Defi Governance Endgame

This is going to be a rant about the unfortunate end state of Defi if we continue along our current trajectory. The fundamental issue is that we’re using money for Sybil resistance. Simply using governance tokens as a vote count makes money correlate super-linearly to influence. Vitalk recently wrote about this problem and it has been bubbling on my mind for some time. So let’s start with the basics, how we got here, where we will end up if we don’t course correct, where we want to be instead, and things we need to build to get there. 

Going back to basics, we need to reach consensus. We do this by voting amongst a set of peers whom are unwilling or unable to collude with one another to dishonest ends. Why are they unable to collude? Well, if they collude in public others will rally against them. As a memetic schelling point, fucking over dishonest actors that are attacking the system is tried and true classic (see ETC). So, to conspire effectively, they have to do so in secret. However, as the number of peers in a conspiracy grows, it becomes exponentially more likely that a leak will occur. This creates a game theory scenario that makes it profitable and safe to be honest and risky to be dishonest. The more peers you have, the more anonymous they are, the better this system functions.

In a blockchain, you inherit the problem of Sybil resistance. Addresses are free. So how do you ensure that all voters are unique peers? Bitcoin uses proof of work to answer this. With proof of work you get voting proportional to the cost invested in your vote. Once you vote, that investment is gone. To offset these costs your vote comes with a lottery ticket and occasionally you win the lottery so the expected value of participating is positive. In any case, the actual source of Sybil resistance is money, proof of resources invested in a vote that can’t be faked.

Proof of stake reuses this capital based notion of Sybil resistance. It’s just that rather than your investment being lost with every vote, requiring a constant stream of investment to maintain security, your investment is reused from vote to vote which makes the system more capital efficient. However, it’s still money proportional to influence.

Governance tokens basically just extend the ideas of proof of stake. Starting with the food token craze, every project today seems to need to launch a governance token. Without it, there isn’t enough marketing power. Outside of some notable L2’s people just don’t have enough reason to care about your new thing amongst the thousand other things happening in Ethereum unless there is money to be made. A governance token distribution gets people to market their bags. This is all fine and well until it comes time to actually use those tokens for voting. Then several problems arise.

The first problem is voter apathy. When only 20-30% of the tokens are voting it gives outsized influence to whales. Most projects today don’t directly tie rewards with voter participation. Even if voting uses snapshot and doesn’t take gas, it still takes time and Defi is already past the limit of human attention if you want to be informed and vote on every project. There’s just too much going on for even someone who devotes their entire day to this space to keep up with everything. Clearly if active voters are a required component of an ecosystem they need to be incentivized directly. This goes against pure buyback + make/burn philosophy of Maker and argues in favor of systems like Kyber that explicitly direct profits to voters.

The second problem has to do with coordination amongst a large number of peers. It’s an age-old problem that technology has helped with but not perfectly solved. You can’t feasibly have thousands of people in a meeting with an open mic. Active voices have to be throttled or every meeting will just turn into twitch chat and the system can’t function at all. So how, do we throttle it? Often by voting weight. You get a handful of the most influential players in a room and talk about the most important issues of the day. This creates tiers of users though; those with a mic and those without. If these tiers are defined using money, even indirectly, then those with money have a categorically different amount of power than those without. Not just proportionally more influence, not just on another scale, but entirely new types of influence. If I donate $10 to a senator I get an automated thank you email; if I donate $10M to a senator I get direct personal access. Using governance tokens as we do leads to using money as a filter for tiers of power.

Inevitably, those with power sculpt the systems they control to their own benefit. Getting political for a moment, I’m not suggesting a world where everyone has an equal say in everything. We live in a world today where we can’t agree on basic facts even when ignorance is literally a matter of life and death and people are dying by the millions. We clearly need a system that filters influence to those with knowledge and good intention and money is at least partially effective at this. However, each necessary participant in an ecosystem should have and retain influence. An incentive system should not allow the influence of any necessary actor to entropy to zero over time. Capital, as it is structured today, does this.

Basically this comes down to wealth inequality and wealth accumulation. These are not only at historically dangerous levels but accelerating every year. In historic times this was about when French heads started to roll. I doubt that kind of reset is viable today. The wealthy, at least the type I’m talking about, are no longer country locked, reachable in any material sense by a mob, or really accountable in just about any way for the evils they do.

Outside of just the top sub point-something percent the real wealth of the world is controlled by institutions. This capital is essentially granted the rights of people while being effectively international, immortal, unjailable, and legally beholden to short term shareholder profits above all else. It is a dangerous recipe. They write the laws in their favor. They can manipulate and endanger global economies and and are not accountable for the harm they cause. Even when they get caught and convicted in court they pay a few days worth of their profit in a fine and continue on with business as usual. Most people underestimate the magnitude of the wealth gap. It’s staggering. These players have way more capital than the entire crypto market cap. I really can’t bold that enough. The gap is staggering, and growing every day. The pandemic only widened it.

If you are thinking that Defi is going to change that, think again. While Defi has democratized access to the financial system, the opportunities it provides still require capital. Once the regulatory clarity for institutional custody and voting procedures have been established these players will own crypto. Many already do and are aggressively hiring crypto capable talent. I was recently searching for a job, so I know this firsthand from actual interviews. These institutions have the capital to acquire a disturbing level of influence in this sector.

Wealth inequality is here to stay and it’s only getting worse every year. If we tie influence to capital, then as this process unfolds inevitably all the influence will go to those with all the capital. Whatever cypher-punk anarchist utopian visions crypto founders had, we aren’t heading there unless some large changes are made to correct our course. So what can be done? Vitalik has several answers to this in his post. I’d like to build on a few of them and offer something of a vision for a possible future.

The first has to do with a different solution to Sybil resistance than money. While I’m aware of projects like proof of humanity and BrightID I don’t think a social solution can’t be eventually gamed. This problem exists today on reddit with bot networks that post some repeatable meme and then the bots all kick in and upvote it to bootstrap another account past the Turing test level. It’s an endless cat and mouse game like spam filtering. In the long term I think we’ll see a bunch of Legos taking the form of things like POAPs, some on chain history oracles, social attestation systems, and implicit KYC systems such as being whitelisted by Nexus Mutual and others. While no Lego is perfect the combination of them will be sufficient to enable systems like quadratic funding to function. This can reduce the influence of capital to alone to something resembling a logarithmic function rather than a super-linear one. I’m working on one such Lego personally.

Next, we need a standard delegation mechanism across Defi like xToken provides but tied to celebrity rather than “economic mandate” which is just too abstract to answer every governance vote.  Delegation systems allow informed, public participants to rise in influence above their personal capital. I want to be able to shop for my representative to each project, delegate to them, and split the voting profits with them. Obviously this requires that more projects begin directly tying rewards to voter participation which I think is inevitable. xToken is currently the frontrunner on a standard delegation platform but they can’t support a representative system at the moment because they can’t fragment the liquidity of each xToken. Either they’ll need to crack that underlying problem or another platform will come along eventually and replace them. As a bagholder I hope it’s the former.

Lastly, we need a good pattern for reputation systems that interplay with voting influence. If governance tokens represent monetary influence then reputation is inalienable (perhaps decaying) influence. Reputation would be earned against some combination of objective functions that each project could define. Maybe someone can build some type of pluggable oracle that maintains reputation scores for each project on an L2 which could integrate into snapshot. The details on this need a lot more thought. How are those objective functions defined? Is your monetary influence capped somehow by your reputation? Is there any way of discouraging people from “selling” their reputation by incentivized delegation? How do you solve the nothing at stake problem for those voting with reputation? It’s a complicated topic that needs a lot more work but I agree with Vitalik that we need to move past straight token voting and this is one of the most straightforward ways to do so.