Bringing Decentralized Governance to Tech Platforms with Andrew Hall | a16z crypto research talks

Transcript

all right so welcome everyone uh today's uh may 16 crypto research seminar i'm very happy that we're being back hopefully you saw us talk on thursday but he's back again today andy hall from stanford university he'll talk about bringing decentralized governance to platforms thank you long time no see um cool so let me give you a sense of what i'm thinking about what i'm talking about today it's a it's a fairly conceptual talk please uh do interrupt uh with questions throughout so i don't know if any of you saw this video or recall it but this is part of an epic pun intended conflict between uh epic games the the maker of fortnite and apple that is actually still ongoing and roughly

speaking the timeline uh was that epic intentionally snuck in a way to get around the app store's payment requirements apple then removed them from the app store and then they sued apple and on the day they sued apple they also released this kind of amazing video which is a parody of a really famous old apple uh television commercial about big brother in 1984 and they're kind of implying that in its role as the steward of the app store and the sort of monarch of the app store if you will um that apple has taken on the role of big brother and there's some i think remarkable

quotes from the lead from one of the legal filings that epic made of course take these with a grain of salt because they're from epic's point of view that expose i think some really important governance issues with modern platforms that i'm super super interested in both in terms of my academic research as well as my work in tech and so just two quotes i wanted to read out quickly both from this legal filing uh quote developers have faced an inefficient and opaque app review process riddled with arbitrary decisions and errors coupled with poor customer service as apple has gathered more and more users into its ecosystem and locked them in the importance of the apple ecosystem to developers has increased to the point that nearly all developers rely on apple

but apple does not need to rely on any single developer i'm super interested in this really important conflict because i think it exposes some very core governance issues that we've experienced in other parts of society in the past that have to do with when you're subject to a system of rules that you didn't play a role in shaping and that if you don't have sufficient outside options you might have no choice but to live under and this problem of how these large tech platforms are creating these systems of rules and enforcing them it's just super it's ubiquitous across the space so i'm going to give you a few more examples um to show you why this is important so of course a totally different example

from apple here's twitter banning the at the time sitting president of the united states and maybe somewhat surprising given the politics involved quite a few global politicians actually came out and said we don't think this is a good idea and so in particular uh merkel as well as macron and others in in the eu who again on the basis of their politics might have thought maybe trump shouldn't be on social media saying that this is actually very problematic that a private social media company can just set rules arguably arbitrarily and knock someone as important as the president of the united states off their platform and because so many users might

be locked into this platform that might have a meaningful effect on who's able to see what this is not obviously limited to twitter we know that the exact same issues play out on facebook on instagram etc i just want to highlight though the issues are also not limited to sort of like binary rules that you write down they're also about algorithms and so another long-running concern of this nature has to do with facebook and the way the news feed the ranking algorithm prioritizes some content over others which then creates incentives for news producers and news producers actually arrange their businesses to do well under the news feed ranking algorithm and then

when the algorithm changes it sort of picks winners and losers sometimes accidentally sometimes people have argued maybe intentionally um and so people are again it's another governance problem that people are quite worried about so how do you create these ranking algorithms is affecting a bunch of people in in unpredictable ways this also isn't just about uh content on on existing platforms it's also about platforms of the future so we're already seeing this play out i see this in my own work at meta lots of difficult issues about how vr platforms are gonna regulate their content or conduct as the case may be i also wanna highlight it's not just

about uh you know web 2 platforms it's also about these newer platforms that cater to web 3 but that have certain web 2 characteristics and so you think about for example coinbase which has faced its own difficult content moderation issues or you think about openc which in addition to these content moderation issues is also exploring actively these sort of automated techniques for for content moderation at scale similar to what meta and others have done and so this is a super wide-ranging set of problems that i would describe as platform governance which is something that i've spent a lot of time working on and i'm super super interested in and part of what's drawn me to web3

i would define it as the system of algorithms product architecture rules rule making procedures appeals processes and other institutions that shape how platforms are used and the thing i want to talk about which is a continuation of what i talked about last week is that today in the way these platforms are governed they're governed largely like monarchies and i talked last time about some of the pros and cons of that but basically this idea that platforms are allowed um to sort of unilaterally set all of these rules and to govern their platforms as they see fit obviously subject to a variety of kinds of regulation but with a lot of latitude to decide how they create their rules how they enforce them

when they enforce them what kinds of appeals you have to the rules how they rank content etc etc and what i want to kind of focus on today is two potential benefits that i flagged last time to bringing more stakeholders more users creators developers etc into the process that governs these platforms and so i've as i mentioned last time i think there's two potential key benefits uh to exploring this first i think there are arguments for why you might actually be able to encourage more growth and innovation on these platforms if the people doing the innovating and selling their products or developing their products are actually involved in

the governance of the platform and also for these big uh legitimacy issues that we're increasingly facing especially with social media platforms but also with other types of platforms these i think are two potential benefits um to democratizing web 2 platforms in different ways so i'm going to talk about those two in order and uh and please do stop me at any time okay so let's talk first about platforms economic growth and what's known in economics is the holdup problem um i'm going to draw here on some work that i do with a frequent collaborator of mine ethan bueno de mesquita who's a professor at the harris school at uchicago and is a political economist and game theorist

our argument in this piece that we're kind of trying to build out into some new research and so would love your feedback on is this idea that actually the way you govern these platforms just in the same way that the way you govern a society structures its economy and can lead to good economic outcomes or bad ones this is like the puzzle of economic development in just the same way how you govern these very large online platforms could affect outcomes in similar ways so platforms actually need right to attract all these different people to the platform obviously they have to provide a service that's valuable to users they then have to create this flywheel where they're also bringing in the developers who make the apps or the sellers who sell the goods or the creators who make the content whatever

in a way that makes everyone better off and so the more of those contributors you get to your platform right the better service you have for your users the more users you have the more compelling it is for these contributors to come in where this problem occurs this hold up problem that i want to talk about is after a platform gets very very large the exit options for these various contributors to the platform may diminish right and that's sort of a big part of the argument that epic games is making is that it's just not the case that the developers can say i don't like your rules so i'm going to go over to some other app store they can do that in a very limited way but they lose a lot of potential business as they do that and the more locked in they are to this platform the more

severe this hold-up problem is so what is the hold-up problem the hold up problem is that once you're at scale as a platform it's very hard for you to promise that you're not going to use that power to start taking more for yourself and giving less to the developers and the creators and and so forth and you know if you go out today i've been trying to quantify the size of this problem across platforms it's kind of hard but certainly at least anecdotally when you talk to people who work on these platforms you notice this across many platforms right growing sense that they're not getting a fair share growing reluctance to invest in learning how to build for a new platform if this problem hasn't been solved and so i'm

not at all trying to argue that today there aren't you know lots of apps on the ios store or anything like that but on the margin think about someone just getting out of college or something who might want to make an app maybe they're a little bit less excited they than they used to be like when i graduated college i went to stanford in 09 everyone wanted to make an app of course some of that is because it was new and there was just more low-hanging fruit obviously but another part of it is today i think there really is a strong sense that there are certain parts of the deal that are just not as compelling as they used to be and the hold-up problem isn't the fact that you are in equilibrium mistreated by the platform it's that anticipating that you choose not to invest in the

first place and then and therefore from the perspective of the platform they're not getting they're not maximizing the amount of contributions that they could be if they could somehow promise we're not going to raise fees later we're not going to have these you know these hard to understand rules and use them in ways that advantage us we promise you we're going to stick to a fair deal if they could do that they could get more people building on their platform but it's very very hard to do that and i think we're in a really interesting phase right now where platforms are just starting to grapple with this because these conflicts are getting so severe so we had you know obviously you saw the the example of fortnite which is maybe the most the highest profile but there's tons of other there's there's

third-party sellers going to war with amazon there's like the etsy boycott i don't know if you any of you saw that this is happening all over the place uh and so i think it's a it's becoming an increasingly large problem and a business opportunity i think is my argument at least for platforms that can innovate and how they're governed to create a credible promise to their contributors that they'll treat them well and just to kind of you know we haven't formalized this yet though we're working on it but if you were to formalize it i think some of the key parameters that might influence how large this problem is how serious it is would be sort of how large is the equilibrium platform in this space right if there are very large network effects then you're going to get that scale you're going to get that lock in potentially

how much platform specific human capital the creator or the developer needs that's going to drive lock-in right the bigger the investment in your time and energy is the more reluctant you're going to be to engage in a badly governed platform similarly to that on multi-homing like if it's very hard for you to multi-home that might that increases lock-in further um and of course like for creators especially imagine like uh if you can't port your data sorry so you build this big following on tick tock but you can't bring it with you anywhere then that leaves you more locked in that leaves you more subject to this problem ex-ante maybe people will become more hesitant to engage in the platform in

the first place quick question here so another thing one could do is like just issue formal commitments right like you know commitment devices are are harder but not impossible and smart contracts make them easier um how do you think about like is it about inability to commit or is it lack of incentive to commit um i think so definitely my historical thought is both oh sorry yeah so the question was are there other ways to solve the commitment problem and has the problem been driven more by not being able to commit or not having incentives to create the commitment i think both have definitely been in play i think if you go back to the early days no one knew how these things were going to end up people were just excited to get involved and so it wasn't obvious

yet that this problem was going to occur but one of the lines we have in the piece is sort of like um you know fool me once or something like that it's like people are learning now that this issue could be there and so for the future both if you're thinking about new platforms or if you're thinking about incentivizing effort on the margin this could be more important now so there could be stronger incentives than there used to be in terms of ability to commit we've been thinking about it mostly like if you're in the web 2 world it's very hard to commit to this stuff i mean you could you can certainly try to use the legal system to make the commitments but we know why that's not a super strong commitment because the costs of litigating this stuff are super high and the outcomes are super uncertain and so

our thought was along your lines that web three the blockchain might actually open up the space to create those more credible commitments now okay so we're going to think about this uh in kind of uh well we're political economists so we're going to think of it in kind of a political economy style way which hopefully is useful um which is how do you solve uh commitment problems of this form when you can't rely on just signing a legal contract together well one thing you can do is you could observe that this is very similar to the problem that societies face so if you if you believe at least which i do these stories about economic development that one of the key things that drives economic development is the

ability for the state to credibly commit not to expropriate you in the future not to erode your property rights which makes you willing to invest to start a business whatever then you can follow the same kind of logic so there's some fantastic work in the political economy of development that's about how society solved this problem you want people to feel empowered to to start businesses to to engage in economic activities you need to promise them in some way that you're not going to screw them over after they start to do that otherwise ex-ante they'll be reluctant to do so how do you do that well of course you build a strong legal system with property rights harder than it sounds maybe it doesn't sound so easy uh but you also combine it with democracy and that's sort of the

really important insight from this literature is that if i tell you today you know this is what the tax rate is going to be this is the process for how you own property if someone wanted to take property from you here's the process for how you would do that and i promise you it's just going to stay that way forever you probably wouldn't believe that because if you looked at the history of human societies you would observe that those kind of promises don't tend to stick in particular if the promise is made to you by someone super powerful like a monarch then you might think to yourself the monarch can't tie her own hands like later if she sees you've made a lot of money she's the lawgiver to the land she can just change the law and this is exactly what

happened historically and i mentioned this in the last talk this was the problem that the glorious revolution was intended to try to solve in the uk it was that the king was repeatedly rewriting the laws in order to extract more resources from society we see this all the time it's by no means limited to the uk um and democracy was part of the way you could make they thought dynamic commitment that you could say not only are these going to be the rules today but i'm going to give you a process by which all rule changes in the future happen and they involve you and so how does that magic trick create commitment because of course the powerful person could just say later no i fooled you i'm not going to actually honor these

elections or something it's a complicated process but the idea is that it creates a coordination point for voters so in the future if i violate that promise it could be very hard for you to punish me but at least you have this clear focal point which is that i promised you the election and then i took it away from you and so the idea in this literature and admiral and robinson is probably the most famous in it although it's a pretty deep literature is this idea that elections giving uh the franchise to society is a key part about how you make long-term commitments the sort of like procedural commitments that undergird commitments to to develop the economy our idea is that and it's not even just our idea a lot of

people are talking about this in web 3 is that there's actually a really deep parallel here to web3 which is that if you can give these developers these creators um whatever contributors this platform this kind of like permanent commitment to governance power then they can really start to believe this is a special platform that i can really build for in the long run because if they want to change the maybe i like the rules today they're taking a reasonable cut i don't want them to take no cut because i want the platform to succeed i just don't want them to take too big a cut maybe the rules on what's allowed and whether my app can get delisted and stuff is very clear i like these rules i want to believe that even

after this platform gets very very large it's going to stick to these rules maybe i believe that more if i've been given some arguably somewhat immutable voting rights over any future changes to those rules so that's kind of the analogy that we're thinking through and so i think what's interesting about this isn't just um that it could be in the interests of these contributors to have the system but that it actually could be in a platform's interest too to the extent it actually increases the total amount of investment that's put into this platform by people so that's the idea and that and the general idea here is just that we've learned through the history of human societies that good governance is

actually vital to driving growth maybe the same logic applies for these platforms so that's the that's sort of what i would call the commerce problem or the economic growth problem i mentioned there's all this second one which is legitimacy so let me talk through this one now and then i'll talk about some of the tools that might actually be useful for implementing this in practice okay so the legitimacy problem is different from what i just described the legitimacy problem is similar in that it's when you're at a large scale you start to worry that if the rules being set on this platform are important for society and if people aren't so easily able to exit the platform because of its scale

that you're now setting rules that are consequential for society but you're not setting them through legitimate mechanisms and what i mean by legitimate mechanisms or a legitimate system of rules is that it's a system where people believe that the rules have been set through a fair process and this is a you know a deep area of political philosophy is like when are rules in the real world legitimate we tend to think one really important way they get legitimate is if they're set by a government that has some story for why it's been empowered by its society and perhaps the democratic elections of course the rules also need to be not crazy because no one wants to follow

crazy rules but if the rules are reasonable and they're set through a fair process then people are more likely to accept them as legitimate which means they're more likely to want to follow them people don't like to follow rules that aren't set through fair processes or processes that are at least perceived to be fair and this in turn makes this super important for the global policy conversation around social media i would say one of the most fundamental drivers of the desire across the world to reign in social media platforms is precisely the feeling that they are because of their scale setting rules that are affecting the flow of information in society and that they're not setting those rules

through a legitimate process and so if you can implement some kind of good governance so that you do create those rules through a fair process it might actually be very good not just for society but also for the business okay so and then how i guess the question here then is how democratizing can help with this in a different way than it helped or at least i argued it helped with the hold up problem so i think there's like an old and super intuitive idea that a system of rules is more fair or more legitimate when the people who are actually subject to the rules participate in creating them that's kind of obvious and there's a more subtle point here which is okay fine why don't we just

have the democratically elected governments write these rules for us which is obviously a huge part of kind of the movement in regulation today and i would just say there's some advantage i think obviously they could be complimentary but there's some particular advantage to having these user communities themselves write the rules if you can get them to do it and some of the reasons i think this are first of all these communities tend to be spread across many countries it's not actually obvious how any particular country could write these rules the people who participate in these communities whichever one you want to pick are different from the people who participate in real world government so

if you try to use one to write rules for the other you might not get very good outcomes and you might not even get very legitimate outcomes and related to that these are weird nuanced contexts where writing rules is very difficult and it's actually a little bit hard to believe that someone quite removed from the communities is going to be as informed about what would be a good rule and what would be a bad rule so those are some reasons why we might want this i guess i'll call it community governance to be at least one part of how the rules for social media are built okay so those are like just high-level conceptual views on two reasons you might want to try to

democratize large platforms and i think the question is how do you do this in practice so we talked a little bit last time about some ways you can borrow from the history of democratic governance to kind of innovate in web 3. we talked a lot about kind of the need to move beyond direct democracy and now i just want to kind of lay out some ideas for how we might do this both for web 2 and web 3. all right um the first thing i'll say as we like apply this in practice is just to note that one thing i see over and over again is that it's hard to discuss decentralized governance for large

platforms of people because there's no common shared definition of what the word or the phrase decentralized governance means and people use it to mean totally different things it's actually quite complicated to figure out how to apply it to a large platform so i want to tell you about this is at least one idea for how to divvy up the different things it can mean so one thing you can do is you can have a platform it's centralized in the sense that the rules apply to everyone the same rules apply to everyone but you could use decentralized means or democratic means to set those central rules and that's kind of how like the federal government in the u.s works or the central government in many countries

works right so so it's centralized in the sense that it's a set of laws that apply everywhere but it's decentralized in the sense that it's not one monarch making the rules and so some people might call that decentralized governance other people might not it's definitely a super important thing for how platforms operate in practice the second thing is who enforces the rules and so you see lots of examples of this too where maybe the rules are actually set by a central actor but how they're enforced is decentralized so you have these communities where maybe there's a set of rules like reddit for example has a set of central rules and they partially rely

on admins to enforce those rules in certain contexts you can imagine pairing these two together you could use democracy to set a set of central rules and then you could also ask the community to help enforce them or you could do some other combination of the two this difference between making the rules and enforcing the rules is is super important um and matt and i talked about it earlier like they're both it's actually kind of hard to figure out which is more important in which context in some context you might think that making the rules is actually the easy part but getting it right on enforcement is super hard and so you see for example i mean just to draw my own experience at meta lots of times where the concern is not about the rule but about the fact

that it was applied to this person and not to that person and the more you have that kind of sense that the rules are being applied unevenly the less legitimate the system obviously is and just from a practical standpoint the worse the user experience is especially imagine you you could have some perfect set of rules but you massively over enforce on them and it makes for a terrible user experience something like that so that's one one way it plays out but it also plays out the exact opposite way in fact it might be an interesting exercise if you're interested in this kind of stuff to go read the actual rules of basically any platform i think you'll be hard pressed to say that you think that they're written super well they're incredibly confusing they tend to be at a almost bizarre

level of detail like some of the you should a good example i don't mean to pick on them at all because it's completely not unique to twitter but like go look at twitter's rules around manipulated media which is what they use for example to apply to the um the hunter biden piece in 2020 it's an extremely weird it's not just like there's a rule it's like under these conditions we do this but under these conditions we do that and it's like if you turn it into a flowchart of an extremely complex flowchart and a lot of them have this characteristic that's in large part because it's just a super hard problem to write a coherent set of rules um but it might also be you know point to some pathologies in the way large platforms write down their rules and so in a case like that

the problem is not just okay we ought to enforce these rules and we're not very good at or it's a hard problem it's also like if the rules themselves are super unclear then that's also very bad for users if they're getting enforced on and they don't understand why even if the enforcement is actually done correctly under the system of rules or something like that and so these two problems go together and i think both are amenable to sort of some of the insights from democratic governance that you might be able to use tools from democracy both to write more effective rules and also to get the community to help you enforce them so that i think that's that's kind of where my interest lies and if you think about all the examples

i put up at the beginning i think it kind of applies to all them so if you even think about openc or coinbase these are tools that might be useful um to platforms that are kind of like web3 adjacent but which still require some kind of central layer of rules and regulations i'll just mention because it's super important it comes up a lot when i talk to people there is of course this other meaning of decentralized governance which is like you kind of get rid of the platform and it's sort of like a we don't see any examples of this in practice yet but it's a good dream some people have that you'll have like a completely decentralized social media platform where there are no central rules no central anything and it's on the

blockchain let's say and then people can write their own competing kind of like wrappers that sit on top of it and some can have one way of ranking content some can have another some could have one system of rules some could have another system of rules and then you'd have competition among users on which to use and that could have huge gains for both the economic pro the hold up problem and the legitimacy problem potentially if it worked but i think there's some very interesting questions as to whether and how it could actually work and i'll just point out that under a variety of circumstances like if the returns to scale and the ranking algorithms are really higher the returns to scale and the moderation programming is really high you might end

up with aggregation and basically have all the voters end up hurting into one service or something and then that service would have all the same exact problems of of the web 2 platforms because it would be the service that everyone's locked into uh and it would have to decide what's allowed to be said and so forth and so i think it's a super super interesting area and something i'm interested in but i think in the short run at least solving these sort of like how can you democratize current platforms is a is a very helpful first step to think about okay all right so how do we actually do this the obvious first thing to do is to have the users just vote and so maybe you uh you write the policy so

this would be under this framework of central rules but voted on by people and centrally enforced you write the rules and you make two versions of them and then you ask the users to vote on them i think it's kind of an intuitive uh simple idea that brings at least some amount of legitimacy into the process of how let's say a social media platform writes his rules and this isn't just hypothetical this is something that was actually done i don't know if anyone here remembers this i don't um though i was a facebook user at the time this completely passed me by um but facebook actually tried to do this it was a super long time ago 2009 like 100 years ago um but that's exactly what they did

they there had been a big backlash to some of their new community standards so they said you know what we're gonna write two versions one that's kind of like what we've been thinking about one that's more responsive to the backlash and we're gonna let the users decide and not only that but on a going forward basis that's the mechanism we're gonna use to to update these um terms in the future and so at the time this was kind of a big deal and i think one thing that's kind of remarkable about it is that actually the person who led the revolt against the proposed terms came out and said publicly i think this is a great idea this is how rules should be made and so there is this kind of like brief moment in time where it seemed kind of

like a one of these tech optimism moments that was like god this is how we're going to solve the governance problem for social media we're going to give power over to the users it's kind of similar to some of the rhetoric in web 3. it fell victim to precisely the same problem we talked about last time in terms of dow governance which is that in the end basically no one voted no one actually kind of thinks 600 000 votes is sort of remarkably high based on my expectations um but it was a 0.3 percentage point turnout rate for the facebook user base at the time and i really like this is a headline from los angeles times facebook governance vote is a homework assignment

no one did which i think fits down governance pretty well too and so what happened here i mean there's a lot to debate about exactly why the participation rate was so low but it's not that surprising that if you ask people they're on this platform for other reasons i think a lot of them rightly feel it's not their job to govern this platform they're not getting paid to govern it um why should they take the time to read these super complicated proposals and choose between them and so we talked last time about the the paradox of voting rational inattention i think it's sort of natural that this is what would happen if you just went straight to kind of

like a simple vote of the entire user base of really any platform um and if you and if you do if you're interested the wayback machine has all this stuff you can actually go read the proposals and they're like very technical and the differences between them are quite subtle and so there's actually this like third document you had to read that was like the differences between the two documents i was like no one's going to put in the time to do that i imagine of these 600 000 voters how many of them actually read the documents who knows surely less than 600 000 long i i want to say it's like 40 pages each or something like that how did they arrive at this like did did they think this was going to work okay so actually that's a very good

question because that's an important part of the history of this they were quite wise to put into the initial procedural promise that they provided to users that they were only going to implement the vote if at least 30 of the global user base participated so they clearly knew there could be they were smart enough to know there could definitely be an issue and they were right i think that they thought at the time again there was a lot of optimism around this stuff at the time um i did think they thought they might make it to 30 percent um but they were at least anticipating that could be an issue and the funny part is af in the aftermath of getting the 0.3

percent turnout rate they actually decided to implement it anyways and then they proceeded to have several more votes and the participation rate actually doubled on the next vote to 0.6 but ironically the vote had to do with like turning off voting for the future proposals and so the system died after that point um but the interest in the idea has never really died and it's still very much around today and it's actually funny to talk to people about it today because not everyone remembers that it happened in fact even at meta the vast vast majority of employees didn't work there in 2009 and so it's actually a really interesting piece of tech history that's

not known to all the participants anymore there is of course a much more cynical interpretation of that whole situation because they knew damn well that they were going to get 30 of the book and it sort of was a pr stunt that's definitely possible certainly in general i think that companies often face that incentive to make a show of listening to people the only thing i would say and maybe this is even more cynical is like i think there are potentially large advantages if you could actually get the user base to make these decisions so this is something you'll see in the press today is so sorry at meta for example mark still today speaks about the desire to

get the community to govern itself and the way i think more critical elements of the media interpret it is not that it's that kind of cynicalness but that rather of course someone who has to face all these hard decisions would much rather push them off onto other people and so it could go either way like if you could get high participation it might make your job as a platform easier in some way um it could yeah it's hard to say though of course the voters could make bad decisions and you could end up having to bail them out it's hard to say between

but like some of the things that like i can imagine there being some things that people may be concerned about like the i'm going to answer a different question and then i'll answer your question my first the first question answers are there other historical analogs to this and i think you're pointing out exactly the balancing act that we see in the history of democratization which is there's some you know in the literature they're called like the elites

um someone has a bunch of power and resources in society and they realize that it's in their interest to extend the franchise because of these dynamic commitment problems i was talking about and so what do they do like in equilibrium the bargain they offer is like just the epsilon amount of power they have to give up to stay in power and so about they don't want to give up too much because then people will vote to you know redistribute and whatever and so they give up just enough power and you definitely see in all the discussion around these sort of voting systems at large platforms you see a balancing act where there's kind of like how much could we give up before yeah either we get decisions that

are bad for the business or also as you pointed out like there's some things we just can't give up legally because we're required you know aml whatever whatever law you want to pick and so there's very much that same balancing act in terms of analogs to companies doing this there are lots of examples where companies do something that hurts their immediate bottom line because of some long-term commitment value so you see tons of this i would i would describe lots of self-regulatory regimes like this where like you know um like nuclear producing energy companies get together and they agree on a set of standards that are actually costly for

them to implement and so it hurts them in the short run that they have to bind themselves to this organization's rules but in the long run it makes people believe in the nuclear power more and they make more money or something like that you see this is like you know emissions agreements there's all these kind of like self-enforcing agreements where companies come together they're not democratizing they're typically giving power to some kind of independent commission but it does have that feature of it's costly to them and the commission does tend to make decisions that are costly for the immediate bottom line but there's typically a belief that in the long run it's valuable and so you could you could think about whether the same thing is true here and that's actually one of the major issues i want to raise

so just to get to how would you actually like implement this this is super early thinking i want to stress but like you can imagine a platform issuing some kind of governance token to its contributors to its developers to its creators what have you and to your point that it could be that it's very carefully specified what these votes are so it might not be like you get this token and you vote on everything it's like once a year there's three take it or leave it proposals you get to vote on and they're only in these issue areas and there have to be some kind of balancing act between if the voting power you give up is utterly symbolic then you get none of the gains in terms of solving the hold

up problem but if they're if the votes are too important you might accidentally like destroy the platform you hope that there's some balancing act because if you get the incentives right the developers who hold the governance token shouldn't want to destroy the platform but it is a tricky incentive problem because of course they have their own private interests and depending on their discount rates and pivotality and a bunch of other considerations they might conclude it's in their interest to vote to take a bunch of money uh you know to like imagine that votes on future fees on apps they might vote to make the fee zero because in the short run that's very good for them even though in the long run if the platform has no revenue they're screwed so there's very delicate balancing that has to be done there um

but i do think that basic idea is kind of interesting it's like what if you gave out these tokens what if you put them on the blockchain so that it's especially credible that the company can't mess with your voting power could that help to solve the hold up problem that's kind of that's like the idea of course it could also be used for the social media problem you could distribute these tokens to users ask them to vote on stuff then the key question becomes well we know people won't vote so what do you do to get them to vote some set of ideas is you could use the token right to build in some kind of direct incentives to vote um you could also use other things like delegation so i think delegation we talked about last

time might be a way to help mitigate the participation problem so you can imagine taking everything i just said and layering on delegation so it's like we give out these governance tokens to the key stakeholders of this platform uh and they don't just get asked to vote but actually they have the option to kind of like essentially elect a representative who will make decisions on their behalf for some period of time and if we make the problem very simple and make it very easy to delegate it's possible that that could improve the participation problem yeah so i mean it's in this part of the talk we're snaps trying to kind of talk yeah fundamentally different you know

just sort of the design space doing building in one versus the other the context with developers or stuff like that i'm envisioning a system with known identities um as the easiest use case where there's even there there's a bunch of tricky problems to solve i think it gets even more interesting if you open it up and you say anyone can come and trade these tokens and also develop and be anonymous and there i think you immediately get all these all the same problems we're familiar with from web3 governance so not only sybils but also governance attacks where you worry that someone comes in and buys up a bunch of the tokens to like destroy the platform which i think is super interesting

so far my basic idea has been you know if you have a platform with a set of known important developers or something you start by distributing it to them and maybe you make it non-transferable again to start i do think there's some advantages to going all the way to transferable especially one thing ethan and i explore in the piece is like it actually has this cool feature that if they're transferable if they trade then if the platform has a strong private belief that it knows what to do and that the voters are wrong it can buy out the voters to implement the idea which is very similar to corporate governance in a cool way um i guess one point here is that even

right so in the known voting case or at least this still enables weighted votes i guess yes which i guess like you say is more than corporate governance yeah yeah so i think i we might have touched on this only just very briefly in the last talk but i think one of the most low-hanging fruit open questions in decentralized governance is just like what are the conditions under which you prefer token-based voting weighted voting what are the conditions under which you actually want like one person one vote very loosely we see in society that there seems to be a view that like when you're deciding normative or societal issues you prefer one person one vote on business issues you like the idea that the person with more at stake gets more

say but i haven't seen anyone lay out like a really clear set of models it's like this rationalizes why you want this year and that there my intuition is for at least for the hold up problem case the weighted voting is important like you want the bigger developers to get more say i would think um but i haven't seen that actually like super deeply formalized though i suspect out in the corporate governance literature they're probably something okay i want to mention two other mechanisms that i think are super interesting um so one is sortition you could call it a citizen's assembly or in this context maybe a user's assembly but there's a there's an when when you know participation is going to be really low and that people aren't going to become

informed there's actually a very logical argument for why you might prefer to create a randomly sampled set of people then the reasons for this are twofold threefold maybe one is that by randomly selecting them you break the serious adverse selection problem you have where if participation is very low and it's something that most people aren't interested in then who are the people who choose to participate they're probably very very unusual anyone who's ever read through any notice and comment comments either from the federal government or actually meta publishes theirs from the oversight board you'll see that the average comment is not from the average person it's from a very

unusual and very strongly people with strong points of view tend to participate and you might not want that for your system if it's supposed to be representative and so by actually randomly sampling people you might get a more representative group and then the idea of sortation or at least startition as i'm proposing is first of all you pay them to participate so that you break otherwise you could have this attrition problem where you've randomly sampled at the beginning but then most of them drop off because they're not interested and once again you have this unrepresented group but if you pay them you can get a representative group and by paying them you can also convince them to really pay attention and study the issues and so you can actually and and

countries do this now with some success ireland in particular has done this quite successfully you randomly you get basically like a jury you pay them for their time and you ask them to do this service and to study an issue experts come in and brief them on both sides and then the group makes a decision i actually think this is a way more realistic for the social media setting i think this is way more realistic as a way to get democratic input than to try to put it out for a vote because it's not just that it's representative it's that you get some informedness through the briefings of course this is expensive to do you wouldn't want to do this a lot but if you're a large platform operating at scale and these governance problems are severe it might

actually be worth it to pay for this some in you know in limited set of cases yeah um so there's two let me answer them a couple different ways first

there's a big difference between an actual jury that is adjudicating a dispute that's like one particular thing and then in sortition the way i'm describing at least it's more about like creating a policy and it's very the reasons very different is that in the jury case well there certainly are some values you can bring to bear we tend to think there's like a correct answer and do they get it or not within a citizen's assembly you're like you're trading off different values and so it's not obvious that there is a correct answer to compare it to there are however like in dish of did people do this in a you know in a well-meaning serious

sincere way did they put in effort did they participate and when people have studied that the evidence is pretty good that actually people are surprisingly willing to hash things out and to put aside their preconceived beliefs and stuff like that but you can't really correlate it with any any particular outcome on the jury side you could do you can do that and so you could ask a legal expert how would you decide this case and then you can compare it to the jury i think the consensus view there and this is kind of nuanced is no the jury is nothing like one of the experts and litigators i'm married to a litigator will talk about this a lot that the jury is just a complete crapshoot but there's a very subtle thing that goes back to legitimacy that's really

important which is that legitimacy in these kind of collective decisions is not imbued exclusively or even primarily by a view of whether the jury comes to the right answer but rather by the idea that like this is a fair process obviously if it's a complete like made-up coin flip that's not legitimate because it's not even like a story but if you had an expert who was super good but they made all these decisions unilaterally i think a lot of people would prefer the jury to that even if the jury is a little bit loosey-goosey with the law because it seems more fair and this is something ethan and i have been have been studying it's this idea

of process legitimacy that when you're when you're dealing in situations or you're making very difficult and fraught decisions from the perception of people who are subject to the rules it's often more important to them that the process is fair then the outcome is what they want and so you and and so i think a good test of whether one of these systems would succeed would be do people agree to abide by the decision even if they disagree with it because they thought the process was fair i don't know that sortation would get there in a super fraught environment but it would be worth trying all right the last one i'll talk about which is quite different from the others but i think complementary to them is this idea of independent bodies of

experts it's actually super connected to what we're just talking about and so so facebook has done this with the quote unquote oversight board it's a group of experts they're compensated through a like unrevocable trust and so they're in some sense independent from the company though people certainly debate exactly how independent they are and they make decisions binding decisions on a small number of disputes so like facebook has decided to remove a piece of content someone appeals and then the board decides should it go back up should it stay down or vice versa they now they do it in both directions and so how does this connect to decentralized governance i mean it's like it's almost like a very small but potentially important amount of decentralization

where you're still making central decisions for this whole platform but instead of the ceo making them or people making on the ceo's bath this other group of people who are paid separately are doing it and there's some idea that like that independence is more fair and so it brings some amount of legitimacy and obviously that legitimacy is kind of a function of well how many decisions do they make and how important are the decisions and people certainly debate today where the board lies on that and then lots of people have used a similar model and so spotify recently announced this for example an advisory council so i would say it's less independent than the oversight board it's it's directly funded i believe and it's not the decisions aren't binding but it

still has this feature that it's like a different group of people being brought in um to make this decision and so i think just in terms of like how these methods all fit together where it seems like we're heading i don't know if there's a good thing or not but like basically reinventing all the different branches of government within the governance of these large platforms so i think it's a really interesting space for web3 because it suggests a bunch of other ways to use web3 that we haven't really thought about yet not only in terms of voting not only in terms of delegation but even in terms of sortition or even in terms of like you know you could imagine giving governance tokens to a small number of independent experts depending on the use case

okay so that's that's all i have to say i know i'm almost out of time just to mention here like next steps i'm used to giving talks that are much more like research i've done this is a new space i haven't actually done most of the research here yet but i'm starting a bunch of projects that i would love to get feedback from you all when they're done uh in this space so one of the things i'm most excited about is i'm going to try it could be kind of expensive but we'll see what we can do to launch a survey of platform participants so so developers creators and so forth across all the major platforms to track over time their views about you know how locked in are they what are their outside options do they think the rules that they're under are efficient or fair etc so that's like one

part of this um ethan and i are exploring a lot of sort of like how can you build the right theories of platform governance so what are the incentives what are the strategies and so forth uh and as i talked about last time i'm building this um with dan von air we're building this um sort of like coherent set of data on how these decentralized governance systems are actually working in web 3. and then i'm also you know working actively to try to pilot some of these ideas both in web 3 for like new platforms but i'm also super interested in this idea of like weaving together the web 3 with the existing web 2 platforms that are out there so thank you very much i'll stop there and take any last questions so

correct me if i'm wrong but it seems like the function that we're trying to optimize is the number of users that think that the process is legitimate or fair and if that's the case then i was wondering about when you put something up for voting uh your proposal people are kind of who object the proposal is less likely maybe to read it and vote on it because it might not be yeah accepted at the end so why waste the time so i was wondering about that whether you thought about kind of being proactive in your employ kind of deployed the proposal first and then those people who are strongly object with their proposal are much more likely to raise a concern and then based on

kind of you can then survey the people seeing whether the number of people will think that the system is legit now has decreased uh yeah and then rolled back yeah that's super interesting i think that that raises at least two thoughts for me yeah so one is um there are interesting processes that kind of look like that and in fact in the 2009 referendum i don't think i mentioned this but it had a characteristic like that they didn't do the surveying part which seems key but they did do a whole period of notice and comment prior to the vote so they kind of like they had an initial proposal then they got a bunch of feedback and they iterated which is something we see in in web 3 a lot with these like improvement

proposals and some of the dows have quite baroque processes for how you go from like the initial proposal through the amendment process to the final one and so layering on to that some kind of like real-time sense of how people think is very interesting the second thing i was going to say that i thought you might also be getting at i'm not sure is like that you could actually have a very well functioning process but if it doesn't move people's perceptions then it sort of doesn't matter and i think this is like a big problem um and something where if you're not like pushing out information and engaging in effective ways you could end up with like a great system that actually in practice is

like democratically justified and is bringing in participation and is informed is making good decisions but if most people don't realize that and they might dislike the system even though it's working well and that's a big like pr problem for democracy which we also see in the real world too and so there i think there's some interesting questions for very large platforms we can afford to do this at least how do you like get the word out about what you're doing um i think are both super interesting yeah um um

although like you know what happened everyone get attention um but i was wondering if that is like kind of commonly accepted conventional wisdom that if your baby will be there so i think you're exposing at least two things there that the first is sartician is the worst possible system you can imagine if what you're worried about is accountability between the decision maker and the voters because there's zero formal

accountability like there's nothing the voters have no control over whatsoever and this is why it's never been regarded as a particularly compelling way to govern a society because you would worry if on an ongoing basis other than aristotle who seem to think this is a good idea most people seem to conclude this is a really bad idea not only because you could get people who have no no idea what they're doing and didn't even want to become politicians in the first place but you now have given them no reason to fear that they could be fired or anything so if you think accountability is very important uh this is going to be the worst possible system for some of these value-based judgments it's a one-off if you look at how citizens assemblies work in practice they're almost always a one-off decision made by a relatively small group of

people where you can hope that some kind of social glue encourages more responsible behavior that leads to the second thing about whether compensating them helps or hurts i don't think that's at all settled you can think i mean there's lots of interesting work on other cases where compensating people actually leads to worse outcomes there's like the israeli daycare finding although i think it's actually been questioned whether it's a real finding out but but it's the example everyone points to where like they start to find parents for being late to pick up their kids and then more people are late because they interpret the fine to be like this is an option i can pay and get and be late um whereas social pressure is actually more effective at getting them to be on time it's a stylized story it's probably

made up of people are super into it um but that is that's definitely a thing that could happen like it could be that paying actually sends a bad signal that this isn't your civic duty um for a platform there's so little sense of duty that i think it probably is important to pay but there's definitely trade-offs and i can imagine in a dow where there is a strong sense of civic kind of duty maybe maybe you don't want to be there so this is another area that i would

love to study more is if you go around i almost put this on the list of things for discussion topics then because it's so popular these days there's all these ideas around slashing or other techniques that try to build incentives directly into voting and devoting a particular way and i think the promise of these is very interesting that you worry that at you know baseline people aren't putting in a lot of effort let's give them some incentives the problem is it's a super dangerous game to play because if you don't get those incentives just right you produce these incredibly strong incentives for people to start ignoring their private signals and vote it becomes a coordination game where the only thing you're trying to do is the voters figure out how other people are going to vote so you can vote the same way so you can get the reward or avoid

being slashed and that's terrible for most voting systems and so i think it's an open question like is there a way to build in those incentives without creating the really bad hurting tendencies oh i had a question go for it um i put in the chat i guess i can read it out um what do you think about trying to make systems forkable as a way to create legitimacy instead of creating governance rights l-ones don't need to explicitly give voting rights because users can run their own rule set this doesn't apply to most applications however for example uniswap can't fork permissionlessly because the assets the contract holds can't be split yeah yeah this is this is a super deep issue that i'm i've been talking to people

here this week about because i think it's one where there's a lot more to study but basically i think your your question is it's at least exactly what i've been thinking about that the stronger your exit option is the less important all these other more complex forms of governance become like if it's really easy to exit then then your obligations or your opportunities to use democracy to to do better are much more limited i think and the interesting thing tends to come around at least for these web 2 platforms when there's a sufficient amount of lock-in that that's where you start to get these legitimacy problems and it's where you start to get the hold-up problems i will just say in web 3 the story might be slightly different

only because you're using democracy there not only for those goals but also for these goals around like keeping the protocol decentralized and and securing it from governance attacks and the like where you might have other reasons to want to use democracy even if you can also exit but i think for these legitimacy or hold up problems i think though i do want to do some more research on this that um that the stronger the exit option is the less you need that stuff and that's definitely been the there's some very old sort of foundational political economy work about like well why are local governments so much smaller um and less developed than than national or state governments and the idea has always been because people can move from

one city to another the cities don't have the same kind of lock-in and they have as a result much more limited governments um so there is some interesting analogs there too yeah it's a really interesting question cool thank you all