The Evolution of Decentralized Governance with Andrew Hall ​​| a16z crypto research talks

Transcript

this morning we're really lucky to have andy hall from stanford both from the business school and the political science uh and some some combination of the two uh and he's gonna talk about the evolution of decentralized governance cool thank you get started um please feel free to interrupt with questions at any time that's kind of what i'm used to let me i'm going to start with just a brief historical story to motivate this and then kind of get into the details with you and please do stop me at any time so um what i'm showing you up on the screen here is the story of the glorious revolution if you're from the uk you probably know more about this story than me um but

really what this story is about and why i think it's relevant to web 3 and decentralized governance is this is a story about parliament in the uk uh taking power away from the monarchy and distributing it a little bit um to a bunch of rich people in in parliament and it was important because at the time of this thing called the glorious revolution there had been these increasing challenges where the crown was constantly trying to raise money mostly to fight foolish foreign wars and over the previous centuries it had gotten harder and harder to do this so if you learn about for example queen elizabeth's very long reign a lot of her reign was occupied with trying to figure

out how to raise money and that involved taxing and also taking loans uh from nobles throughout the country and over time it had gotten harder and harder to do this and by the time of james the second's reign the crown had been forced to get very creative and basically had started to massively expropriate wealth in the country and so they would say they wanted to take a loan you wouldn't want to give the loan because you knew it was never going to get paid back and they would tell you well you're required by law to give me this loan it's still a loan but you don't have a choice and the interest rate is going to be basically zero and oh by the way we're never going to pay it back uh by the time of james ii this problem had become quite severe and james ii had essentially declared

that he had the absolute authority to change any rules at any time that he wanted and so if you didn't want to lend him money too bad he's going to take it from you anyway and the glorious revolution which had many complicated other causes had this consequence that parliament basically conspired to get rid of james william and mary were installed as the new monarchs and as part of their installation they somewhat informally agreed to this declaration of rights and the declaration of rights included a list of grievances about what james ii had been doing and the ways he'd been usurping powers that were rightly held by parliament and it also listed out the powers that parliament felt were rightly its and not the monarchs and william and

mary agreed to these rules and that ushered in what's called the era of parliamentary supremacy where power that had been centrally held by james the second was distributed out to still a relatively small number of people just to be super clear but to a larger group of people across society that represented the largest wealth holders in the country and somewhat as an almost accidental consequence of this revolution the government was suddenly able to raise much more money than it had been before because now people who had previously worried well if i get involved with the crown i'm just going to get completely expropriated now they felt we actually have this sort of constitution

that guarantees me to some extent that i can actually do business with the government and i won't be hurt later because there's now a set of institutions that were laid out in this declaration of rights to make sure that my property rights were secured and that gave me a process of governance to be confident that if i had a problem if i felt the government was doing something it shouldn't with my wealth i had a recourse a means of recourse to correct the problem and this led to a very large increase in economic growth in the uk in england and a sense that government was becoming in some ways more legitimate um and this is a story that's told i think quite interestingly in a really famous paper by north and winegas today

um that i recommend so why am i telling you that story uh the reason i'm telling you that story is because i think that these fundamental problems related to governance and related to holding too much centralized power has a lot to tell us about uh web 3 governance as well which is something that i'm super interested in so you could think of these very large web 2 platforms that we all know so well as kind of like monarchies with pretty weak parliaments that don't do nothing i want to be clear but that don't have a lot of recourse to prevent the same kind of decisions being made by platforms that were being made by james the second so you think about these

disputes for example between epic games and apple they're very much in this vein it's sort of like apple started to build an app store people really wanted to be involved then the app store got really really big kind of in the same way that james the second got a lot of power and now apple's like by the way we're gonna charge you a ton of money and there's nothing you can do about it you have to follow all of our rules no matter what and i think that that's it's very very similar um to the state of affairs at the time of the glorious revolution there are lots of use cases where that might make sense for platforms obviously it might be more efficient it might just be easier but the glorious revolution i think tells us something about what might happen if platforms actually figured out ways to

distribute power a little bit more and the two things that i think are most important that i think about in my work which i'm going to tell you about the first is this sense of social legitimacy that if a platform is making decisions that affect all of society like for example what we're allowed to say online that it might be more legitimate for those rules not to be made by a single person or a single company but rather by a larger group of people and second more in line with the glorious revolution itself this idea of economic growth that there might be people on the margin at least who aren't invest investing as much time or effort in becoming a creator or a developer or a seller on a platform because they're worried about being expropriated in

precisely the way that the crown expropriated people prior to the glorious revolution and so i think there are these reasons to care about whether we can implement effective decentralized governance at scale to solve problems like these two and potentially um therefore to sort of increase the level of social legitimacy we perceive in our in these platforms uh and to encourage a bunch of new growth and innovation and just to give kind of so that this talk is going to be about how can we do that and what is political economy which is the world i come from have to say about that and just to give the bottom line up front the thing i'm going to try to kind of repeatedly convince you of is that everything we know about how governance has evolved

over time in society suggests to us that it's going to have to however we build these governance systems it's going to have to involve a lot of indirect democracy this idea that we can have everyone in a protocol on a platform whatever vote on everything i think is is not going to work and that's what i'm going to talk about today before i get into that let me just super quickly explain who i am i'm super excited to visit here and to get to know you all and kind of like how i came to work in this area so by day i'm a political scientist a political economist um i do mostly data-driven research on how to build democratic systems so just to give you like a few examples like

recently my lab is studied so what happens when you let people vote by mail how much does that affect participation actually turns out it has very little effect on participation um we go out and we measure the ideological positions of all the state legislators in the country all the candidates running for state legislature to try to understand you know where is polarization coming from uh what is causing it among our politicians done some work linking property records to voter turnout data to understand these kind of like differential participation of homeowners in local politics just giving these examples because they're all about these same governance problems that turn out to appear in in web 3 all the time so like you know polarization conflict

disproportionate representation of certain participants over others these are all sort of like classic problems and part of how i became interested in the space is that on the side i've also been working in tech for the last five years on these types of governance problems and large platforms so primarily i've done quite a bit of work with meta the company formerly known as facebook which obviously has pretty severe governance problems uh and i've i've been really interested in those problems and in the fact that they actually overlap with these governance problems from society the way i got into web 3 and the way i ended up in this room giving this talk is that through my work in tech

i became really interested and i was asked to do a bunch of work to understand how groups online are self-governing and i was googling around one night and i stumbled on this post i don't know if any of you know this guy zerox dennis he's at maker dao he's brilliant he wrote a super long post in the make it up form about the problem of getting people to participate uh in protocol governance and and the post is super thoughtful it's super clearly written encourage you all to check it out but what blew my mind was that the problems he laid out were just completely isomorphic to the problems we study in political science about countries or cities or states running elections problems that you

can't get people to participate that it's too expensive for them to participate that they don't know what's going on that it's a pain to learn about the issues these are classic political science questions and i became super interested in whether um we could improve decentralized governance by taking some of the things we learned from you know 2 000 years of studying democracy and apply them in this space and so that then led me i met porter smith here at a16z we just wrote this piece for a16z on this idea of what does the history of governance tell us about how to implement governance for web3 and a lot of what i'm about to tell you about comes from that piece

okay so let me get into the the meat of the talk for today um [Music] when you look out at governance in the web 3 space if you you know talk to people who are working on governance i think there's sort of three there's obviously lots of different issues out there but there's three deeply interrelated issues that come up all the time the number one issue that i hear all the time is like we can't get people to participate and then that's closely related and in some sense causes and is also caused by these other two problems which is that when no one's participating small groups with concentrated interest can sort of like threaten to hijack the process in various ways and at the same time if you don't have enough people participating

you're also worried that the decentralized workforce or the bureaucracy as we would call it in politics might go off and do things that you as the token holder or the voter don't agree with and because no one's participating no one's monitoring that and even if they're actually behaving fairly well you might just there may not be trust because you're not actually around monitoring and that leads to suspicion and so if you can't solve these three interrelated problems which every democratic society in the history of the world has has stumbled upon then you're not going to create that sense of social legitimacy that this was a legitimate process that led to a legitimate outcome and you might also have these problems of growth

because this isn't a well-governed system it's not a system that you can trust of course there's lots of other problems too like governance attacks and the like that are exacerbated by these issues societies across history have found again and again and again that you simply can't govern through direct democracy at any kind of meaningful scale and the reason for this is kind of obvious the reason is that it's totally irrational for any individual person to spend substantial amounts of time monitoring and learning about each vote if they're very unlikely to be pivotal in the outcome of that vote and so even if you only have say 40 people voting it's already starting to be pretty unlikely that your vote's going to make the difference and so why on earth would

you spend hours each day studying these abstruse issues and trying to vote on them so this is often in political economy called the paradox of voting down's talked about rational inattention it's a classic problem in all voting systems and it's really cropped up quite severely in web 3 because there's been a pretty heavy reliance on this ethos of direct democracy which i think is kind of a mistake and part of why i think it's kind of a mistake is people want to say that if you're using direct democracy it's more decentralized but it's not if people aren't actually participating and once you have that kind of voter apathy we tend to know what happens next and we're seeing this more and more i

think now in web 3 is pretty small groups of people who can benefit from getting certain things passed are going to start pushing to pass things knowing that many people may not be paying attention and we've seen examples of this like there's an example maker where a small group of people tried to get something passed it didn't pass so they just tried repeatedly and when you have this sort of like weakness that comes from very few other people participating you become vulnerable to those kinds of problems and as i mentioned before it also makes it very hard to perform oversight if no one is participating then who's around to assess whether the decisions that are being made or the actions that have been taken on a day-to-day basis outside the voting

context are good or not and so this is why i think it's really important to explore how we can use indirect democracy to make these systems work and this again i think completely parallels history if you look at the history of how organizations govern themselves you see very early on long ago in history a pretty rapid iteration past direct democracy so this is sort of a picture of the early days of the british east india company the east india company had an elected board of directors it was not direct democracy and it came out of a history there had been a history of smaller companies with shareholders

voting on things directly but it was five or ten people and so it was kind of like you could all get in a room and hash things out what started to happen is more people wanted to invest in these things and as these enterprises became larger there's almost instantly this problem where some of the shareholders who were voting on stuff were also the managers of the company and knew everything about what was going on and some of them had other jobs and this was a side venture for them and they weren't around as much and they became very worried that because they weren't around they were getting screwed over and so very very quickly they said this makes no sense i can't show up every time and vote on everything but i don't want to be expected to and not show up and then get

screwed over so i would like there to be a representative body and i'd like to be able to vote for someone to be there on a day-to-day basis when i can't be and make sure that my interests are represented and that's a super important evolution in how corporate governance worked and of course we know the same thing has happened in societies all across the world not just in the us though of course here in the u.s we tend to learn about it for the u.s context is a picture of james madison you know they they put a bunch of work into thinking about how you could go create a representative system where there'd be linkages between the people voting and the people who are representing them and we're starting to see this

uh in web 3 and i think it's i think it's a really important evolution that i want to talk about um please you're talking about history and direct democracy failure in history but yeah presumably it was technologically just more more infusible to run a direct democracy historically yeah yeah the costs of doing this so for sure for sure so if you thought that the main reason direct democracy hasn't worked in the past is because it was too difficult to perform the act of voting then i think you're totally right like technology should facilitate a massively quicker and easier way for everyone to vote the problem whenever we've studied

this and actually kind of comes up in that paper i always mentioned you that i did on vote by mail is that whenever we study within the plausible range of ways you'd run an election what gets people to turn out more or less it has way more to do with the cognitive constraints of figuring out all these things and this sense that a lot of people have that like i don't want to have to deal with this it shouldn't be my job and i have other stuff i want to do and much less about whether i have to go somewhere to vote or whether i can vote you know on my computer leaving aside obviously security issues those cognitive barriers are i think much larger and so and this is i think and i'll talk a little bit about this i

think one of the interesting fallacies that i see all the time in american politics which is playing out again in web 3 is this idea that like well we can put out all this information because everything's transparent and so you're going to know everything about what the delegates are up to or you're going to know everything about what we're voting on and so there's no problem and in the same way i mean basically every year an undergrad comes to my office and says i'm going to fix democracy by making an app that tells you what your representative is up to and the reason those things have very little impact is because my problem isn't access to data it's time and expertise to analyze the data and figure out what's going on what i should do about it so i mean i've hung out quite a bit in maker

i don't hold any mkr anything like that but i hang out there sometimes the stuff they vote is insanely abstrused there's like these flow charts with the cayman islands in them and stuff and it's like i have no i don't even know where to start uh and so whether i can push a button to vote or whether i have to link my wallet and do all this other work is not the key it is a constraint but it is not the binding constraint i think um yeah um let me ask another kind of devil's advocacy like within the us context or maybe within like western democracies in general is the trend toward more direct democracy or less because i would have said it was like we're on a slow trajectory

yeah yeah i know i think that's 100 right we are on that trajectory and i think that there are there's two very different parts to that trajectory one of which i think is unequivocally a good idea and one of which might be a terrible idea although the jury's out the part that's unequivocally a good idea is making sure everyone's allowed to vote and can vote because that's critical to the functioning of a democratic system that's not really about direct democracy but that's a really important part of this democratic trend it's of course in the u.s questionable today whether exactly what it means to be a democracy or whether the u.s is one but it's very obvious that it certainly wasn't one before the voting rights act and so

that's a trajectory towards more democracy but not necessarily towards more direct democracy well the other trend we've seen which you might be getting at is sort of this idea that we should all vote to nominate candidates for office which didn't used to be the case or used to be that parties could kind of hash things out behind closed doors there's some trend towards more ballot initiatives or referendums which are another form of that direct democracy and there's definitely this rhetoric around that that's like a good thing um most of that trend is actually pretty old so like the development of those initiatives and referendums is actually earlier in time but now we're kind of stuck with them like where i live in

california where we vote on 100 of them each election i think most people who've studied them from different ways tend to come out on the conclusion that it's generally not a good idea to have a lot of stuff like that there's a huge trade-off though if you don't have referendums and you don't have initiatives you worry about the process being captured by special interests or by politicians who maybe are missing things that are important in society on the other hand when you start to open things up to these direct votes you worry that you make you it's it's less biased in the sense that like more issues can be considered that special interests might otherwise quash or something uh but it's you might make really bad uninformed decisions and

paradoxically and it's exactly like the problems i'm thinking about in web 3 you can actually accidentally empower special interests more because no one knows what's going on and so this is what you see in california if you look at the things that are proposed in these initiatives they're impossible to parse it's like should a doctor be required to be present for dialysis at private dialysis clinics i have no idea i don't know the first thing about that it turns out that that whole initiative which has been proposed on like five years in a row now is a battle between two different special interest groups that has you know nothing to do with the things most people are worried about um and so i think

i think most people who study these systems tend to conclude it's a bad idea to have too much of that but it's also a bad idea to have none of it but you know you don't have to look far to come up with examples where that kind of direct voting seems like it leads to pathological outcomes brexit perhaps being one obviously some people are into brexit um but it's actually like not a very popular policy today and yet got passed okay so i want to talk about with that motivation in mind what are some concrete things we could do from where we are today to change decentralized governance in interesting ways the first has to do with these systems

of delegation which are which are already being developed and some dows and taking the next step and and making them into like robust systems of delegation i'm going to talk about how we might be able to do that by giving stronger incentives to people to become delegates the second is building out more complex systems that guarantee representation to different groups and so like every democracy in the world does this in some way we're starting to see you know like there's some examples of it now in web 3 that are interesting so talk a little bit about that as a second idea and the third is is building up these systems of oversight that i think are essential to creating trust between the the workforce and the token holders so these are the three things i want to

focus on for the rest of the talk um okay so the first one i want to talk about is like how we can improve delegation as it exists today so as i said there are some dows uh using delegation by which i mean you can take your tokens and you can delegate them to another person and say hey i'd like you to vote on my behalf so you get my voting power this is an idea by the way that's been around forever and it's called proxy voting in corporate governance um uh though it can be implemented in technical ways in one three that are different and super super interesting um and and what i think is important to recognize but not all not necessarily

super obvious when you look at delegation in in web in the web three space is that it is like representative democracy in the sense that the person you delegate to it's similar in a way to if you voted for someone to be your representative you are empowering that person to take decisions on your behalf or to take votes on your behalf and so that opens up a really interesting toolkit that we don't have in direct democracy where we think about how can you make that relationship work well how can you as the delegator have confidence that the person you delegate to is gonna do the kinds of things you want them to do and that's called the accountability problem

in the city of democracy and it's sort of like a super classic age-old thing we've studied a lot and so we have a lot of ideas about how you might make it work that i want to tell you about and i'll just say the history of how that developed looks pretty similar to how how a delegation is developing so far right now we're in this really early phase there aren't a lot of delegates it's kind of hard to convince people to want to do it most of the money or the tokens being delegated are coming from a relatively small source of people um it's not dissimilar from how representation arose in democratic systems and what we saw in democratic systems is this evolution towards professionalized legislatures which is i don't think in

any means entirely a good thing but i think it's in some ways inevitable and necessary and i want to kind of talk about how by making the delegation process more professionalized to use this political science term you might get that stronger sense of accountability that you need if you're going to convince people that indirect democracy can work for web3 so this is a figure from porter in my piece that i think tries to explain why this is important the idea is um if you can create a system where your representatives the people you delegate to perceive that they have strong incentives to do the hard work and like figure out what the right

decision is make the right proposal make the right decisions then that induces token holders to trust the system more and to believe in it more and that gives them incentives to want to be involved and to want to be engaged which strengthens the incentives for the delegates to do a good job because now they're thinking ah if i don't do a good job these token owners are really paying attention they'll fire me they'll delegate someone else and so when you can create that virtuous cycle you can get a very well-governed system and it relies crucially on this idea that i as the person you delegated to that i like care about that that i want you to keep delegating to me and that i

don't want to get fired from that job that's totally critical to the accountability problem and where we see accountability fall apart in in the real world which happens all the time it's from some combination of either i as the person running profits don't actually care about winning because it's it's just something i'm doing for whatever reason or i am secure in the knowledge that whatever i do you're not going to be able to fire me maybe i know that no one turns out to vote so it doesn't matter so you have to create this virtuous cycle where i want to win re-election and i know that if i don't do a good job i won't and that's how you create accountability and we're not really there yet with the delegates i don't think although i do think that people

at least that i've seen acting as delegates are working very hard but as a system and as a thing other people can trust to work we haven't gotten to that point yet and i think there are some concrete things we could do to try to get to that point and so the very first is i think it's essential for the reasons i just laid out you've got to make being a delegate a job people want that sounds kind of obvious but it's really important and that could be some combination of course compensation is part of that and there's a huge literature on like how and when you should pay politicians with this exact problem in mind you know if you pay them too much maybe you get sort of corrupt individuals wanting to get the job and and cheating in the election and stuff if you pay them too little though you

don't get people who necessarily care and you don't give them that strong incentive to not want to get fired in the future um but there's other things beyond compensation that might be more important so obviously one is just prestige this idea that like if i hold this job it's gonna be good for my future it's something that's socially valuable it's something that's professionally valuable it's also i think about giving them a clear job to do so i think one of the really interesting challenges i've seen in maker which is i think the very first dow to really build a delegation system is that it's been a little bit hard to figure out what the delegates are supposed to do and there's at least some people whose view is well the delegate should really be

taking over in the sense of articulating a vision for maker and and putting in proposals based on that vision there's other people who think like no no the delegates are just one part of the system and they're basically just there to vote on proposals that other people make and i think figuring out exactly what their job is is essential to making it something that more people are going to want to do and of course the last part of this that's important is right i think there's quite a few people who don't want to be delegates because they worry that that carries with it some kind of liability and there's been some work on you know insuring them against that liability which i think is is essential so that's one one chunk of things i think is interesting to think about that we can pull in from

governance and other in other places and times is this idea that like if you're going to use indirect democracy if you buy my argument that direct democracy is not going to work at scale you've got to get representatives in once you have representatives in though it's only going to work if you create this virtuous cycle where they want to be representatives they want to keep their jobs and you're paying attention and you're going to fire them uh if they don't do their jobs so that's that's the first term second chunk relates really closely to that there's this idea of liquid democracy that i think is incredibly interesting in web3 and so i i don't want to assume everyone has necessarily seen it but but this idea that was sort of like you delegate to people on certain decisions in some forms of it you could delegate to

different people on different decisions but the idea is that whether it's to multiple people or to one or whatever the whole idea is it's continuous so it's like i give you some money i can take it back if i don't like the way you're voting um and that seems very appealing it's cool first of all because it's something we couldn't really implement in the real world before now so that's cool um and it seems like it has this appealing aspect that it it creates this very sharp form of accountability that i can fire you at any time fire might not be exactly the right word but you know what i mean in the case of elements you could actually fire them because if they go below some threshold they they fall out of the council um but you gonna on a day-to-day basis change

your mind whereas as a voter you're stuck waiting until the next election right and if it's a u.s senate election you might have to wait five years and there's a sense that you therefore have very little control as a voter over your senator and so liquid democracy sounds very appealing in this regard it comes though there's i think there's a trade-off that hasn't necessarily been that appreciated which is that if i'm a delegate and you've given me my tokens and maybe i get paid based on how many tokens are delegated to me or whatever um you want me to kind of like invest in studying things and making proposals and that's like a long run commitment and if i am essentially up for election every day for the rest of time because you can undelegate for me at any time

that might actually erode my incentives to to make long-term investments because i'm being elected literally every day and so you you see this trade-off very sharply in the real world and there's lots of actually good theoretical work on how you balance the frequency of elections with these two forces in mind because we definitely have a sense that if the elections are too spaced apart you have too little control but we also have a sense that when the elections are super close in time to one another it creates very perverse incentives for representatives and if you look at like congress this is like a famous line of argument that congressional elections are too frequent members of house are always running for election that changes the way they behave it makes them pander more it makes them invest in longer term

more serious policy less and so forth and so i think thinking through that problem for delegation is super super important and it may be that rather than having this like continuous liquid democracy you might want to have some amount of guaranteed term lengths uh for delegates and you could think of creative ways to balance that with with liquid democracy ways that could be kind of cool the last thing i'll just say on it is it also might have a voter behavior benefit which is that we know voters tend to participate more if there's like a set time at which everyone votes because it creates a social phenomenon around voting and and that's something that's really missing in liquid democracy um do you have a question

yeah i was i was going to ask like do you feel like delegated proof of stake is being kind of a different category because that's it's not usually called a democracy but it has the same property that you can change that you delegate your stake to all the time i mean maybe stakers are doing a slightly similar task they're supposed to follow maybe a more set set of rules but they do have some like policy 26 right yeah yeah i'm super interested in that i don't have a good answer yet um i've been kind of like studying a little bit of that recently because it seems so important and i worry i i don't know as much about proof of stake but i have worried at times like if no one is monitoring

um could really bad things happen and so i think i think one of the things you're raising that could be really interesting to explore is sort of like how does the optimal rate of ability to redelegate how is that a function of the task that the the delegate is performing um just off the top of my head i would say and i don't know exactly how this applies to to proof of stake situations but off the top of my head i would say the more you're expecting them to do a concrete action on a concrete day that doesn't have like dynamic long-term consequences the more frequent you should probably want the

the ability to redelegate and the more you want them to be looking to the future and making long run investments and understanding issues probably the less frequent you want the delegation to be that would be my guess i'm not sure anybody really has like studied it that carefully or knows but the seems like the concern people express more is actually that people won't move their steak away from a bad yeah a bad scenario like that people will be way too sticky and won't be paying attention yes totally if someone's attacking like they won't like all flock somewhere else yeah yeah yeah yeah that seems especially severe if you're in this world where you can stake but then you get these derivatives and you can go off and play with them

and you don't really care about what you staked um yeah i think that's a so that actually raises kind of a broader point i wanted to make which is i've set this whole thing up about how indirect democracy helps but of course indirect democracy still only works under pretty optimistic conditions where enough people they don't have to make every decision but they do have to make this one decision really well which is they have to decide who to vote for and we see representative systems fail all the time because voters don't do that very well and so there's still a pretty optimistic part here about how do you get the delegators to do a good job and what you're raising is a sort of classic problem in voting

systems where well i don't care i i still have this collective action problem my vote in a in an election has very little impact on the outcome maybe if i staked to someone but i'm only one of many people who stick to them same thing um i think the issue i was raising is that people are really sticky yes and i mean i guess they're like that in congress too where they really have incumbents all the time but definitely like that's the perennial fear of delegating so i think that's a huge challenge and i think that's um in the empirical in the small amount of empirical work i've seen on liquid

democracy in other settings it seems like it has two pathologies one is this that like there's the super optimistic idea that you choose issue by issue who to delegate to and in reality people were like delegating to one person and not only that but people were hurting and they're all delegating to the same person which could be quite bad especially for information aggregation purposes and similarly they weren't reevaluating their delegation decisions over time and i agree it's com it is quite isomorphic to these voting problems where you vote and then you pay no attention and then you you vote the same way again and maybe it's the incumbent maybe it's for the same party you always vote for whatever um some of the solutions i think have to do

with with what i'm talking about here is sort of like one is making voting uh like a social enterprise where you all do it at the same time and you talk about it i think is important the third part here i think is relevant too is sort of this idea of a free press so it's like how do you get people to pay attention in large elections uh i think in general if you looked out at these voting systems in the real world you'd have the same exact concern you raised and if you just predicted from theory you might predict they don't work at all and so it's kind of surprising that they actually work of course we have tons of complaints about how they work but they work better than most other governance systems we've been able to come up with

alluding to the famous churchill quote and part of why they work better than we expect is if you make it this phenomenon people do actually pay attention and they do participate and part of the way we've done that part of it is having everyone elected at the same time so creating term links that seems critical but another part of it is like this other stuff that goes on around elections and a critical part of that is this idea of the free press this idea that you get accessible information about what's going on that helps you understand and care about the issues and given whatever your values are figure out where you fit in like are you like this party or that party and that this i think is something that is

missing right now um in this early forms of decentralized governance that i think will become really important and this is kind of one of my hobby horses is what a lot of dallas talk about i noticed today is oh we're going to build a dashboard that's going to tell you what your delegate's been up to is going to tell you about all the proposals and who voted how and we're going to track the participation rate of your delegate and stuff like that it's definitely useful but it's nowhere close to what we need in terms of a free press to make elections function because what i need is not data i need to understand the implications and these decisions are quite complex i need sort of like independent actors ideally with some

kind of profit motivation if possible um to be studying what's going on and telling me about it so i can say well i you know my values are over here therefore i support this proposal or i don't and this is where i think it'll this will be a really interesting evolution to see first of all are there ways to make that service profitable if not is it something that can be subsidized like it often is in the real world if it is subsidized who should be subsidizing it like is it at the ethereum layer is it at a very large inside a very large dow but not inside a small one i'm not sure but i think if we think about look at what's missing from getting representation off the ground i think that kind of making delegates want to do the job and then subjecting them to

important levels of scrutiny and analysis i think is the only way you make a representative system work any questions on yeah quick question on that can you give us a sense of what you actually think the sort of design problem for those sort of like auditors or curators or something is like you know because your ones first thought is okay maybe there's a profitable market for somebody just be doing that like you know auditing or curation if they have particular skill and doing it but then maybe they're subject to some sort of you know side payments that means they might like manipulate their point what's the what's the frame of the problem i don't know yet i think i think if you look at analogs in the real world you come away kind of stunned that it works at all and i would say it works way better than

i would have expected we have this there's one super optimistic story which i think you were alluding to which is that there's a competitive market for the provision of news and that does seem to work in some cases where where to the extent consumers are actually willing to pay for news um you can get you know good market properties from that but of course we worry that there's tons of pathologies where actually it turns out that people don't value high quality news very much and they view it as entertainment and and then we get you know problems from that and so i think i don't know yet in the web 3 space what the nature of the sort of like

consumer demand for this kind of analysis would be i think it might be kind of high only because in kind of a previous era of of democratic politics where there were kind of like high levels of interest in civic responsibility there was pretty high demand for it at the time um and you might think with sort of like optimism around around web3 among certain dedicated participants who do have quite a bit of money there might be a market there and then you could create competition and you know reputations for high quality news that lead you to sell more sell more of your news footnote it's funny like so news is one good analog of this i was almost

thinking more of like analysts you know sort of providing information to shareholders or something where where there's like a big you know value incentive associated with getting it right in a way that would like some types of political decision making maybe there isn't you know it's too much of a tragedy in the commons but i guess you're right that like you know for most types of decisions it's actually the news that's that's providing this sort of oversight super interesting thanks yeah yeah yeah we should talk more about the analyst thing because i definitely you see some of that happening it would be really interesting to understand where it's going to head okay all right so the second part is kind of like constitutional engineering so right now in you know in most of web3 though not all

you have this very very basic setup right where the token holders do everything or decide on everything um you know corporate governance to some extent has worked the same way in the sense that the shareholders are thought to be supreme of course these days there's been this enormous highly vitriolic debate over whether other stakeholders beyond shareholders should matter or not and how and if so how and it's just a huge huge deal in corporate governance right now with no obvious answer as far as i can tell so far and we see the same thing happening unsurprisingly in web 3 where there i think are our well-founded concerns that it's actually hard to get a

meaningfully decentralized self-governing community off the ground if it's dominated by a set of large token holders so there's like you know there's a famous vitalik post about this lots of other people have pointed this out too that if you rely entirely on token voting it might be hard to create that sense of legitimacy and that sense of sort of a widespread shared interest in the success of this enterprise when all the power and all the rents are concentrated like that and so we've seen people um explore a lot of possible ways um to address this to bring in uh meaningful roles for people who aren't large token holders and and some

of those people could be the people who actually work uh for the protocol could be other kinds of active contributors people who are you know making good rather than bad posts on the forum and so forth users of the protocol small token holders we see the same exact thing in democracies and so we've seen an evolution over time in how systems are set up in ways to guarantee in some sense representation for certain groups and i think that this is heading in exactly the same direction and once again i think what's interesting is web 3 is kind of like rediscovering issues that have been around for a long time but also tweaking them with you know with new technology so just to give

you one example of this because i find if you if like me you're from the u.s you may not know about this because the us is an outlier in this but huge fraction of the world actually has quotas in place for women in legislatures um in fact i think it's the majority of the world maybe not by population and so i think this is one really important example of where you engineer a system and you don't just say we're just going to let voting do everything you say from the outset we're going to design the system to guarantee a certain type of representation i'm just offering this as one example there's tons of other examples you can think about bicameralism in the u.s is basically a

guarantee to the states that they'll you know that the smaller states will have an outside portion of power majority minority districts in the u.s are supposed to be a guarantee that minority populations have representatives of their own by giving them a majority vote within a within a certain district there's lots of other examples so you can you can build a system to weigh different voices in different ways and we're starting to see this i think because there is some perception that it might be bad to rely rely overly heavily on token-based voting you're seeing the evolution of new systems like this so a couple of examples so one within the framework of token voting of

course is just to give uh certain groups of certain people more votes through retro like retroactive air drops for people who've contributed to the community or something like that so that's one way to try to to get at these problems but what we're seeing now i think which is in some ways more interesting and more creative is the creation of new governing bodies where the selection into that body or the power of those bodies is not directly related to the number of tokens you hold and so optimism's citizen's house was one kind of interesting optimistic if you will um attempted this where the idea was there's going to be a token voting and then there's also going to be this other

chamber that you get into not based on how many tokens you hold but based on your contributions to the community we'll give you a special token that gives you the right to be in that house and that and the citizens house is going to make a set of decisions about the community particularly around public funding for projects separately from the token holders so it's almost it's a kind of bicameralism where you're creating it's almost like you know the token holders are the the senate or the house of lords or whatever you want to call it uh and the citizens house is like the the lower chamber um it's a weak analogy but it's sort of there and you can imagine going much further in creating different kinds of assemblies that have different powers

where membership in that assembly is a function of something other than the tokens that you hold and depending on your problem and which stakeholders you think are important you know you could design it to make sure that those stakeholders are represented um finally this is where i we haven't seen as much yet but just based on the history of governance in the real world i would hypothesize that we will see it is this idea that well if you're going to rely on a system of representation through delegation then that opens up this possibility of actually defining groups of delegates so instead of just saying anyone can become a delegate you could say we're actually going there's going to be a group of delegates who have explicit groups that they represent

because they're from that group because they're familiar with that group whatever uh and they're going to have some amount of voting power that's not just a function of how much has been delegated to them freely by people and this is kind of like the quota system from the past slide so you can imagine you think there's a group maybe it's small token holders i don't know it could be any group who you think is underrepresented in the process you could guarantee them a delegate and that delegate could have a certain you know amount of power these are things i expect to see happen as the stuff evolves okay last part i want to talk about and then i'll try to to wind up here um is this problem of oversight

and as i said before when you have a system that relies on direct democracy and participation is inevitably very low one of the problems you then face is that you have this people get this sense not always justified by reality that the people who are working for the organization are not being monitored and could be doing all kinds of bad stuff that's bad for me as a token holder that i'm not seeing because i have the sense that the process is kind of like run out of control and this is exactly the problem i was talking about earlier when i talked about the east india company and why boards of directors were set up precisely because as the number of shareholders got larger the ones who weren't managers of the company became obsessed with this problem that

the managers were secretly conspiring against them sometimes because it was true and it happened other times because it's natural to become suspicious when another group of people has incentives to screw you over and you're not able to monitor them and this has been happening to some extent in maker i'm not an expert on this so i'll just say my impression is that actually the core units have been working quite diligently and hard but there is not a lot of oversight and this has led to all kinds of dramatic accusations that the system is growing and growing and growing out of proportion with its needs that the budget is being overrun as a result it's a classic governance problem and it's the same thing we see happen in

politics and so in the u.s we have exactly this problem that there are a ton of people whether accurately or inaccurately hold a very strong perception that the bureaucracy which is the political equivalent of the workforce is unmonitored and doing all kinds of things that voters and taxpayers and whatever don't want and there's this whole political movement right i'm showing a particularly dramatic example here steve bannon called it deconstruct the administrative state i don't think he turned out to be super successful but uh there's this idea it's hardly unique to the us that when you can't monitor um this kind of suspicion and anger and distrust festers

um and that i think will be a huge problem in any decentralized organization that grows to a large enough size and i think it's precisely some of the growing pains maker maker seems to always be first in exploring these problems and coming up with pretty interesting solutions to them this is i think exactly the problem they're facing it's sort of like if you don't have a ton of participation and ton of monitoring how do you create that trust and i just want to emphasize creating those mechanisms of oversight is critical not only because it prevents bad actors from doing stuff within the system but also because it creates trust so even if the workforce is already doing a good job that's actually not sufficient you need them to be doing a good job and to be trusted by the token

holders for the thing to work in the long run and so creating oversight is important even if you think today the system is working well this is kind of the idea of trust but verify um and so how could we do that i think if you had you know look again to other systems that have representative democracy it's some combination of the delegates themselves have are tasked with some amount of oversight and they're given staff with which to perform that oversight or they delegate in turn to some kind of committee like an audit committee or a group of people whose job it is to study and to monitor what's going on uh in the workforce and there was a kind of a proposal like that at maker that became

super super controversial and hoping that you followed it um uh but there are these ideas out there to start forming these kinds of things and it's supernatural right that's like the job one of the jobs nominally at least of corporate boards of directors is precisely to perform this sort of oversight just like through yeah public nature of blockchains to monitoring but actually easier than traditional settings yes i definitely think so i of course there's always like what you can see and what goes on in the private chats and stuff but yeah i think it's a it's a very different starting point i think it's a big challenge a big challenge you saw in the history of corporate governance which you also

definitely see in democratic governance is are these information asymmetries where the people working on a day-to-day basis know so much more about what's going on that they can they can run circles around you as sort of like an outside shareholder or even an independent member of the board because they just know so much more to start it and you know at baseline the outside people couldn't see anything about what was going on and then there's the growth of all these rules around how audits work and what kinds of information have to be disclosed and sdc regulation and all that same thing in in democratic governance like public records laws etc and they're hugely burdensome and they don't work super well they're way better than nothing but yeah i think blockchain is starting from a much higher potential level

where lots of the stuff that you want to know about is already public and what you need over and above that is the sort of analysis and research and interpretation of it which in some ways i think is is the easier barrier compared to systems where the actors have strong incentives to hide everything see i think that's an interesting point okay so those are like the three potential paths forward that i think are like both interesting and promising and likely i think to occur as governance moves forward i just wanted to stop here at the end so we can turn to kind of discussion and questions with just a few notes about um where i where i'm hoping my research and

other political economy research could could go towards in this space um so i think the most obvious and important thing for people studying governance in the space is going to be is going to be to try to develop you know a science of this sort of like what is happening and why is it happening so that we can understand how to intervene and to make the things we think should happen happen and so we need like uh we have a bunch of ideas from the study of governance in other areas but we need to develop them and test them in this space um to start to get kind of like towards a mature idea so i've thrown out a bunch of hypotheses but how do i actually start to test them and figure out what works and so um some of the things i just want to

list out in the space that i think are particularly important and interesting to me at least is sort of this question of participation who participates and why how can we get more people to become informed and to participate in good ways and where i think web 3 is different from what's come before is in the fine grain incentives you can give to people through the way you design the tokens through the way you design the system and so that's new and totally different and so how do we develop a science of how to do that in good ways for governance i think is super interesting second is like kind of coming from the study of legislatures where i spent a lot of time is sort of like how do you make this representative system function well so if the first question is about

how you get voters to function well the second is how do you get delegates to function well and how do you organize conflict so we have this idea from the study of politics that uh conflict is inevitable and that if you're going to have conflict over how to make a collective decision which we see all the time in web 3 of course how do you make it stable so that through you channel the conflict into a process that one is stable in the sense that it leads to predictable outcomes that everyone agrees to and views is legitimate and so that the conflict itself actually informs people about the stakes and so we go back to this question of like how do you get people to pay attention and know when to

unstake or when to delegate when to undelegate a lot of what we think happens in politics is that by having a debate in public about the key issues and and sharpening the like what are the actual disagreements that we have um is how people learn and how you get to a kind of structured political system and so how do we how do we get these sort of like clear conflicts um to be channeled and to be stable in this way is i think a really important question um and as i mentioned just the sort of like early work that dan and i have been doing on this is we're developing this dashboard that in addition to showing you kind of like who's participating and

when across daos it's also going to explore what these patterns of delegation look like across daos that use delegation as well as what the actual votes on proposals look like and what the uh kind of like structure of preferences over those votes is so i just want to show you some very very initial stuff that i did on the airplane yesterday on this note so i'm showing you here this is super preliminary is we take all the proposals that are voted on in each dow and so we have addresses for the voters and we have uh the different proposals and we have you know the number of tokens voted so you can think about it's like a matrix where the the rows let's

say are the proposals and the columns or the addresses and each cell is like the number of tokens voted in favor um that opens the door to a you know there's a very long and i think super fun uh research tradition in political science and political economy about interpreting voting data and clustering people into groups into coalitions essentially and so a very easy it's not the fanciest way to do it but a very easy way to to start to do that is to just do a principal components decomposition to to this matrix of votes so what i'm showing you here is and if you if you haven't done that before it doesn't matter this is um i'm showing you the screen plot that just it gives you a sense of how structured conflict is this is for

compound and so the fewer dimensions you need to explain the variation in these votes the more structured we think the conflict is and what's really interesting here is this looks we think the us congress is super one-dimensional not entirely there are other issues that matter but most the variation in how peop legislators cast roll call votes in congress is largely explained by this single first dimension that has to do with like how left wing or right wing they are the scree plot for congress doesn't look that different from this one for compound which suggests way more structure to the conflict within compound than i would have expected i am not an expert on compound governance i'm not even going to attempt to interpret

this for you but i just want to give you a hint of where we're heading in our research is you can actually then you know if you have a space like compound where the first two dimensions do a lot of the work and explaining how different people are voting you can actually plot it and start to look at who's voting with who and so you could think about clustering this into groups of of actors that seem to be whether intentionally or purely accidentally voting in in similar ways to one another and that might for now this is just kind of fun but it might lead us to some kind of more mature idea about what is the nature and the scope of conflict within these daos and how can we channel it effectively there's no a16zs right here right near the cal blockchain group

um yeah this is the same scree plot uh for uniswap and what it's showing you is it's not nearly as as low dimensional it you actually it looks like you need five dimensions to sort of like get a good explanation of who's voting on what and so it's much less structured i don't know why that is yet but that's where i want to go in this research is like take things like this across lots of daos and start to understand well why are some so structured and wire some less structure so i take it here you looked at sort of specific motors and the corresponding volume vector some normalized projection i'm just plotting the first and second

component from the pca yeah and like is it easy to say like the various quadrants like yeah like when you summarize it can you summarize the political position that so that's where i want to head so that is totally possible but it's completely qualitative so what you have to do is go back to the loadings figure out which proposals have high loadings and then read them and understand what they're about and that's definitely something i want to hire a bunch of students to do in the near future because i do think for context where there is structure it it almost has to be the case that there will be certain issues that are

explaining the different quadrants of the space so yeah that's that's exactly where i want to head is there other enough proposals that you could just run a text algorithm and try and figure out what like automatically potentially yeah potentially people tried to do that for congress and they've never quite cracked like how to do it well but um but it's definitely possible i thought it worked okay for like was that the scholarship yeah yeah so you can scale yeah so they're the nice they already know the dimension of conflict they're trying to like predict whether you're republican or democrat so they have kind of an easier job because it's like i know there are two parties and i'm going to see which words predict which party here it's not clear who the parties are

um but i but we could probably discover them by looking at the words that are most associated with the proposals that have the most loading yeah yeah um do you try looking at like the causality between who votes and like who commits their vote first or who yeah or yeah who announces yes i'm super super into this i haven't done it yet but it's on our list of things to do obviously the data does have that time dimension in it because none of this voting isn't private no and this is one of the areas i'm most interested in dan is i think really interested in trying to implement different kinds of private voting is

that we know when voting is public and done over time it creates super interesting equilibria in voting strategies and so there's a great paper on this um by a friend of mine pablo montanious about the us senate where voting is public and done um in like it's like partially alphabetical but there's some randomness to it or something yeah it's super cool yeah yeah and there's this kind of unintuitive aspect to it which is like in some ways you think you'd like to go last because then you've seen what everyone else does but then that means people in front of you if they know enough about your preferences can like force you to vote in a particular way by

voting before you if you're gonna be pivotal or something like that and so it definitely creates all kinds of weirdness um the fact that the voting is public is also super you know where quasi public is also super weird and you worry about it with that cluster if they're all making those decisions independently or just copying yeah yeah for sure yeah that's definitely a direction we want to go in okay so this is i just want to wrap up um uh where i think this is heading or like what really excites me is this idea

that there are these classic problems from the history of governance that are relevant to web3 but web3 is also different in new and really interesting ways and the reason i've gotten super excited about this as an academic is that i think we can learn a bunch about governance in general from the space too and so i just wanted to kind of list out some of the things i think are so interesting so i've spent my career looking around for natural experiments because what i tend to do in most of my papers is like collect data on some governance outcome and connect it to some change in the structure of the governance system to try to get at causal effects of like okay so you put in place term limits what happens or you allow people to vote by mail what happens

and you have to search around it's super hard because actually most physical world governments aren't changing the rules very often for kind of obvious reasons and there's tons and tons of friction but in web 3 people are changing the rules quite frequently and in fact one of the things we're doing in our dashboard or we're trying to do at least is to actually like automatically detect changes in the governance structure through the through the smart contracts and so where the hope is it'll be this kind of amazing laboratory where different communities are are mutating and changing the rules a lot and then of course we can get all the data um right off the blockchain and understand what the consequences of these reforms weren't so i think it's a just a super exciting um space that i'm really enthusiastic to do research in and then

the last thing that's my last slide um is also i'm not just interested as a researcher i think what's gonna what's coming out of this already is super super interesting from a governance perspective in the real world too and so i mentioned before i've i've done a bunch of work outside of academia at very very large platforms i'm super interested in the enormous governance challenges that they face both as i said the social legitimacy challenges but also the challenge of getting buy-in from developers creators sellers and so forth i think there's just like this potentially really interesting opportunity to democratize these large platforms in kind of the spirit of the glorious revolution that i started this

talk with uh and to bring democracy to some of these decisions about you know hard policies on social media platforms or policies about how developers make money in app stores things of that nature where we could use democracy it's clear that it won't be direct democracy facebook actually tried to do a global referendum i don't know if any of you remember it it was a super long time ago the participation rate was i think three basis points or something like that and so they cancelled it i'll talk more about that in my next talk but representative democracy the kinds of delegation systems that that web 3 is now exploring i think could be a really big deal

for fixing some of these big governance problems in web 2. and that's what i'm going to talk about in the next talk next week so i'm going to stop here love to take any any further questions thank you [Applause] differences in the data between governance systems focused on tuning protocol parameters and those focus on treasury allocation i haven't looked at this question yet but i think it's a really important question porter is actually really quite interested in this question he's pushed me pretty hard on it because i tend to like think in terms of democracy and porter tends to think in terms of like money um and if

the purpose of the dow i think matters a lot for what kinds of governance you want to have and so like one thing for example porter really pushed me on is like uniswap is very simple it doesn't need a complex governance structure maybe because it's somewhat minimal um maker is making slightly more complex decisions maybe it needs a different structure if you ever had a successful social media dow or something the decisions i can tell you having worked at meta will be much much harder um so i do think there are probably different ways to structure them this goes to this question of like the science of the space like what is the optimal structure for each purpose and in terms of the data to get at the question directly no we haven't looked

at that yet but once we only have a small number of dials in the dashboard today once the dashboard is built out the goal is to make exactly those kind of comparisons this will basically have a bunch of you know attributes of each of these dowels that you can slice the data by there'll be some risk of you know over interpreting different correlations which i'll have to be careful about but my goal will be to get at actually kind of natural experiments of changes in different things that are going on and how they affect governance outcomes yeah so to follow up on that do you have like uh i don't know a short list of dimensions that you would say affect whether you want to have one kind

of system versus another one so like for example i can imagine that like one difference between um i don't know like political governments and corporate like well not being a corporate terminal but i assume most of them are just interested in the same objective they want to maximize profit but maybe they disagree on the best way to do that yeah whereas like in a political government everyone genuinely disagrees about what they want to do yeah and i'm wondering like what are other demands like what is that a relevant dimension and what are the other relevant dimensions like if you had a short list i think that first dimension is critical because the governance solutions in a space where you have hold a common value and different beliefs is is very different from when you have different values

um uh i would say i would say on the corporate shareholder side two other dimensions that differentiated an important way or one that differentiates it one that makes it more similar to democracy the one that differentiates it is the ability of exit the ability of exit is critical and that's obviously to some extent also present in web3 so so how do you create a governance system where people are voting but they can also quit at any time um uh you might have kind of like a lower obligation for certain kinds of governance in that space right so like part of why social media has become so problematic is the sense that the platforms are so large in some sense you can't exit and therefore they

need certain kinds of democracy to legitimate themselves that if you can exit you might not need um so that's that's another dimension the third i was going to mention is um the accountability problem so one thing that goes on in corporate shareholder governance is what you're saying is like it's a common value problem and they're just trying to figure out the best way to run the company but there's also this principal agent problem which is uh i'm over here i'm a shareholder and i'm worried that the ceo is you know enriching him or herself at my expense and that's pretty similar to problems we see in democratic government and in problems we at least worry about in the web three space so those are some of the dimensions i think are most important

uh in addition to the one listed out in the chat which is kind of like and what is the purpose of this enterprise um i have a completely unrelated question this is um are there any so like you mentioned that now that things are either more digital or on the blockchain there's lots of different technical tools that are possible that are not possible with the kind of voting teams that happened like 10 plus years ago yeah um i'm wondering are there any like fancy algorithmic schemes you've seen that actually seem interesting or do you generally feel positively or negatively towards development

solutions yes yes i do i'm probably an optimist compared to the average political scientist on this um i do i'll give you a few that i think are really interesting first i think liquid democracy is genuinely very interesting i think as i allude as i talked about earlier there's a critical trade-off and i think liquid democracy as implemented right now is too probably too far on one end but there's a whole spectrum there's nothing to prevent a form of liquid democracy where there's periodic redelegations or something i haven't seen anyone study that but it would be i think that would be really worth exploring and i think that there's something about that idea that's quite quite

interesting uh another one that might even be more important is just the way you can design the token and so these ideas about you can kind of like encode into the token any laws you want around how voting works so things like compulsory voting i think are super interesting excuse me um but you don't have to make it compulsory instead you can like fine-tune the level of incentive or disincentive you give for different types of participation i think that's super interesting and related to that at this same kind of coin if you will other side of the coin bad pun um uh at the delegation level you can also create fine-grained incentives they're like with liquid democracy i think there's like trade-offs you have to be

super careful about i've seen some stuff i think maker actually already does this and i i don't know that it's the way i would do it necessarily is like trying to create your compensation as a delegate as a function not only of how much you're delegated to but also like your participation rate and stuff like that and and i think on the one hand i think it's super promising because i think if you look at models of accountability like one of the reasons accountability in elections is so hard is because the the power the voter has is so coarse like all they can do is hire and fire which is crappy and if you look at like you know um this stuff in in labor markets you get this much more fine-grained incentives because you can

pay people and give them bonuses and stuff like that and promote them and so getting towards that i think can be really really promising uh and and in fact i think is so promising it could even kind of like reflect back into real politics and giving legislators better incentives but we also know that the problem of being a delegate or a legislator is super multi-dimensional and you worry a lot about incense creating incentives along one dimension at the expense of others and so we see this kind of like effort allocation problem a lot in politics where like the only thing you care about as a voter is like how many votes i show up to well i can like you know maximize that and then spend no time actually studying the problems that are and so you know we

have classic um classic papers on that problem of uh the multi-task problem so i think that's one where i'm super interested too okay so like you mentioned liquid democracy it sounds like so far you're kind of like presupposing a certain aggregation rule it would just still be like a majority code base after delegation and like one of those like i i haven't read a lot but some papers that get submitted to cs theory conferences are like what if you cap the you know like what if you cap the maximum influence that you know you have and like in some models i don't know that helps get rid of some you know the bad thing effect and i'm wondering like

in general is like i don't know if that stuff has even made it like political theorists is that kind of thinking helpful or mostly missing the point or yeah it's interesting i don't know but i'll give you just my sense my sense of what i've seen on liquid democracy so far has been that it's been quite focused on the common value problem where like you're trying you have different beliefs you're trying to aggregate and stuff and i i'm not aware of any papers in cs or political economy that look at the accountability problem which is um if if the delegates are then going off and doing stuff are they doing a good

job do they worry about being undelegated to and i think that difference could be important for your question because capping the influence of nodes is probably very good from an information aggregation perspective i would guess but it could be it seems ambiguous on accountability because it kind of like caps my incentives as a as a delegate to get more delegation so i don't know my real answer is i don't know but that's my first thought i think it's a that whole area i think is is a is super ripe and actually i work with uh um super interesting political theory um professor at the harris school uchicago's name is ethan bueno de mosquito who's super interested in these questions i'm hoping

hoping to persuade him to write a paper on liquid democracy and accountability uh lyra asked what kind of decisions do you think could benefit from both votes being secret yeah this is a great this is a great question a passion of mine so the most important there are a lot of reasons to think secret voting is valuable there's some arguments for why it could be bad also the best arguments for why you want secret voting i think there's at least two the first and probably the most important is to prevent vote buy-in and vote buying could be good in some circumstances like we don't tend to think it's that problematic for someone to come in and buy a bunch of shares in a publicly traded company and like influence the direction of the

company through that mechanism but depending on what the nature of the buying is and what the decisions the dao is doing it could be very bad at the minimum for the legitimacy of the organization but also potentially for the future of the organization um if it's very easy to pay people to vote the way you want them to vote and this is a big debate i know there are some like purists who think vote buying is purely good in this space but i think that we could come with a lot of examples for why it can be very bad and the whole evolution of the secret ballot or the australian ballot in democracy was precisely about this because in the past it wasn't secret you actually in the us

and in many other countries too you had a you would take a ballot and the ballot would be colored according to which party you were supporting and the party would watch you put it into the ballot box and then give you a beer um and uh and we've seen a lot of very good benefits from moving to secret voting because it prevents that if i don't know how you voted i can't easily pay you to vote the way i want you to vote because you might take my money and not vote the way i wanted you to um so that's like i think one way that secret run can be super useful the other is in terms of information aggregation um we tend to think you know there's sort of like conversate jury type things if you're familiar with that like

social dynamics of public voting can be very bad for information aggregation there's some great work on this in the context of the federal open market committee where there's evidence that like they disagree with one another less now that the transcripts and the votes are all public and you worry about getting this kind of hurting where someone important someone you care about comes out and says they have a strong position on this issue your vote's going to be public now you're hesitant to vote but maybe you have a private signal that what that person thinks is the right thing to do is actually the wrong thing to do you now are breaking the information aggregation process so that's another argument for um for secret voting so and there's a lot there's lots of others too yeah thank you everyone i'll leave it

there you