On The Governance of Decentralized Autonomous Organizations

Transcript

Uh, hi everyone. Can you hear me? All right. Hi everyone. Uh, my name is Vuk Parahari and I'm from the Moxplank Institute for Software Systems and today I'm going to be talking about Dows and how Dows govern and specifically the different design choices that they make and their implications. Um, now we all love DeFi. Um, and the reason that we love DeFi is because you can basically do a bunch of actions in a decentralized manner and you can just place your trust in the smart contracts, right? And in a similar vein, Dows were basically created as a way to basically own and govern DeFi protocols in a decentralized manner. The

way that this works in practice is basically you have your nice DeFi application up on the left. Uh, then you have your governance contract uh in the middle. I don't know why the thing isn't showing up. And then you have basically your DAO that basically uh controls the governance contract and uh if you want to make any change to the DeFi application, you make it through the governance contract, right? And the DAO basically consists of like a group of token holders and that's what a DAO is, right?

And essentially uh you might think like okay like all all you need to do if you want to participate in governance in a DAO is uh just hold some tokens but that's not actually true. Um all these rules of how DAO governance is actually done is actually placed in the smart contract. So for example um the DAO actually decides who they actually allow to participate in governance and who they don't. So for example, if you want to participate in some proto some Dows, you actually have to register what your wallet address, you may need to put your tokens into a staking contract, right?

And uh these uh these governance decisions are actually written down in this governance contract in terms of codes, right? And now that many jurisdictions are actually starting to make laws on Dows and uh stipulate what an DAO actually is, it's actually important to understand the different dimensions along which a DAO is made. So for example, this is a law from the uh Wisconsin DAO law which actually states here that an algorithmically managed DAO may only form under this chapter if the underlying smart contracts are able to be updated, modified and otherwise operated. Now this really makes no sense because under this law basically like a

unis swap contract could be called a DAO because it can be updated, right? Like updated or modified by who? Does the judge have a key? uh what kind of powers does maybe like maybe they mean that law enforcement should have a key, right? But like it's not clear what kind of power they want in the Dow, right? And so in order to make these decisions a lot clearer, what we did is like we actually analyzed the top 51 Dows on Ethereum and essentially we looked at the Sorry, I don't know why there should be text on top of the things, but I don't know why they're not showing up. Uh so we analyze the top 51 Dows. We looked at all their smart contracts, their governance contracts, uh, and basically their whole developer documentation

in order to come up with the different dimensions and design choices along which a DAO makes their decisions, right? And so we came up with this nice big list uh, which I'm happy to talk about later on. Um, but at a high level, basically a Dow is making two decisions. Okay. So the first decision is actually who actually has the right to govern and who does not. So who actually has the right to vote and participate in the governance right the second part is actually what the process of governance is. So first like that a proposal is created then like that the voting actually takes place then essentially that a proposal passes and then that it executes. All right. Um and so let's

look at the first thing. So who can actually participate in governance? So the first thing is actually owning governance tokens is not enough in order to participate in DAO governance. Okay. So for the 51 39 Dows actually required you to register uh in order to actually uh participate in governance. And this is kind of weird. Why actually require it? Well, it's basically gatekeeping. You don't want some people to actually participate in governance. Now why not? because that other person might be Binance um who don't actually own their own tokens, right? Uh they actually tried to register to vote in unis swap around three or four years ago and there was such a large community outcry that they actually dregistered themselves.

Claim it was a mistake but we don't know. Um and so in terms of registration there are two two ways generally that this is done. Verified but only two Dows really do this and then the most popular way is anonymous. Okay. So in short like with verified registration what you have is basically you have a government ID you have a wallet and basically there's some centralized party that you actually call and then essentially um you know like the centralized party says you can participate in the Dow now you can own own tokens only two Dows really use this it's Nexus Mutual and Goldfinch um and it's not very popular because there immediately introduces this centrality

right um second one is anonymous registration where essentially you have um you have somebody they have a bunch of tokens in their wallet and they basically just call a governance contract and just say okay I want to register to vote. All right. Uh now the downside of this of course is that the same entity may have like three or four different wallets right divide all their tokens across them and then say I want to register. Um, and this kind of gives you at least the thing like you know it looks decentralized but uh it it really isn't. And the second part is if this guy is malicious you really don't know how large your malicious party is,

right? Um the second way the the second dimension along which registration happens is like token holding versus token staking. So in token holding essentially all you need to do is just hold tokens in your wallet and register and you have the right to vote, right? So essentially you can just buy it on like some sort of a public market and you immediately get the right to vote. Uh and this actually has uh led to a bunch of attacks, right? So someone just goes into a DEX immediately buys a bunch of tokens and basically creates a malicious proposal, right? On the other hand, with token staking, you actually send your tokens to a smart contract and they're locked

there at least for a certain time, right? So, if you actually want to take it out, you actually have to you can't do it immediately. You actually have to wait uh a certain time, right? And the idea here is actually first if you do something malicious, then you have some economic impact, right? Because your tokens that are locked in there lose value. So it makes uh attacking uh the DAO a little more impractical. Uh the second part is actually you actually need to be monetarily vested for some time in order to actually get the right to vote, right? And uh this is actually a big problem that we have in Dows right now is that you actually you really have to decide if you own some tokens whether

you actually want to vote with them or whether you want to exercise basically like the monetary value of your tokens, right? So you generally in Dows you can actually put them in a DEX and get some value from your tokens and also have the right to vote. Most Dows don't allow that. Most Dows require you to just like keep holding on to it if you actually want to vote. If you sell it in any way, you lose completely your governance rights, right? And in this vein, we actually wanted to analyze like um you know which one of these are winning. Um, and if you actually look at how much tokens uh are out are owned by like custodians and like onchain deexs, basically like a

large percentage of them are owned by them. And this is like a lower bound on basically just looking at like a few of the top CXs. Um and actually for 44% of the Dows you have more more tokens actually in the market than are actually registered to vote. Um yeah so it seems like people are way more interested in exercising their monetary right rather than keeping it to actually uh govern. Right. Now that we've looked at you know who has the right to vote let's look at how governance power is actually exercised.

And so here if we're looking at the life cycle of a proposal, right? So first you have to actually initiate the proposal. Somebody says, "Hey, I have a great idea for a change. Let's do this." Then there's generally some sort of a voting delay to say, "Okay, let's talk about this." Uh then in the voting period, this is where the voting actually takes place. There's some things about like what kind of voting happens like whether you have to vote for something or actually actively against something. Um then there's the post voting period.

This is where the vote votes are counted essentially and you know based on the rules they're certified. Yes, if if it's passed and no if it hasn't passed and if it has passed uh there's some sort of an execution delay, right? And then the proposal finally takes effect. It's executed on chain. Um and so between the time when a proposal is initiated to when it actually executes, somebody can also veto, right? Uh that just means cancel the proposal. Uh so for the purposes of time, I'm just going to go into these two things, but happy to chat about everything later on. Um so with certification, right, like this is a

pretty important thing. I I would wager that like none of us knew before 2020 that the vice president of the United States had to certify the election. Um so like these are things that you don't really realize how important they are in the smart contract until like something actually happens, right? Um and so the first type of certification is basically decentralized, right? So here anybody and people that aren't even in the DAO can just say okay this proposal has passed right and how do they do that well they just called a function in the governance contract the governance contract basically checks the rules and then basically it says okay check mark proposal pass right uh sec and realize here you don't actually need any sort of

a trusted third party in order to certify it right uh the second one is centralized generally this is done done for dows with offchain voting. But actually this is also the model that many layer 2 Dows have like arbitum and optimism, right? Uh so there's a proposal that gets put offchain uh and basically it passes the trusted third party sees it. They say okay it's passed and then basically puts it on puts it on chain. Of course there's the there's the caveat that something might pass on chain but this party might just not decide not to actually execute it.

Um, in a lot of layer twos, they actually have the power, the the the certification authority essentially has the power to just say that any proposal that they want to have passed has just passed regardless of the voting outcomes. Okay. Um and so in order to basically you know like uh get rid of this centrality in like this off-chain voting and stuff there was uh there's another paradigm called optimistic certification that allows anybody to certify by putting down some collateral onchain right so essentially here I see something offchain I put down some collateral I wait a couple of days

if nobody objects then essentially it passes I get that collateral back Right. Um essentially now what happens if someone malicious puts it in then basically within those two per two days somebody challenges it says this is the right proposal and the decision generally is made by an impartial third party. Note here that actually you are giving up your DAO's autonomy uh by using this system because at the end of the day um this impartial third party has the ability to actually um make decisions on behalf of the DAO. It's not the DAO itself. Right? And then you know the winning one gets some cash back and

then the winning proposal is executed. Right? So if we look at veto right essentially someone proposes a proposal and you know the proposal may have a bug in it like it may be malicious right so you don't want these proposals to actually pass so somebody just like somebody can come in and basically cancel it and just note that the veto is generally more important for Dows with with ex with just onchain voting and like where the entire pipeline of the uh proposal is done onchain Right? Why?

Because everything is automated. So something that doesn't pass something malicious that passes is actually ex executed on chain. Whereas for offchain there's always this trusted third party, right? Um and the veto is actually something very necessary in any for any DAO to have like why? Because this is what a proposal actually looks like, right? like you allow people to just propose with like a certain threshold and nobody by just looking at it actually knows that there is a bug in it. This was a real proposal from uh ENS where they found the bug after the proposal has actually been uh done and everybody had to vote against it. If the

uh if the proposal had actually passed, nobody could have stopped the execution of this buggy proposal. Uh so the veto is actually something that's very much needed. Um generally the veto power is assigned to two people. So the first one is the vote initiator. So if I initiate a vote, I can cancel it myself, right? Like of course this can lead to something where like you know like it's a proposal that's required in the DAO. So, I propose it and then uh if I'm like a little malicious like or I don't like it, I can just cancel it and just make the make it longer for the proposal to actually execute later on. Um the second one is a specific veto authority. So

basically here, right, a specific wallet address has the power to veto essentially any proposal, right? Like this is generally looked as a guardrail. So if there's something malicious or something or a bug then somebody can just come in and just cancel the proposal. Uh but this is huge implications for the centralization and security of a DAO especially if you look at this recent well not that recent but uh like uh four years ago Venus protocol essentially a majority of their token holders voted on a proposal to basically take uh a lot of the DAO's treasury and there was a

vetoer that basically canceled it and so this brings into question of okay like who like like centralizing of a DAO versus basically having guardrails against it and essentially who in essence like controls the DAO. Um uh Yep. Um Yep. And so why are we doing this? Well, actually this uh our characterization of all these Dows actually brings in uh allows us to study uh like what can go wrong in case of like attacks. We actually map out like the different security implications of each of the design choices that you make. Right? So I'm just going to give you a quick

example of basically compound which was exploited for like 25 million. In the end, they the the exploer decided to cancel the proposal. But um I'll just show you quickly how actually this happened. So essentially they exploited three critical vulnerabilities that we've gone through today. The first one is like anonymous registration. So as we looked at before, right? Like nobody knew how large the party was because essentially they just had it divided across like 28 different wallets. Like uh yeah uh six around the true size of the malicious party was 680,000 which is more than twice what their best estimate was. Um the second was voter apathy and vote

sniping. So they do all their voting onchain. Onchain voting has a cost. Essentially, people just thought the proposal would fail and at the end of the day, like right before voting ended, basically 520 82k votes were cast and essentially the vote passed. Just to note that like there are Dows that have um that there are Dows that actually have um guardrails against this written into their smart contracts. Uh essentially like Lido and Arbitum. Essentially what they do is they only allow you to vote against a proposal on the last day. essentially making sure that such proposals don't pass. And the third one is a lack of oversight. There was no veto authority and surprise surprise in response that's what they

did. They basically added a vetoer. Um I think it has to be uh elected every six months or something. But um essentially that's what they did. So in conclusion, we classify Dows around very security dimensions and essentially we realize that every time that you have questions about like security or usability or efficiency of a DAO, like the direction that they lead to is essentially like centralization, right? Like take an example of delegation, right? Delegation sounds like a great idea and it is a great idea um because it increases the amount of votes that you actually cast. But it invariably actually centralizes the DAO

because you're actually giving power to a smaller number of people, right? And so now the other the the other thing that like our work really helps with is now you can actually answer some of these questions very well. So if you're like a lawmaker and you actually want to like create some sort of a DAO law like you can actually make it clear like what exactly you want uh the power to do, right? like who gets to update it, who gets to modify it, or I if you want a veto power, maybe you want a certification power, right? Um maybe the key should be with the judge, maybe some law enforcement, but I think it actually allows you to uh make these decision,

make these decisions um as a lawmaker a lot more cleanly. Yep. Uh thank you very much uh for uh listening. Happy to take any questions.