Exploring Cluster Mapping with Joel Miller & Umar Khan

Transcript

[Music] welcome to the green pill. network podcast if you're just joining us we're building a coordination a network Society of thousands of hackers dreamers and doers focused on using crypto to bring positive sum digital systems to the world this podcast features the people who are doing it and is published every Tuesday and Thursday at least if I don't have a schedu in snap view with the guests which I had this week so you're probably listening to this on Friday or Monday if you want to learn more visit our website at Green pill. Network where you can download the green pill book for free join the Discord or become a member of one of the local green pill. Network chapters my guest today are Joel Miller and Umar Khan Joel is a researcher and PhD candidate who's been working with gitcoin for the last

six months on Ed which stands for the quadratic experiment and development program and Umar is a data scientist on the program workstream which is formerly known as the public goods funding workstream on the podcast today we are talking about cluster mapping which is a new mechanism for doing cibil inclusion resistance and we're going to talk about how it was rolled out on gitcoin in gitcoin Grants round uh 18 which just happened so basically Joel has written a paper on cluster mapping which you can access a link to in the show notes we also did a greenfill episode just on cluster mapping about nine months ago with Joel uh TVR it's basically a way of dampening collusion between participants

that are trying to siphon off the quadratic funding matching pool and uh uses some really fancy and also powerful algorithmic tricks in order to do that so I think this episode is for you if you're into following gcoins Evolution if you're into collusion resistance and siil resistance which are very Niche but also very important parts of mechanism designed as we're trying to figure out public goods funding in the post crypto internet so um this is a fun episode because it Blends the theory of cluster m mapping the leatest latest in mechanism design research with the practice actually deploying it on getcoin grants and I think Umar represents the practice uh Joel represents the theory of what's being

done here and I'm sort of playing host but I also jump in and sort of add to the discussion from time to time on the episode which is kind of fun to play both po uh host but also uh founder of getcoin and have a lot of this stuff in my head so the three of us have made for a good and educational conversation about cluster mapping at the frontier of mechanism design before we start the episode quick disclaimers I'm a founder of getcoin I hold GTC getcoin does work in the ethereum ecosystem I also hold eth and other tokens in the ethereum ecosystem but I'm doing this episode because I think cluster mapping is one of the most powerful mechanisms for doing collusion resistance in the future and it's important to highlight the work of the people who are doing the mechanism design here so this is a fun

episode if you are into mechanism design public goods funding quadratic funding or gitcoin without further Ado I give you Joel in Umar enjoy Valle Dow is an open Global Collective working to finance and democratize the governance of synthetic biology Technologies to protect the future of our planet research and progress into synthetic biology is critically constrained by the world of nation state regulations and bureaucracy from middlemen but the potential that synthetic biology has for the world is massive it can produce a world in total harmony with nature we can grow stuff using cells like software it's awesome and the best part is we no longer have to pull carbon out of the ground to grow and maintain civilization don't you think we should be funding this Valley da does and that's what they're here to do to try and make the

solar Punk World of Tomorrow a reality today so join Valley Dow right now and save the planet at Valley dow. bio hey Umar hey Joel how's it going I'm all right great it's uh raining but it's fine great to have you here excited to talk about cluster mapping we did an episode with Joel on cluster mapping before and now we've gone from Theory to actually seeing it in practice it get coin grants and so wanted to create an episode about that and uh what it means and why it's important and what's the the future of cluster mapping so uh Joel maybe I'll kick it over to you first what is cluster mapping sure so cluster mapping or cluster matching depending on

which uh Forum post you're reading by the way quick aside that's my bad for letting the names get out of control I'm you know I'm all right at doing math I'm not good at at creating names for leave it to the meme Lords over here to figure out Joel the big rins anyways cluster mapping uh matching whatever you want to call it is uh a kind of mechanism that builds off of quadratic funding which gatecoin has been using for years a mechanism that we all know and love and tries to tweak quadratic funding to make it one uh kind of more resistant to civil attacks and collusion and two kind of from a different perspective better at really kind of capturing this idea of social well welfare maximization uh in

the presence of normal human life so I know that that second explanation is a little bit more like head in the clouds so maybe I'll just like try to describe the Civil resistance collusion resistance stuff for you guys so essentially you know quadratic funding has dis built an assumption that all agents in the world are like totally isolated from each other right this is like a super super classic assumption in economics right itic funding kind of comes from this like mechanism design economic space right and in that field it's very very normal to assume that you've got this almost you know Hoban world where everybody's out to kill each other not kill each other but you know

everybody's maximizing their own soci social welfare or everybody's maximizing their own utility you know everybody's very selfish and uh pretty much what you see in the real world is that um that doesn't really happen right we have groups of people that collude we have civil agents which are kind of like if you want to think of it this way groups of people quote unquote that uh collude very very very effectively right and so what happens is that this assumption that quadratic funding was making about everybody being super isolated and selfish just totally falls apart and it causes big big problems the long story short is that cluster mapping is a way of dealing with that problem Yeah so

basically I my sort of layman understanding of it is that cluster mapping is a way of measuring the social distance between different contributors and when they're in the cluster together they're socially similar to each other either because they give to the similar grants or they have similar attributes associated with their identity you sort of collapse them into one identity and and then what that does is um it basically allows the qf algorithm to upregulate cooperation across social distance so if I'm from Colorado and someone from Istanbul uh votes for the the same Grant then because there's a big distance between them then that's a signal to the mechanism that there's a

lot of value in that Grant that's using physical distance instead of social distance as a um as an example so cluster mapping is a way of basically clustering different contributors together by their similarity and then dampening their matching if they're similar and increasing their matching if they're socially apart how is that in terms of it from layman's terms no that's perfect yeah you're exactly right yeah and maybe I can also just point out that like this technique also incorporates this idea of silver resistance again when you think of sibles as people who are like really really really socially similar right like the exact same you know what I mean and so yeah cluster mapping is a way to kind of say okay let's get the Civil

resistance but then also even when you got two people who are just might be more similar or more different there's something we can learn about like how good a project might be from looking at the people that contribute and saying okay how different or how similar are these people one of the other just like in terms of playing name salad today one of the other names for this is also I think like Boral CUA and that name really appeals to me because I think the whole idea of just plalism talks about the diversity of communities and how there should be space for like multiple multiple communities and what cluster mapping and coral qf method seem to do is like make it so that the public goods supported by the most diverse communities receive the most funding

which is just uh really really great and yeah it also feels like pretty cool that we're at this blend of both pluralism and like uh radicalism in terms of like quadratic fundings uh s rate effect uh in terms of the amount of like the way the way contributions get matched it just seems like a really nice uh intersection where it's like not only do we want projects that have support for more uh more people to get support but we want the projects that have support for more diverse communities to get support I really like that because it just like it really speaks to the strength of the breath of how much uh a good supports people so let me try to do one rundown summary of cluster uh qf and then we can talk about how we've

integrated into gitcoin grants from there so basically quadratic funding is a way of having a central matching pool that is allocated according to the results of a crowdfunding campaign in which the number of contributors matters more than the amount get that they give in the in the crowdfunding campaign so basic basically uh if Joel has a grand that gets $100 from 100 contributors and I have a grant that gets $100 from one contributor then Joel is going to get like 99% of the matching pool which is really beautiful because it means that the median new contributor to the matching pool can give uh $1 and it can have tens hundreds or maybe even a thousand dollars worth of effect on the outcome of the pool but basically the the Achilles heel of all of this is what if you make up identities or what if I

were to go in and to attack the mechanism by telling people if you contribute to my grant then I'll give you a kickback or something like that so basically how do you deal with the Civil attackers which is people making up the identities and how do you deal with collusion uh which is basically people paying each other in their Social Circles to um contribute to the grants cluster mapping is introducing a really interesting mechanism for dealing with that in which basically you measure the social distance between the people that are contributing so basically um Umar if you and I contribute to the same grants and uh then we're the same identity but as soon as you contribute to a different Grant and I contribute to a different Grant then we become only 50% the same

identities um and so on and so forth so if I was going to collusion or cibil attack the mechanism in cluster mapping I basically have to give to a bunch more public goods in order to hide the fact that I'm collusion or Cil attacking the mechanism from there and it also has this property of elevating the preferences of people who are socially distant from each other who are dis who are different from each other so that's from like top to bottom what is qf and where does cluster mapping solve one of the main Achilles heels of qf yeah I think I think that's what makes it so attractive to me as well which is just like it makes it more expensive for the cbls and it gives a benefit to the real people um because you're now getting more matching because you're more diverse you're more unique one of the

really interesting you know like Joel is someone who's like really deep in the in the theory and uh you're someone who's been dealing with the Civil problem for years now have have really like explored the landscape of different possible solutions uh and I'm the new guy putting it into practice um and trying to see how well it works and you know one of the things I like to do is just like try it and see what the results are and one of the cool things was like you know this round you really pushed to just try cluster mapping and see how it worked because of that push and we did it we were able to like get all this great data on how it works um and what the community response is which is also really really valuable because like in whatever our future models of qfr I

guess like it really depends on like what the community wants them to be and and them like pushing us forward as well as us pushing them forward the two criticisms were is it really effective at squatching civils and then second if I'm a real person does this discount my matching amount one of the things that's become really clear from the data is while those are both Fair concerns to have in theory uh in practice um they're they're not really problems so like the first thing is like does it does it really SC sibles and like what we can see from the data is that it definitely does because when you Custer People based on their donation profiles one of the things that stands out is that the vast majority of people donating are oneof one clusters 80% of people are completely unique and Joel actually

brought up a great point to me about why this is the case uh in terms of the number of projects in the round uh Joel I don't know if you want to dial in on that for a sec sure yeah so the way that we clustered folks for g-18 was to put them in you know a cluster corresponding to like the exact number or the exact set of projects they donated to right so if you think about there being like say 30 projects right for each project I could donate or not so that's two options for each project so um the number of possible clusters you could be in is 2 to the 30th power which is like nuts that's really really big that's uh and computationally intensive as we as

we learned yeah it's it's so massive that like chances are if you're just a normal person doing normal stuff you're going to land in one of those two to the 30th spots that nobody else is in and that's like totally supported and backed up by the data like 80% of people are completely unique oneon-one clusters and yeah just like the explanation of the fact is there's so many different spaces that people can land in uh really helps make it clear that oh this this isn't that much of a problem but then the the interesting thing with the symbols is there is a small number of really big clusters where there's like four 500 doners in a single cluster they're all donating the

exact same L and those are the Clusters that really get reduced matching with with cluster mapping and it's honestly it's it's good to see that they get reduced matching because the size of our matching soite and you know because these are indeed likely civils or at the very least accidentally over coordinating people then what we're able to do is reward the projects that have more diverse support support across social distance because that fix some of money gets to get redistributed according to that and this to me is also better than the other alternative of of how to sort of work with sibles which is um de just like blackbox postp feling that just is uh not as transparent to

the I think in terms of how we do things it's so much stronger to have it as supp of the mechanism so I guess that's one thing is like 80% you know don't get the matching reded they actually get greater matching because of complely unique and then some people are are wondering do I need to change my behavior do I need to direct more projects and I think the answer for most people is no you're ready completely unique you don't need to change your behavior at all and at the same time if you are someone who's only donating to sve one project or two projects then you might want to consider picking up a few more that are of interest to you because there are fewer projects you're donating to the more likely it may be that you fall to the same clusters as other people but the

more projects you choose to select the more you almost verify your unique humity by what it is you choose what unique fingerprint you make in terms of the projects that you have and yeah I think that's really cool Umar if you can take us through end to end gg8 was a couple months ago cluster mapping is something that you built some infrastructure for I wonder if you can talk about that infrastructure and basically what yall built and how it supports cluster mapping definitely so in order to run cluster mapping what we need is really good data on beat Nots and the round in a format that is really manipulatable blockchain data is often messy and I think one of the biggest

value adds is is transforming that data in format that it's really easy to uh work with and so like one of the things that we had for GG team that we haven't had before uh is Grant CTL and slor region data uh which is a nice JY um and region data is this really cool database that uh came up from gr CTL which was super valuable and Wy you started yeah I'll maybe do a minute on this I'm G step out of the host and play guest for a second here but uh Grant CTL is basically just a way of extract transform load uh that's a common Paradigm in computer science it basically pulls data from ipfs from the graph from RPC calls from anywhere you could possibly want it and

puts it into a common postgress database and then that allows data scientists like Umar not to have to [ __ ] around with data normalization pulling it from a bunch of different places dealing with API rate limits and services being down just to have the data all in one place and be cleansed so that you can do what you do which is data science so um this is actually a pretty common Paradigm in web 2 people doing data analytics just like separating the concerns of getting the data in one place from doing the science and that that was the foundation upon which you could start messing around with cluster mapping so just 30 seconds on what Grand CTL and regen data is back to you umer yeah and uh it makes it much easier when I don't have to also play data engineer um and can just focus

on the analytics and Science and so like uh this was amazing and and shout out to uh Bliss as well who was uh uh doing a great build on this and so with that data infrastructure the next piece is the algorithm and this is where Joel was like really the key man um because Joel wrote the code that does the cluster mapping and so what we were able to do and he didn't just write the code for one cluster ma it was like six or seven different algorithms and so we were able to run this data through all of those different operates uh and compare the results and um just kind of do some like sense gut checking on which one of these are most uh similar to the results you would expect if we were indeed using the sibles that were detected um and so like

the one algorithm that performed the best compared to where we expected the sibl to be um was uh the one we ended up using for this round I think the code that I wrote kind of looks like the Cheesecake Factory menu or something because there's so many different little tweaks you could make to all of these ideas and there's so many different directions you could push like the kind of fundamental concepts that we're working with so yeah it's easy to just spin up a bunch of uh algorithms and we're coming so this takes Cil resistance on gcoin U which by the way quadratic funding uh is the Achilles heel is a very powerful mechanism but it's got two Achilles heels the one is one is the need to continually raise matching pool funds I'll just set that

aside for a second because that's his whole other podcast episode and then there's cibil and collusion resistance which is basically it's you get such a good matching R ratio as a new user to the system how do you prevent people from making up new identities or bribing people in in order to allocate the matching pool in an unfair way and there's an impedance matching between the size of the grants rounds and the amount of civil resistance that you need so basically you don't have to you don't have to protect against sophisticated attackers when you only have a $25,000 round right no one's going to spend 100K in order to allocate 25k so there's an impedance matching between the Civil resistance and the size of the rounds and they're going to grow over time as as it hopefully getcoin grants grows but

basically now we've got a multi-layered civil resistance approach you've got getcoin passport which is a privacy preserving way of getting people to reveal things about their identity so that we can have confidence that they're individual humans but even if they get past that if they're able to spend more in order to forge identities uh to get into the system now we've got cluster mapping on the other side of things where basically people if they want to attack the system Umar if you've got a grant that I want to make sure gets a lot more of the matching pool not only do I have to spin up a 100 different accounts to contribute to you Umar but each of them have to contribute to different public goods in order to make their identity more plural and so basically we're raising the cost of the attack just automatically using the

mechanism as opposed to having to go around and play whack-a-mole with all these different Cil clusters now the mechanism just filters out the Cil clusters and you're almost like using your opponent's momentum against them because in order to attack the mechanism they've got to give to other public goods on gitcoin um so basically it's like a way of I always think of like Karate Kid style like opponent like using your opponent's momentum to fund public goods and I think there's something elegant about that so I don't know that's that's my little Riff on on what we're doing here I'm curious if if that brings up anything for either of you it's like a little Judo maneuver like oh thanks for coming in fun public I would just G to say one of the like exciting things about this too is like you know the speed with which we

can see research coming into practice like I I think there's like two papers that kind of laid the groundwork for the cluster mapping we did one was the the doto paper with hja metallic and Glenn like a year and a half ago and I think um someone called this out R Ro called this out this speed at which like research can get translated into actual implementation of Technology we three is like really high compared to most other Industries um and then Joel wrote another paper which was like about a year ago partnering with lean Ericson who's also doing some amazing work at bigcoin about collusion oriented cluster matching uh and various uh techniques for being collusion resistant using quadratic funding um and that I I think

that's been out like less than a year and or or Joel you can say about talk about that way better than I can but like I just think it's cool that we can go from you know papers to uh having it in practice um this fast and also I just want to like shout out massive kudos to to Kevin because like if you weren't pushing on us to get this into practice much faster it probably wouldn't have happened this fast it definitely wouldn't have happened this fast uh so also just like appreciate you well I'm sorry Joe you're gonna say something oh yeah I was just going to say yeah it's been around 10 months since that since the second paper but yeah I'm super proud of how fast we've been able to push on this stuff and for me as somebody who spends you know I spend

most of my time just in the math world uh it's been really fun to see all of the little details that you really want to think about when you go from the math world to the real world right like all these little implementation things that actually end up being super important right and all of these other ways of kind of measuring success or failure uh for me have been totally fascinating to think about is one of the fun things about having done 19 rounds of Grants on gitcoin um is that we've entered this sort of Loop in which the theorists of which I would call Joel uh Eric uh Glenn wild GL all of Glenn's disciples I'm not sure if I can use that word in Academia

but uh the people who revolve around uh you know quadratic funding that the theorists now are in sort of the synergistic lock step with the people on getcoin who are actually putting this into practice where uh we're in the UDA Loop together observe Orient decide act repeat and so basically the getcoins really good at acting well not really good but we're getting there uh at doing qf rounds getting the data and then we observe it and then we pass it over to you and you help us Orient okay this is the way you can solve cble and collusion and then we decide together through the governance process and then we act again and and so there's like I my dad's a theoretical physicist and I know there's

this like relationship between people who are observers people who know how to run telescopes and the people who are actually doing the math models of the observations and understanding the physical phenomena and it kind of reminds me a little bit of that but with a different complex system a social complex social system instead of a complex physical matter system and so I'm really interested in spinning this UDA Loop forward with you Joel on on cluster mapping and doing it in such a way way that the the good guys the blue team is ahead of the attackers um and also in a way that's legitimate with with respect to government governance at gitcoin to do it um and I'm just curious if you all can paint a picture of where do we go from here what is the future of cluster mapping or disclusion and Cil

resistance that getcoin look like from y y'all's perspectives and uh maybe I'll go to you first on this one I think one of my big goals is to turn off post round seling if it's a black box today you know after the quadratic funding round there's this period of time where myself and and others before me have like gone through all of the data try to detect the symbols and filter them out and the methods we use to do that have been close Source completely private and because of that people have less visibility into this governance mechanism how we're all like like to be making decisions and I think it's really important in order for us to like collectively make decisions truly in a decentralized way where you can trust

that you're of a process that really your voice is heard and it's heard in the way you want it to be heard um we really need to turn off that post round fling so that's like one of my big rlls just so the audience knows what that is so there's a process through which um someone who's imbued with legitimacy by getcoin governance is going through and finding the Civil rings and kind of making a judgment call based off of the parameters given to them by governance about what a Cil ring is and then they present those results back to getcoin governance for ratification and and legitimacy and the problem with that is that it's socialware it's an assurance created by humans in somewhat subjective although prioritized decisionmaking as opposed to trust whereare uh things like

math and cryptography incredibly neutral mechanisms for squelching the civils that's the debt that's being paid down here progressively is is that right yeah definitely and a much clearer explanation and I think like what's so exciting here is that like now we have these two things that are making it so that we can get away from using the spell stream that we didn't have before one is like better qf mechanisms with Joel and like we've started upgrading this mechanism but we absolutely haven't stopped and I'm really excited for the the the work we're going to do on this next and then the other is the ever improving gitcoin passport which is rather than being this retroactive uh removing of sibles from the ecosystem it's a proactive removing it's this gate in front of the ecosystem that really

protects us from those bot accounts getting in uh and so I think like as these two things get better and better will be able to become a much more transparent governance mechanism that will make us all like collectively make better decisions and like more transparent decisions uh Joel next same question to you what's next I'm really excited to keep working on uh all this stuff and I'm especially excited to integrate um to see how we can integrate with passport more like Umar kind of alluded to I think that you know not every passport stamp is really indicative of like uh a social kind of attribute we might really care about so for example you know if you have got a Twitter account that might make you less

likely to be a Cil but if probably doesn't really say much about where you are in the social world right on the other hand like something that Eric and I have been talking about is kind of like focusing in on like maybe a more select set of stamps that might be harder to attain um but then can also serve as like really meaningful markers of social diversity and I think that using donation data is a really fantastic idea and I'm also excited to see if there's potential for using stamps and other types of data pretty much the mechanisms just aren't complicated enough yet we need to make them more complicated or make them more legible as they become more complicated um I feel like there's a trade-off space

there uh and we see it also within the ethereum protocol is that uh there's a trade-off between legibility sometimes and and what you actually want the mechanism to do and sometimes you find this really elegant very concise way of doing something that you want to do and it and that's the sweet spot so Kevin flipping the question back to you uh where do you want to see this guy oh no I just feel the shroud of me being the host of this podcast eviscerating around me um well so I mean I guess first off getcoins moving to a world of multiple mechanisms in the future it was just it was actually uh pairwise quadratic funding back in the 1.0 days but it was this giant monolithic codebase that no one could go in and change the mechanism because it was so monolithic and um and

so now we've moved to this decentralized modular ecosystem where you can sub out different things at least in theory you could sub out different identity vectors you could sub out different algorithms um and we're going to move from quadratic funding to a bunch of from if quadratic funding is a is a screwdriver we're going to move to a toolbox of having a screwdriver and pliers and a hammer and uh a bunch of different mechanisms so in mechanism design speak that's qf and QV and mechanisms that I think are on the road map are like conviction voting and insurance contracts and rank Choice voting and stuff like that and I see this this future in which we're going to have sort of like a toolbox of different mechanisms and you know maybe some people will prefer vanilla qf because it's less computationally confusing or

uh intensive and it's less confusing and or maybe they don't believe in all this plurality hippie dippy [ __ ] like they're not bought in on that and so those choose reg regular qf or maybe they'll choose a different mechanism but then there'll be people who are deeply like Glenn wild acolytes that are like [ __ ] yeah let's use cluster mapping out of the box uh on on getcoin so I think it's like there's an important choice architecture here in in that anyone using Grand stack or aloe should be able to choose which mechanism they they use and so a plurality of mechanisms and each of them having different values tradeoffs between uh Democratic and technocratic retroactive and proactive and there's probably a bunch of different vectors that people can choose of so I just look at this as one tool in

the toolbox albeit a really powerful one for allocating capital and funding what what matters in your community and so we'll we'll move from gitcoin innovating on um on one direction to sort of like a plural of directions around mechanisms and you know maybe eventually plugins to Grant stack and stuff from there but um you know I I I think that I I'll also add the caveat that there's a really amazing product leads and developers at gitcoin who are making a lot of these decisions these days back in gitcoin 1.0 I think I would be making a lot of the decisions just pushing it to production and the team would find out when I sent a tweet that the code was live and it's not that's not how it works at all anymore so like I'll just caveat my answer with that it's really up to the

governance and the product leads but that's what I'm excited about oh also Joel if you write a paper about this I'm excited to have getcoin mentioned in the academic literature one more time again so is that coming up uh it'll hopefully be coming up eventually I've got a lot on my plate but I mean there's so much interesting experiments that you could write about that I think are all really academically interesting and you know besides the kind of empirical stuff personally I think that this is a really interesting example of kind of like algorith algorithmic alignment being carried out in a very kind of like concrete way right like you know with AGI and stuff there's all this talk about like what's alignment how do we align stuff you know

like you said Kevin uh choosing to use plural qf or vanilla qf is really a question of values it's a question of what kind of things you want to value and what I kind of observed at gcoin from the conversations I had was you know there was a framing in which you could say Okay vanilla qf just does not align with our values as a community right and so we very concretely kind of chose a way to align this algorithm and it seems like it works which is fantastic right and so I yeah I think there's a lot of interesting stuff you could say about what's Happening Here on that topic I think like I was talking about mom about what I do and she was like why would you have an algorithm

making the decisions on where you send money rather than people like couldn't people make better decisions and one is like Mom we're Pakistani like if you look at the Pakistani government it's one of the most corrupt money stealing governments in the world and uh what what what's like really clear to me is like you want people to be making the decision at their local level in terms of what matters to them but at the global level we want to make sure that there's no corruption and the best way to do that in my in my opinion is a transparent algorithm or mechanism and so like less

money stealing corrupt government officials more algorithms and mechanisms that work really well and are hopefully like understandable to people I think will be and like who knows maybe somewhere down the line we into like AI governance I mean if I could but in here I would even push back on the idea that an algorithm is making the decisions in quadratic funding I mean think about voting just normal one person one vote voting right well that's really a very simple algorithm right and but is an would you say that an algorithm is deciding who gets elected president right when Americans vote for president like it's the people that are casting the votes right and so sure you know quadratic funding it's not like a robot that's

divorced from reality doing stuff it's just a way of kind of collecting everybody's opinions and doing something with it I think of it is almost like a govern it's like routing the opinions into um in a certain direction and doing it in incredibly neutral way I actually think that AGI if we ever had like an AGI governance world then you'd have the algorithms actually making the decisions and there wouldn't be a lot of legibility or credible neutrality within that algorithm and so um yeah I don't know I think of these coordination mechanisms as routers of our values into actual funding decisions that are uh credibly neutral and transparent and uncorruptible and I think uh so I don't know it'd be interesting to uh to see if

we can get your mom to come around and uh maybe experiment with the tech and then and then not only understanding it conceptually but also getting the muscle memory of it uh maybe maybe people start to come around to it on that topic I'm on uh and also so uh hi hi Carlos's mom there's a there's a meme at getcoin that all of the transparency reports that we published the only ones that actually read them are our moms that for the audience but uh they're they're proud of us how transparent we're doing and how much impact we're doing yeah definitely yeah but the Democracy I don't know that does also kind of feel like an algorithm even if it's just like a very simple one like who gets more votes but I guess with the AI it's like will it be listening to

people in order to make it decisions or will it make its decisions independent of what people are thinking they wanted and Laura I don't know but there's a there's a really good podcast on this one uh with Sam mman and Joe Rogan and the the first like 45 minutes in they're just mind well we're getting towards the bottom half of the hour what didn't I ask that y'all want to say and let's see who's good at thinking on their feet uh Joel I'll go to you first on this one well unfortunately you're such a fantastic host that you you asked absolutely everything that I want to talk about today um yeah I guess the only thing I really want to leave off on is um I really like what you said about

you know these different styles of quadratic funding capturing values and uh I just kind of want to raise my hand for the Viewpoint that uh yeah values are really important in Tech and it's a good idea to think clearly and closely about what you value and then try to implement that in your Tech and day-to-day life whatever I feel like I've talked to many people who see uh you know algorithms as these kind of platonic objects that are divorced from your opinions and your values but I don't think that's true so yeah that's just my suggestion to everybody not that you should listen to me of course but yeah are same question to you is there anything I didn't ask that you want to

say if you are interested in getting involved in this process in terms of like governance how we incr Q app how we Huns uh just reach out uh you can DM me on Twitter or on telegram con e um and you can also just hang out on the get Clan garet forum and yeah that's pretty much it and looking forward to St well um we'll have a link to the paper Joel that you wrote about cluster mapping in the show notes we have a link to the previous episode that we did on cluster mapping together in the show notes and this will be added to the getcoin playlist on the green pill YouTube channel so that we can follow the progression of the evolution of getcoin over time so this was a fun conversation Joel I think you're a total Galaxy brain and really excited to put

your work into the world so thank you both thank you thanks for having me Coen thank you and 's Mom debate me call no just kidding I'm sure you're really nice but yeah thank you again for having me yeah thank you very much thanks y'all all right peace and [Music] love [Music] we just heard from Umar and Joel I'm really excited about them because I think cluster mapping is at the Vanguard of mechanism design and collusion R sible resistance is an Achilles heel of quadratic funding so solving the big problems with Joel and Umar with in in

cluster mapping as they've deployed it on the getcoin network thanks again for listening to the green pill. network podcast again we are featuring the people who are building a coordination a network Society of thousands of hackers dream end doers using crypto to bring positive some digital systems to the world I think the mechanism design that Joel has done and the implementation at getcoin is an example of that if you value our work please remember to subscribe rate review it really helps people find the show thanks again and I'll see you next time peace and love