DAO Governance failures and ways of mitigation

Transcript

[music] [music] Nice. >> Nice. So, I had this stupid slide here, but I believe that it's completely useless, right? because neither of you have a lot of stake or uh tokens, governance tokens, or neither of you participates in identity based dows because they simply don't exist almost yet. And uh I'm going to give you some crazy numbers. So I've got the data from almost everywhere like snapshot, tally, aragon, dows, dow house, snapshot x, salana dows as well. hundreds of

thousands of proposals uh LLM's let me read over two,000 proposal texts and uh this number is not only for the small Dows it's coming up only to 6% even in the biggest Dows so imagine you have uh a lot of tokens in circulation and only 4% actually votes okay let's uh find out what's what's happening here What actually is governance? I think this is a very encompassing huge word and sometimes we mean governance just proposals, votes and upgrades, but it's a small part of it, right? Uh Dows are

also bargaining power like how much power you have in order to turn things right or left. Governance is also your exit. For example, if you run an Ethereum node and uh you got your client doing weird stuff, you can switch to another client or you can actually exit and divide into two Dows if you don't like what the protocol doing with its treasury etc. I'll show you later. But Dows are also in governance are social fabric. it's code, norms, culture, people if they exist uh in the DAO. And I want to tell everybody here and

especially those who don't know much about what happens in our beautiful industry, we have this term immutability, right? But we never have 100% immutability. When we talk about immutability is just things that are really hard to change that require a lot of coordination. But anything can be changed. Bitcoin can be changed. Ethereum can be changed. Anything there is immutability. 100% immutability does not exist. 100% security does not exist at all. But first and foremost, the most important thing in governance which most of the protocols completely ignore and you'll see that later is coordination. is how community converges on one particular

decision. So we found out some pretty nasty numbers. Uh in the last year we have about 2,000 active Dows out of uh 100,000 created in the last 6 years alive. And as you can see like one to three addresses they propose eight in 10 things in the DAO. Okay. So it's mostly team proposing and in snapshot as you can see it's most of the half of them you can't even propose an outsider. Okay. So again, the team decides everything and uh as you know, Gary Gensler left us recently and SEC is not trying to tell you, oh,

you're not the centralized protocol. We're going to we're going to get all money from you because uh well, you're basically decided all yourself and your security. But uh after Gary Gensler left us, well, most of the new projects just don't have DAO at all because they don't need it because Dows were mostly a facade to evade legal action by US government. Okay. Well, probably it's obvious already at this point for everyone and uh it's great when DAO doesn't exist. If you really have immutable application, immutable smart contracts deployed on good secure chain for it's for example like when you have version one and then

you develop version two but version one still exists forever on chain you can withdraw money anytime but the contract is just not upgraded. Okay, in this case maybe you don't need it at all but in most cases you really need it. Uh let's look at the voting power. I think it's all obvious as well. Voting power mostly comes from the top 10% voters because all of the Dows are what they're token weighted Dows and money is not decentralized at all. And this is what happens. Uh surprisingly we thought that oh you have this rich VCs and funds investing early into projects and then deciding the fate of

these projects. Well, I found out in personal conversations with funds and VCs that this is not true. They just don't participate at all. And there are three reasons. Number one, legal advice especially for the biggest US funds. So there was that case when uh Samuels I don't remember who he is some guy he went against Lido and uh in this case those funds who were participating in governance they were partners of Lida and then they almost became liable and uh after that lawyers say to funds that just don't vote okay you'll be on safe side. Uh the second problem is that uh they hold their funds in custody and

custody providers don't care to integrate uh any kind of voting features because they don't add value and yeah of course there is no incentive if there is incentive to participate in governance for them. Well it's uh basically not enough in the end funds are just here to make more money and there are fixes to those and they fall into four buckets. Number one, guard rails. This is something that uh stops Dows from mistakes. Basically, prolonging decision- making. Uh next are markets such as uh meta governance as I call it when you have uh DAO voting in another DAO. And we have futur now which

is very interesting and helpful concept. Decision quality is something that we have after vote and before vote which is very important because vote itself should be really seamless because well the decision is already taken by the people before vote right and identity which is completely missing in our DOS. Well, uh this is what happened mostly this year and these are very interesting cases where the governance went wrong. Uh at the first one token of power very interesting someone bought the majority of tokens and proposed voted executed in one transaction and that's it and money

is gone. Uh second one is bongda. What happened here is seven wallets voted for moving $20 million out of treasury and it's gone. Barnbridge is a dead uh project and uh the debt project with the DAO. So attackers just spent uh a little bit of money. It was like less than a thousand bucks in order to have a vote and get the rest of the treasury out of the barbridge and uh term finance going to tell you later. Neutron well simply uh tokens bought minutes before the vote voted money gun no guardrails crazy so

term finance they had a guardrail but it was the guard rail that could be changed so the attacker bought less than thousand bucks 90% of the voting sub supply because nobody votes uh 60day veto nobody voted why because in their case you have to turn your token, lock your token for the vote and there was a snapshot before that vote. So nobody could actually come and lock their tokens for this vote. So no vetos and this proposal set cool down to zero and uh 8 million drained.

So seven kinds of guard rails for the DAO. Most of the token weighted dows should have ideally all of them which is uh voting delay past block snapshot a council with a veto. This is uh a little bit centralized but still saves us sometimes. Uh some rules should not in in the smart contract of the DAO should not be changed through a new proposal like in term finance. Of course, uh those that should be monitored and uh they should have disputes by let's say ClariS sometimes you can exit which can be good which can be bad caps thresholds and of course if your project is going down please uh

shut down the governance and uh distribute the rest of the treasury left. So what happens with the uh with the council votes? Okay. So oh no that's not the council. So term finance right I already told you about it. Uh nouns this is where the community split in two and for good or for bad the treasury also split in almost two but at least uh it continued existence as two different dows. And as for the security councils, drift was a very bad example because well, you have a security council as a savior of last resort. But what happened is that uh two out of five

signers were hacked through basically social attack like North Korean, not North Korean hackers doesn't matter, but they lost almost 300 million bucks. And uh Arbitrum, this is again pretty centralized, but Arbitrum froze uh 71 mil before this year and now they also froze some during Binance hack. So it it worked. Uh meta governance. What happens here in curve? Uh you have V tokens which you lock and if you lock them you get more voting power through the time. But uh well people don't want to have their capital locked. That's why we have uh meta protocols on top of these dows and eventually all the voting power went to

these protocols like convex st and yarn and uh yeah basically all these dows metadows decide what's happening in vote and most of why these dows decide what they decide because they get bribed okay well what could y expect from token weighted dows uh footer key is very nice uh we finally have some data on during the last years because of the metadawa they had 150 markets and it works like that. So you go into the market like prediction market and there are two markets. So, and if you vote uh you you bet money on yes and you bet money on no. But uh the

thing is that uh you vote for yes proposal if you think that the price will go up if you accept this proposal and you vote on no if you think that well if we reject this proposal the price will be higher than if we accept it. So basically in this market actually it is not as attacker in your interest you can lose money if you will be going against the well rational decision right but imagine if you have a treasury of 10 million bucks or a million bucks and the futuri market is just uh 25k traded well you can actually lose this 25k but uh anyway drain the treasury in the

So decision quality is very important thing that most of the Dows don't do and uh decision quality is when we discuss the proposals when we compare alternatives and when we also look at oh this is what we chose and these are the consequences and what are we going to do later and almost zero Dows in uh I mean on top of Ethereum do that but Ethereum the one which has no DAO does this the best because uh Ethereum has no vote. Ethereum has no onchain governance but they have public APS core dev calls and a lots lots of discussion like these discussions we have here and real coordination it doesn't take just vote

and voting mechanics it takes uh that anyone can propose and argue. You should have several options several alternative not just uh yes or no. Everybody sees transparently what's happening, gets feedback and uh right to exit. And unfortunately, especially recent years, most of the protocols uh have this thing called optimistic governance, which is a very pessimistic picture because in optimistic governance, it means that the team proposes something and unless everyone objects, so everyone acts and acting costs something for people. Well then proposes go through and in most of the time it goes through it's uh completely dysfunctional in DA terms but very

functional in terms of business because teams can push what they need very fast. Uh [snorts] so we still for some reason don't have Dowos with identity. It's crazy. All of you heard about quadratic voting right? Zero Dows with quadratic voting. Yeah. And then we have 12 plus proof of personhood protocols. Most of them very much care about privacy. Most of them completely non KYC. Yeah. Like just uniqueness checks, private biometric uniqueness checks. We already have a lot of humans going through these. I mean, we're pretty much used already to using biometrics.

And it's great especially if they're private, verifiably private and nonqyc. And we have a lot of attestations. So attestations let you earn reputation, let you earn let's say additional points to prove you're not just unique human but you're unique uh human with developer experience and well giving some traits to it and uh snapshot they have these primitives integrated zero DOS use them completely zero and what was very funny there was a paper from circle and they said oh impossibility theorem of quadratic voting because you can create as many addresses as possible. Well, brava, you can just use

identity and and it will be all right. So, in identity system, it needs uh unique human verification at the bottom of it, reputation from contribution or attestation from outside data sources. votes, you can't uh prove to a buyer, meaning that if somebody tries to bribe you and usually it's uh very well sometimes it's it's cheap to bribe people. It costs like 2530 bucks to bribe people. But you can have a technology that either you cannot prove to the briber uh how you voted or you can prove him that you voted for him and then you just redirected your decision

to another person and he still uh pays you money but you voted for another one and delegation minutes limits in order to evade the centralization of voting power. We actually had an experiment by optimism in uh identity in one person one vote. These people were hand selected by optimism but still these were people that you see here as speakers and in 2022 there was citizen house. it was at least responsible for the distribution of airdrop how it's going to be handled but eventually we found out that optimism is not interested in decentralization of governance and uh citizens house was just like a guard

rail to object against but you imagine how hard is it to coordinate the 100 people in order to object on something and in the recent vote token house just say oh the airdrop distribution will also be decided by us so the citizen house. It technically doesn't do anything right now. And uh for some reason, we are pretty much the only ones building Dows built on people, not money, where people can actually have a voice, can contribute. And uh how do we do that? We do facial recognition, private, of course, in order to figure out if you unique. Uh but in order to have a vote, we don't need tokens from you. We need you to

contribute because well, we want not the stupidity of the crowd but the wisdom of the crowd. So we have to choose carefully who actually has a vote in the DAO. So once you contribute and you get your first milestone accepted, so do something actually prove you can do it. Then you get voting power. It's really hard. It requires a lot of coordination. But coordination is what governance is about as Ethereum shows us being well one of the best Dows without having onchain uh governance. Okay. And uh we had other experiments like proof of humanity DAO. They had vouching and video checks of people but uh unfortunately they are not that active as a DAO right now. though the protocol

still works as civil resistance mechanism and uh IDA as well as a Dow not active at all. Polka dot will try hybrid model but it's like uh rich people choosing their elected people and uh as you know elected people are eventually becoming corrupted and what I really vouch for is direct democracy liquid democracy and we don't have them in crypto but we have them in real world probably people living in Barcelona or other cities know about the existence of things like participatory budgeting and it has crazy scale over 30 million euros distributed on

average every year in Barcelona followed by people proposing stuff other people voting on it and budget is really being distributed and things are really being done and it works already across 36 countries and well I was I was very surprised to find it out. So if uh people can do it locally, why can't we do it in the internet, you know, borderless? So my advice of what to do if you really want decentralization, not just uh power and money from cryptocurrencies and uh blockchains and protocols. Well,

create guard rails everywhere. Otherwise uh you will be drained. Then uh pay a lot of attention, invest a lot of time and participation. It's very hard to get outside participants. But if that's only one team deciding everything, then what's the reason for it? It's not decentralized at all. And then please use identity primitives. And if you are already like token weighted DAO then at least use it for community signaling like one person one vote to actually see uh what people think not just uh your team thinks and yeah we

need more experiments but uh you know our institutions that we live in like our governments I mean real ones that we live in that we're citizens of are really old. They haven't changed almost in hundred of years. Now we finally have tools to experiment a lot. And once we experiment at least on a small scale, maybe we'll pave the way to them to our dear lords to change the systems as well because well they believe democracy doesn't work. They believe most of the people are stupid. Well, how about we prove them that we can coordinate together and uh we can live in direct democracy. So, thank you very much.

That's that's it. I hope I have some questions from you. >> You do. [laughter] >> So, we'll start at the top and thank you very much for that. We'll start with the top one. With so many attacks, we need to go proactive on white hat recovery. How can you defend yourself legally to take action that way >> legally? How can you proce yourself legally against people who don't care about legal? Well, you cannot. So just well having a lot of white hats it's great but some people are ready to do crime and uh with AI we got more and more of those people because before that they

didn't have the technology and knowledge to do crime but now they have so you cannot protect yourself legally just you have to write nice code you have to have guard rails and uh that's pretty much it be ready for a hack any day. >> Perfect. That's great advice. And then obviously you you've given some negative view. You you you highlighted lots of the failures of Dows. Lots of people use that to critique the whole model of Dows and decentralized governance. What would you say to critics in defense of Dows?

Well, we never experimented with Dows much. As you can see, Dows as I showed you Dows are facade was legal facade to look decentralized. In fact, teams decided everything. outside contributors never contributed most of the proposals done by the team so yeah it it gives transparency but I would say that we didn't have DAO experiments to say that Dows failed okay that's it >> understood um and then the question here isn't identity an increase in liability >> that's a great question which I don't have answer to well well you sufficiently decentralized and you have like 1,000 people participating all over

the world. Who is liable? Well, in this case, nobody is liable. Even uh with the recent guidance of US administration, because you're all like equal participants, I believe nobody is other than maybe the deployers of the protocol. >> Understood. Thank you. And then I like this one working for a market maker. Uh how how can a DAO protect itself uh when a token price is crashing over time? protect itself. [snorts] Well, token prices are markets. I believe all the DAO can do to protect itself is to do more work. That that's it. Write better code, do more work. Good point.

Brilliant. And then and then a follow-up to our earlier question. If you can't self your if you can't defend yourself legally, then then you can't white hat. Uh more of a point than a question. So, we'll leave that one. Um >> yeah, I agree. I I know many people who were about to if you can't white hat just uh use private communication channels with the team and uh of course you get less rewards uh rather than when you white hat and uh if you whitehead just please uh your tokens should stay where you just don't move them to Monero. Okay. And uh you will you will be most likely all right but usually you just uh send the bug report or possible exploit report to

the team if you are really scared. >> Perfect. And one last question just to finish off. How does the end state of governance look different than governance for investors in public markets? >> Oh that's easy. In public markets uh you have uh again one dollar one vote one stock one one vote and uh this is not you see long ago in the history not everybody had a vote only rich had a vote only that those had who had lent had a vote but uh right now we kind of have some versions of democracies even with elected corrupted members but now we have tech to do direct democracy so the end state will be indeed different,

but we don't know how it will be. Will it be one person, one vote? Will it be something else? We'll see. But >> [music]