform and its contributors, rather than a way to introduce any efficiencies. More bureaucracy and politics are not a good thing. This is true of both meatspace and the internet. To become lindy, protocols will eventually need to adopt a political framework that destroys itself completely over time. Else, they are beholden to the same forces that tear apart any system reliant upon functional, altruistic, engaged participation in perpetuity. In the next section I will cover current thinking on the evolution of governance and visions of what a degoverned endstate could look like.

Where Governance Is Going

The first step in moving towards a degoverned state is establishing a program of progressive decentralization that enforces a gradual handover of power first to a DAO, then a systematic elimination of the DAOs power until the protocol has ossified. A number of strategies have been proposed to facilitate this transition by evening power across token holders. Novel forms of governance have also been a subject of discussion as replacements to one-token-one-vote in an attempt to improve existing governance systems (while they are still needed).

New Schemes

Quadratic Voting

Quadratic voting is a system that limits outsized influence by a small number of influential actors with an unequal share of capital. The cost of effecting outcome x is related to the number of votes squared; cost = (votes)^2 . Now, the cost to the user increases faster than they can acquire new votes. Therefore there it is a diminishing return to the voter as they try to gain more influence.

The mechanism for quadratic governance was created by Vitalik in 2019, but Quadratic forms of governance have been a topic of discussion within and outside crypto since at least the 1980s.

Futarchy

Futarchy is a voting based system in which the participants place bets on a future outcome using prediction markets. The goal or desired outcome is measured by a numerical metric. A liquid prediction market is created that allows participants to speculate on whether or not enacting a certain policy will generate a value of the metric that is at least at the level of the desired outcome at some point in the future. Originally introduced by Economist Robin Hanson in 2008, it has been adapted to DAO governance by Vitalik Buterin and others.

A DAO governed by futarchy would create a two token market for each proposal at the beginning of the voting cycle. The outcome at the conclusion of the cycle is determined by the relative valuation of the two tokens. By linking the future wellbeing of the DAO to economic incentives for participants based directly upon their opinions of a certain proposal, incentives align between all parties. The profitable outcome, therefore, will typically be the most profitable one.

Futarchy has been criticized by some because it fails to account for the self referential nature of markets, the potential for price manipulation of the prediction assets via derivatives, and the difficulty in reducing subjective goals to numerical metrics (how would one measure happiness or inclusion?). These concerns were made in the context of governments, which must coordinate to make complex, subjective decisions all at once, but become less relevant for a protocol DAO, that makes decisions that are often measurable (such as earning yield for the treasury via various strategies) and fall within a specific domain. As of June 2022 futarchic DAOs have not been implemented in large numbers, despite ongoing discourse on the topic from Vitalik Buterin and others. One example of a futarchic DAO today is Zeitgeist, which uses prediction markets and futarchy to govern its prediction markets protocol built using Substrate.

Examples of Progressive Governance in Protocols

MakerDAO

MakerDAO is one of the oldest DeFi protocols and therefore is among the most robust in regards to governance. The Maker foundation was dissolved in 2021, putting full control of the protocol and its finances in the hands of the DAO. In May 2022 Maker founder Rune Christensen published “The Endgame Plan”, which aims to lead the DAO to “a predetermined, immutable end state many years out in the future”. Similar to the many subsidiaries that exist underneath Alphabet Inc, which each manage separate budgets and succeed or fail on their own, Maker seeks to compartmentalize different aspects of the protocol and isolate their decision making. This begins with focus groups of voters who coordinate their decision making before voting, then evolves into task forces of experts in particular discussion. These task forces then become DAOs of their own and are delegated authority over their area of expertise by the main DAO. Thus, the central DAO must only focus on the “core” of the protocol, which generally does not change/require frequent governance, and the creation/destruction of various MetaDAOs.

This design minimizes the number of decisions the central DAO must make, and optimizes each thread of decision making to contain subject matter experts. Maker as a protocol is far closer to being ossified than any other and is thus positioned uniquely to drastically advance its governance towards minimization in the near future.

Reflexer

Reflexer Finance is a stablecoin issuance platform with Ethereum tokens as an underlying asset. Its stablecoin, RAI is not pegged to $1 as is common place, but has a floating exchange rate. The aim is to create a dampening effect on the underlying collateral rather than targeting a particular dollar price.

Reflexer makes use of the PID (Proportional Term + Integral Term + Derivative Term), a concept from Control Theory, in order to set interest rates in response to changes in the underlying value of the collateral. One advantage here is that parameters for such a system can be determined upfront and adjusted only when absolutely necessary. For everything else the PID can manage itself.

Reflexer also aims to minimize governance through multiple avenues, including:

• Time locked governance: once changed a parameter must endure a cooldown period of T seconds before being changed again
• Action locked governance: a parameter n will have a range of values that are allowed
• Governance ice age: after deadline D most governance will be locked, denying further intervention. This is key for reducing social dependences and creating a robust long term solution.

The protocol can also shut itself off in cases of severe emergency. Should community desire to intervene in places that the rules of the smart contracts do not allow, a redeployment would be required. $FLX is set on a clear path towards not being used as a ‘governance’ token at all.

Babylon Finance

Babylon Finance brings investment club style vaults to Ethereum allowing users to pool funds together under a particular strategy and accumulate yield. Babylon finance therefore is essentially a series of factory contracts, that allow users to deploy their strategies and earn yield on them with other users. There are still, however, other parameters particular to the factories themselves that can be edited. Although Babylon admit that some governance action may be required in perpetuity, they are also pursuing a policy of governance minimization, based on upon writing from Ehrsam and Robinson’s 2020 article “Governance Minimization”

As such the $BABL token holders will only control fees, treasury deployments, and whitelisting for integrations with additional DeFi in the medium to long term. The separated nature of each individual vault (called ‘gardens’) and the overall governance related to the factory contract may never be fully ungoverned, largely because at maturity its only function is to collect fees and do something with them; this is not something that necessarily needs to remain static in doctrine forever. Despite this, Babylon have put governance minimization as a core principle for development.

Conclusion

In addition to the projects mentioned above there are dozens of others who are also pursuing a program of governance minimization. I suspect that much of DeFi will successfully navigate themselves away from “Governance Hell” given the growing body of research, both theoretical and practical, that suggest minimizing governance is the ideal solution.

Endgame Scenario - Conclusion

Governance solutions and innovations today revolve around minor tweaks and improvements to increase efficiency and bring about a roadmap to minimization far into the future. But what could this end state possibly look like? In 2050, assuming humanity is still alive as a species and economically functional, 100 million or more humans could be interacting directly with DeFi protocols and their native tokens. Millions or billions more will interact with DeFi indirectly through normie-friendly platforms.

Modularity and compartmentalization are the most important early factors to accelerate minimization of governance. Narrowing the scope of what is currently governed by one particular group makes ossification far easier while also incrementally increasing efficiency along the way. MakerDAO, discussed above, is an early leader on this front. The less parameters can be changed, the fewer opportunities any particular actor will have to disrupt the function of the protocol.

A series of ossified modules will makeup the entirety of DeFi. If one were to be extremist in their anti-governance views, then everything above the level of complexity of a factory contract ought to be decomposed. Whenever a factory deploys a new vault/pool/etc, the deployer(s) have the option to set their parameters. If they need to be changed later on, then the contract must be re-deployed; liquidity will follow to the new locale if-and-only-if it finds these changes to be acceptable, rather than being forced to accept updates that effect their capital without proper voice.

Boxing out voters is the best option for crypto’s principles to survive going forward. Any system that can be governed is prone to capture by actors with bad intentions. Nobody in crypto wants to hand power over DeFi protocols to big banks, silicon valley investors, the Government (big G, the one in DC), financial institutions etc, even if these institutions eventually begin using the technology. The ideal end-state is one in which tradFi/tradTech play on our terms - they do not deserve a say in any future systems no matter how many tokens they may own. Ossifying protocol parameters as soon as possible will protect DeFi from malicious actors in DC, Wall Street and San Francisco. “Permissionless” means anyone can use/interact with the deployed contract binary, not that anyone can define or reshape its future going forward.

Sauces

Disclaimer

I own some of the tokens mentioned in this piece. I do not advocate for or against investing in any of the mentioned tokens, or any governance token at all, based solely upon the information herein. NFA DYOR.