The Next Generation of Decentralized Governance

All right, so yeah, I'm really happy to be here. Take a moment to just enjoy that for a second. Hey, everyone. So I'm going to be speaking about decentralized governance today. I'm Trey Kioptririx.

I'm a dinosaur. I am the chief governance officer at EigenLayer, or at the Eigen Foundation. That's where I'm working. Previously, I worked at, I was a contributor to Yearn Finance, where the gentleman who's leaving right now spoke before Gabe Shapiro and I both designed a governance system called Constrained Delegation, which informs a lot of the work that you just saw in Borgs and also what we're building at Eigenlayer. I also was a co-founder of Coordinate, which is a system for decentralized compensation.

So I want to start, why does it even matter? Why do we need decentralized governance? Why is it important? I think we talk a lot about governance minimization, which is really important. If you can minimize it, do.

But there's no way to minimize things at the frontier. You can't reduce the most new, confusing, misunderstood things into immutable code. There are periods where human beings have to come together in subjective reality and reason things out and make decisions together. We need some form of collective decision making, and we always will. And the forms that we have now and have for centuries aren't so great.

They lead to wars and various other things. So beyond just the ability to launch a token and not get sued, decentralized governance has the potential to do much, much more. It has the potential to give us ways to not stop killing each other, for instance, which we haven't figured out as a species. It's a tall order, and there's a lot to do to get there, obviously. But, you know, here we have three predators that are already doing it.

So we're going to take our cues from them. I want to talk a little bit about what DAOs are doing now before we talk about the future of decentralized governance. So today, DAOs are really good things, like increased transparency in organizations. Beautiful. More permissionless access to information and control.

Capture resistance. more permissionless access to information and control capture resistance all these on-chain technologies allow for much greater security and to reduce capture by small groups give more access to people and they allow for decentralized decision-making which has been an incredible advance over the past you years, eight years. But it's not all happy stories, right? There's a sad dog as well. For instance, DAOs are pretty bad at power concentration.

If you look at a lot of what seem like decentralized DAOs, you find that under the hood, they're maybe not so decentralized. They're also bad at skill-task matching, which is something we'll talk about a little bit more. There's tons of principal agent problems all throughout DAOs and low accountability. And while they're doing decentralized decision making, I'd argue that it's pretty ineffective low quality decision making in most cases. So what's the next generation?

So far we've done well at decentralizing. But what's next? And it's pretty simple. It's making them decentralized and effective, which we don't really have any examples of today. But we're getting there.

And so that's what we're working on at EigenLayer. So EigenLayer, we're doing Ethereum restaking, helping create a new ecosystem of AVSs and taking things off-chain, putting them on-chain on Ethereum to expand what's possible. This is a big, ambitious vision. It's a big, open ecosystem that we're building, and we need effective and decentralized governance. We need the people building on EigenLayer to have control and ownership over what happens on the commons that we're creating together.

And it has to be, you know, for now, it's a tricky question. Decentralized or effective, right? You can kind of choose one. We don't have the technologies to do both yet. So in the beginning, it's kind of okay to be a little bit more centralized.

And if Lefteris is out there, we're not a DAO yet, right? We're trying, but we're not yet. We're starting more centralized, and we're going to decentralize over time. And that's because we're focusing on quality decision-making. We're not going to give that up, and we believe that we can do both.

But later, over time, it really does have to be centralized because you can only trust, you know, in the beginning the participants in the ecosystem need to know that Eigen layer is making good decisions, right? But over time after the, you know, honeymoon period is worn off, you need to trust that they're not going to screw you. And that needs to be decentralized. So we have this process we call incubation, which is our version of progressive decentralization. And in progressive decentralization usually, you know, you have a foundation which has the power, and it controls everything, and then it gives a chunk of power to a DAO, and then another chunk of power to the DAO.

It's good. We're adding more gradation to this. So our version of this incubation, we take decentralized structures, and we operate them in the open as if they were decentralized, but they're controlled by the foundation. They're centralized to start. And we're not pretending they're not, right?

But we're doing this because that's how we learn. That's how we can learn to make these structures actually operate in a decentralized way. And then over time, we decentralize them bit by bit in a public, you know, open discussion with the community. It's because we have two contradictory things that we have to do. Like we have billions of dollars in TVL.

We can't play with that. We have to make sure that that's secure. So we need both a minimal, secure, and stable governance system to minimize risk, but we also need to do something that hasn't been done before. We need to make it effective, too. We have to be radical, innovative, and experimental in order to invent this future.

So in order to do that, we've created what we call version control governance. It's like a two-track system of governance. We've got the core system, which is the secure layer. It's handling just the minimal stuff that's needed to make sure the protocol is working. And that stays, you know, if that needs to be more centralized, okay, it's as decentralized as it can be.

And then we have the vision track. And the vision track is where we're experimental. And we're creating the systems that will be able to merge back in over time into core once they've proven themselves to be effective. So let me tell you about what we're making. So it's I can gov is what we're calling it.

It's this is going to be a very brief overview. We haven't launched this stuff yet. We will soon. So if there's a lot of questions, some of this might not make so much sense. I'm going to do the best I can.

We'll be publishing a lot more of this over time. So in designing iGov, we started kind of at the beginning. What do organizations need to do? And there's kind of three things. There's decisions, there's decision makers and then there's decision making and that's how organizations both centralized and decentralized function decisions are the same stuff that we've always had what should we do who's in charge how much does it cost it's variations of this so we don't need to worry too much about that we're not fundamentally changing what a decision is but decision making decision makers there's a lot of room for change decision Decision-making has done a lot of development over the past five years on-chain voting, off-chain voting, multi-sigs, rough consensus, all the different things that we use to do decentralized governance continue to be developed, continue to need more advancement as well.

Decision-makers need a bit more work. So decision-makers are things like unique human beings, which supposedly can make good decisions in democracies. We also have tokens, which supposedly can make good decisions in DAOs, or delegates, where you take 100% of your voting power and you delegate it to another human to make decisions on 100% of the things in a DAO, regardless of what they're good at or not. These are not such great decision makers, in my view, and so we're trying to improve this. The first step is we're not doing token voting.

Token voting itself, I mean, this is a bit of a nuanced one because we are doing token voting, but we're not doing normal token voting. Token voting in its normal form, you know, is what I said. You have a governance token, and whoever owns this token now can say things about any matter. They're a security expert, or they're a capital allocation expert, but this isn't true. And one of the myths, I think, that we need to break and douse, I think there's this idea that's kind of beautiful that we're just going to be able to put together a big group of strangers and the collective intelligence of that group is just going to solve our problems.

We're just going to give our responsibility to this group and they're just going to make things work. And that's a myth. That's not true. Decision making, business, operations, organizations, these are things that require responsibility, accountability, relationships over time. You can't just give this to an anonymous group of people.

Now, anonymous groups of people can make certain classes of decisions quite well, but they can't do these in a vacuum. You know, Vitalik's written about this convex, concave decision making. And it's true that you want to estimate an average of certain types of decisions that a large group might do better than a small group. But again, these decisions don't happen in vacuums. So what happens is if you give this power of an organization to just the wisdom of the crowd, you might get decisions that work in a small box but they don't work in reality, like giving hundreds of millions of dollars for grants when you don't have the ability to really oversee it.

And this is not just one group. Many groups make decisions like this. But if you're a small group of people with relationships and accountability and things at stake, you probably won't make those decisions. Another thing about token voting is that, you know, going back, tokens, using a crypto-economic token to, you know, provably make a decision about something is a great system. That's great.

The problem is who are the deciders that are using that? What is the structure around it? Often there's this idea about governance fatigue, like this is the problem. Governance fatigue, if you think governance fatigue is the problem, you're seeing the thing completely wrong. It's not, that means you're giving people the wrong jobs to do.

What you need to do is we need to give people the right jobs to do and then and they won't be fatigued. So in Eigengov, the starting premise, we're not sacrificing on decision-making quality. All decisions are made by small groups of high-context, high-trust domain experts. That's the foundation. That's where we start.

Why? Because we do have a superintelligence on this planet, and it's not an LLM, and it's unsure if that will even get there. The greatest superintelligence that we intelligence on this planet and it's not an LLM and it's unsure if that will even get there. The greatest super intelligence that we have on the planet is a group of seven people that trust each other and have worked together for a long time or fewer. That is the smartest thing that we know of and we've known this for a while.

But a lot of research shows is after about seven, intelligence drops about 10% per person. And once you get up to large group sizes, you get pretty dumb. So we're trying to hit that sweet spot of a small group. And this is, you know, like a family, like a small team. This is what does the best decision-making.

So we start there. Why? And then we can talk a lot about why, but if you look at the relationships, so work in life and creative work, it's about relationships. If you have a group, you have two people, one relationship, three people, three relationships, then four, six relationships, 10 relationships, 15. Each relationship needs to be maintained.

Each relationship is a vector that can have misunderstanding, conflict, tension, confusion. So as you grow, it becomes unwieldy. But we can manage in small sizes. This is slime mold. It's not a non sequitur.

So if we wanted to keep scaling and just think we want a homogenous group of decision makers, the best we can think of is maybe slime mold. So slime mold is all these different cells. They're all kind of the same cell, but there's lots of them. And they make decisions as a group, and they do pretty miraculous things. But, you know, it's kind of limited.

What we need is actually something more like a fly brain. Now, this is a heterogeneous group of centralized and decentralized actors woven together in really complicated ways to make complicated decisions. This is what we need for DAOs. And so, I can go starts with this idea of small teams making decisions. And it starts with the idea of councils.

And councils, a variety of councils will talk about them, is how all decisions get made. Now, how do the councils get populated? Elections are a great way to put popular people in charge of things, but not a great way to necessarily put the right people in charge of things. We've created a system called endorsements, which is a way to create reputation. And decentralized reputation is a research area that we're super invested in and will be very important in the future.

A lot of people are working on. So we'll talk more about endorsements and also decisions. Decisions is how we actually weave all this stuff all together And we're working with incredible partners like Tally and hats and EAS and others to build all this stuff So this by the way is speculative so left terrorists don't yell at me if this isn't what happens We plan to make a variety of councils and And using the incubation process, powers will incubate within foundation and then they will be spawned into councils, groups of powers, and then over time as they mature, spawn into more councils, pending the amount of overhead needed to make all these things work together. But this creates a better distribution of powers more checks and balances more specialization so each council is a group of high trust high context expert domain experts in the field that they're making decisions on each council has a limited scope of power in the constrained delegation model which they can act in and many of these will be borgs just like like what Gabe was talking about previously. Endorsements, how we find people for these councils, it's similar to delegation.

This is a sneak peek at the endorsements platform that we're making. But instead of delegation, again, you delegate 100% of your voting power to one person. With endorsements, when you sign up to be a contributor, which is our version of a delegate, you say, what are you good at? So Roger's good at auditing, governance, and leadership. And so when you trust Roger by using your stake diagram and allocating it to Roger to make decisions, it's according to those specific skills that he's now empowered.

And this does a number of things in the system. It also increases your voice. So you'll be able to signal on all proposals. We'll talk about this in a second, rather than just eventually filling council seats. So I want to talk about how decisions get made.

This is what a lot of DAOs, the model that a lot of DAOs follow. So you have token holders that can either become delegates themselves or they can delegate their token stake to delegates. And then those delegates vote on proposals. And those delegates then elect members to councils who can also occasionally vote or veto on proposals. So we changed this model quite a bit.

In our model, token stakers who have something at stake, they're not just holding the tokens to sell them, can be curators or contributors. And I'm going to pause here for a second because this curator one is kind of new. So when you're a curator, you are making endorsements. So perhaps you're a token staker, but you don't know who is the best at OPSEC in the system. So you can just assign your tokens to a curator, and then the curator is the one that's paying attention to all the contributors in the system.

And then similar actually to what Denison was talking about previously in the talk, when you want to make money from staking tokens in a governance system, you want to make sure that if the incentives are coming from people taking action in governance, which was our plan as well, then you want to make sure that your endorsements are going to people that are going to be creating value because you'll be incentivized for it in the future. Curators also solve a big problem in DAOs, which is liveness. Most DAOs, there's an airdrop. People come in, they claim their stuff, they delegate to it, they send their tokens to a delegate, and then that never moves. That token power is just not moving.

That's not a very good map of the world. For an organism to be intelligent, it needs to create an updated map of its environment, and so liveness is key to that. These curators, people who are curators are going to be incentivized for keeping a high quality map of who is contributing, as well as evaluating those contributors over time, which is outside the scope of this talk for now. So you can assign a curator, or you can endorse a contributor for various skills. Now the contributors that get the highest endorsements with most trust and relevant skills can then be eligible for councils and eventually will automatically fill councils and then councils vote on proposals.

But token stakers need to have ultimate authority. So they can veto proposals in many cases that they don't agree with. And in the worst case, they can fork the intersubjective eigen token if the governance gets captured. So and then the last piece here is about expert signaling. So this is how decisions get made.

There's a discussion phase, then there's the council voting phase, and then after that there's the veto period. That whole time there's a phase of expert signaling. So contributors that have gained reputation in the system for various skills can weigh in on every proposal. And this is a sense-making advancement to support council decisions. It's non-binding but it's important and the people that are in the community need to be in discussion right so if there's a controversial proposal and the the community can discuss it, experts can signal on it and the council can vote and the council might have some idea if they're gonna get vetoed or not before that stage happens through this right but you want that small group of people to be able to take anti-majoritarian action because they might not know things that the rest don't and they are there for a reason.

So there's a balance. And that's most of it. So I just want to talk a little bit about the future. We have a lot of research to do and a lot of work to do to make all this happen, right? Because this stuff hasn't been figured out yet.

That's why it's not a DAO. It's not tested. We have to work on it before it's really decentralized. So there's a few areas I really want to focus on. One is we need to create a system of decentralized reputation.

And there's a lot of work there. There's a lot of things I want to say about it, but I don't have time, so I'm going to skip it for now. For decentralized reputation to work, we need decentralized identity. And we also need evaluation systems, accountability, incentivization systems for these types of decentralized systems. Because if we don't lot of these things.

We also need to do productization. DAOs are a mess. We need to apply product development standards to the way our DAO tooling and systems work, so that all information and processes have UX that works, people can understand and figure out how to use. And we also need to have credible commitments. This, along with decentralized reputation, is what will allow for these things to really function at a high velocity and change this kind of messy, beautiful experimentalization period into really the future of work, where high-velocity, small teams of groups come together in ad hoc ways to create products and incentives and get retroactively funded, build things, deploy them and then break up and go on to other jobs.

The nature of the firm is changing dramatically from these large centralized vertically integrated systems to these small groups of autonomous sovereign contributors able to gain reputation, to gain incentive, to make to make use of their gifts in the world because all these problems that we have the Meta Crisis everything is all solvable with all the gifts in all of you and all of us in the world if we can figure out a governance system, a system to support us in offering all of these gifts to the world and in service of solving these challenges and taking our civilization to the next level. So that's everything. I'm Triky Optrix. Thank you. Thank you so much.

Thank you so much, Amanda. That was awesome. And people, you're finally using the application to ask the questions. So we have a few questions here already. Do you want me to...

I'll use the votes. I'll use the votes to start choosing the questions. So the first question is, the endorsement system is overly similar to expertise endorsements on LinkedIn, which were removed due to their ineffectiveness. How is this different than LinkedIn with tokens? It's a great question, and it's not oddly similar.

I mean, we looked at systems like LinkedIn. Without a good system of decentralization, of reputation, without a way to incentivize good endorsements and punish bad endorsements, it's not much any different than LinkedIn. So right now, it's quite similar. In the future, it will be quite different. Awesome.

Next question. Consuls with limited scope can have domain experts that are in multiple consuls, leading to a scope creep. Any plans for council service limits or requirements? Yeah, definitely. There'll be term limits, and there'll be eligibility requirements, and there'll be legal contracting in some cases.

There'll be KYC in some cases. And I don't know, I think probably people will just be able to be on one council at a time, but there might be some cases where it's appropriate to be on two. We'll figure that out as we go. But being on two could create conflicts of interest that we want to avoid. We'll have to see.

Awesome. Tough questions. The current design seems similar to OP's promise of we will decentralize. What accountability will Agen have to actually make progress? We know we don't.

We don't have any. I mean, companies can do whatever they want, right? Like, OP doesn't have to decentralize. I think you have to look at what's in our interest. So if EigenLayer doesn't decentralize, I think over time that it causes a lot of problems for us.

And so we can say whatever we want, but look and see what we do. Awesome. So how to avoid that delegates just elect councils who will create things good for them, then later they will approve? Yeah. So there's three key things.

So I just kept it to decentralized and effective, but actually there's a third one that's really important too, which is alignment. So you want to, it's the principal agent problem, right? How do you know that your investment broker isn't just making trades that stack his bank account or her bank account? This is, I like to turn this around and we actually think of it as the principal agent solution because the alternative is that you are both principal and agent and nobody's got time for that, right? So it's a huge solution to be able to split things up from ownership to control.

But how do you keep that alignment? It's super tricky. And most of our research path is designed around solving exactly this question. Decentralized reputation is a key piece of that. So if, which is also outside the scope of this discussion, but let's say that there is, we believe there's ways to incentivize reputation system using things like trust assignments and the endorsements that we're doing, skill assignments, to make it much more aligned than it is today.

But just wait. We'll share more on this soon. Good, good. Any way to transit tree-like decision structure to decentralized structure? You're going to transit a tree-like decision structure to a decentralized structure.

I mean, I'm a little confused by the question. Somebody want to speak to that? A tree-like structure could be decentralized. Are you saying like a hierarchical structure? We can skip it.

Okay. We can go with the last question. And it's the control of token holders is limited, removing value from the token plus accrued value to console positions instead. Abero is significantly less power than the power to do. Why token at all then?

Well, the token isn't for governance. The token is for interest-rejected forking. It's not a governance token. But why use the token in governance? Well, because token holders are the owners of the protocol.

And if they always have a governance action that they can take, which is to sell their tokens. But before you sell your tokens, why not, you know, you should have voice as well. We believe one of the things I could have mentioned in this talk is that everybody in the ecosystem has value. The token holders, the AVSs, the operators, the LRTs, everyone building on it, all the partners, all have value. And one of the big magic of putting together a governance system is figuring out how to help that value flourish in the right ways.

And what's the right job? What's finding the right job for the right group? And token holders have a really important job. They're the ones whose finances are at stake based on the actions of the organization. And so you want to give them a lever.

You want to give them a voice. The problem is giving them the voice of making all decisions for a DAO is just not a very good connection of skills and opportunity, right? Or skills and challenge. So what is the appropriate voice for them? And we feel that the appropriate voice, it's two things, right?

One is deciding on who makes the decisions, right? Which is similar to a lot of forms of nation-state governance. So they can endorse contributors that then have the power to make decisions. And they're in charge of those endorsements. They're in charge of curating the space.

But then they also have this fear response, like in a body, like we're creating a body. There's another thing I could have talked to, there's an executive council that is a kind of authority, the top authoritarian layer, then there's expert councils, most of the councils making these domain specific decisions, then there's the majority layer of token holders and reputation holders that are curating the space, vetoing the space, and at the bottom, there's this self-enforcing layer of token forking. So, yeah, token holders have a really important role, but it's just not the role that we're used to. Okay, thank you so much. And people, please put those hands together for Tricky After Risk.