Quadratic Collective Intelligence

Transcript

foreign [Music] cool thanks you all so much for joining um I'm here to talk a bit about quadratic funding but specifically through the lens of collective intelligence so uh hence the title quadratic collective intelligence so um my name is Kevin Olson I'm a workstream co-lead of passport and Grant stack um and the git coin uh get coin Dao I should say my past experience is that some impact startups I did my time in the Wilderness at some Enterprise groups

trying to help teach them how to write some software and then a software engineer myself so maybe some of you know git coin through our grants programs or some of our products I wanted to just quickly uh share this if folks aren't familiar with it our vision at get coin is inspiring a world shaped by community-led positive change and our mission is creating Technologies and opportunities to enable communities to fund build and protect what matters so we've kind of shifted a little bit away from just being a grants only platform and hopefully you'll understand that at the end of the day what that means so um the problem everything is a capital

allocation problem I think kind of from our point of view um and where I wanted to sort of take this is uh around collective decision making um if you look at a grants program um you're often looking at a small committee who's trying to come to some kind of consensus or around where to allocate funds we specialize in public goods funding which often does not have a particularly good Revenue model so you don't have self-interested actors necessarily who are you know investing in things you're looking for other ways of arriving at optimal allocations dial budgeting is also a form of capital allocation anytime you have pooled resources and you're looking for where to allocate those contributor Awards perhaps you have some kind of a a bonus

program we've run a Citizens round ourselves ways of kind of sharing back with the community people who aren't explicitly you know earning from your your down necessarily how do you how do you decide what's a fair allocation in there retroactive pgf is another topic you'll see a lot of other implementations in the ecosystem Beyond quadratic funding and what we do but ultimately again a collective decision-making problem right uh treasury management perhaps you have a diversification plan uh for how to you know diversify uh your treasury or maybe you're even just an investment Club you know how are you arriving at group decisions all of these are collective decision making problems that would benefit from collective intelligence approaches so let's let's dive into that

a little bit um so wisdom of crowds this is a video it's very beautiful I promise I can't get it to play let's see um hopefully no sound uh cool um so this is these are starlings right these are a murmuration starlings it's called and it's very beautiful but it's not a great example of collective intelligence um a better one are the social insects um and this is often referred to as swarm intelligence I didn't want to put a bunch of ants up here I think that might be a little bit less relaxing to look at but what you'll see um why this is interesting is you often have a completely distributed group of individual actors who are arriving at uh optimal outcomes you know that an

individual couldn't arrive at right so the classic example is ants who are you know finding food and leaving a chemical trail back to the hive you have bees who are working together to produce hives themselves there's no Central architect or actor who's coordinating all of this this is all emergent behavior from sort of individual um you know actors or or agents within the system um so going Beyond just getting a little Echo here Beyond just swarm intelligence um there's a lot of other things that kind of are thinking of I'm thinking of as the collective intelligence pattern library that we should we should use as we investigate these um problem spaces where we want to elicit the best outcomes um so swarm intelligence maybe we even

kind of can think about that in terms of how we secure our own distributed networks right you have these self-interested minors who have been given a set of you know incentives and they follow along and you get this you know greater than the sum of its parts distributed um uh activity markets are a really interesting one um we're all familiar with that right uh price finding within a market is a form of collective intelligence the example I always like to use here is after the Challenger um disaster in 86 it took a congressional Hearing in the United States of you know a bunch of experts something like three months to figure out which of the vendors was responsible for the disaster right it was I'm going to say this probably wrong Martin Tia

call was the the producer of the the solid state boosters with the O-rings um but the markets have figured that out in about 30 minutes so if you compare what a huge distributed group is able to analyze and look at they were quickly arrived at that and priced that into the stock price of that particular vendor dropped by 12 and that stayed uh all the way through the Congressional hearings until finally the fines were levied against them collaborative filtering kind of encompasses a couple things you might be familiar with this from you know trending articles in a feed you know the idea that you have um lots of sort of digital dust a lot of behavior from Individual actors that's getting summed up to sort of point you at the likely most interesting content

in a system so content recommendation these kinds of things um and then you have things like deliberative democracy this is one where you might see things like um sortition or juries right if you think about how to arrive at a a verdict in a um in a courtroom at least in the United States um you have jurises of citizens right who arrive at these efficient criminal cases and that's a form of deliberative democracy where you're giving the decision-making power to not just one person not just a judge but to a group of individuals with the idea that they will arrive at an optimal outcome for uh for that particular decision one that I like to highlight too is cognitive diversity this comes up a lot where you have you

know the more sort of perspectives and backgrounds you can have represented on a team you often arrive at better outcomes right there's been a handful of studies in this space that shows that the more diverse a team is the sort of more uh Innovation as possible there and likely the less less likely to fall into sort of traps of groupthink and these kinds of things and then lastly is a sort of open collaboration in general right the more sort of open and information sharing you can have within a system you often arrive at uh positive outcomes so reducing the amount of information hiding and siloing within organizations can also give this sort of emergent collective intelligence so I'm going to step back quickly and talk about quadratic funding um just uh so you all know quadratic money there's a there's a link here so

WTF is qf if you want a interactive way it's sort of intuitively understand this there's a little uh sort of um thing you can play with at the bottom to give you a Hands-On sense of how this works but ultimately qf is the algorithm that get coin uses a number of other projects is using I think giveth is launching their qf around here as soon as you Griff back there so qf is good and everyone should be using it that's kind of maybe part of this talk um I'm getting the battery warning watch out I think I'm going to make it through the talk um ultimately what you're doing with qf though is your your treating donations sort of amplifying the ability of a funding allocation problem through direct donations from a community and

treating those donations like votes right but you're amplifying the you know the broad group not just the large donors so you're looking for so the popularity of a project through the number of donors within it it's I'm not going to put the the formula on here but it's the sum of the square root of the the individual donations squared um but what that looks like is um many small donors having more power in the system than a single large donor um what I like about this we actually have um our first research fellow Joel Miller who has been researching this and sort of Innovations on top of qf for us and his take was this I'm going to read it really quickly is that it's not just an algorithm quadratic funding qf is not just an algorithm that web through

communities like because it's appealing of is because of its peeling mathematical properties but folks also at qf for more intuitive reasons people say qf they like qf because it's egalitarian Democratic or empowering So based on Joel's qualitative research on the topic it seems to me that qf serves as a mirror for the community's values the community co-creates an implementation of qf sees that the outcome matches their intuitive idea of what should be happening and then adjusts accordingly for the next grants round so maybe it's a little bit about qf at scale what we've done I'm going to just quickly blast through this we have some great numbers but one of the most important things that I think here is the one in bold there is that if you look at all of the you know grantees from the past few years have come

through a lot of those are gone become very valuable companies all right and if you look at this idea of many many grantees coming through the system and the sort of get coin donor Community you know bootstrapping these projects these open source projects through donations and steering the matching funds at them uh the the total outcome is is a crazy number 28.2 billion market cap of these these projects collectively which I think is is worth really like highlighting that this is not just a vibesy feel good thing but there is some sort of larger collective intelligence that's happening here where really quality projects are getting funded by this ecosystem now the counter factual isn't possible we can't say what would happen if we weren't there so it's a bit

of a vanity metric but I do like to highlight that there's something intriguing and worth looking at I want to zoom in really quickly on our last grants round just maybe because we're going to get into the design space of quadratic funding and and where the kind of call to action is here across our different um categories these are the the different rounds that we ran you can see the amount of direct funding these came from Individual donors you have the matching funding this is what was raised and then the number of projects that hit what's called a matching cap so when they're sort of total payouts were capped at a certain point this is an individually configurable per round um and just to give you a sense of sort of the the numbers that led to this right so in a very large round like web3 open source

uh Leinster took the prize of 14k matched among some other projects but it got 29.7 000 donated directly to earn that 14k but then if you look at another project so for instance climate Solutions right where there's 350k same size pool on the line a much higher matching cap we actually didn't have anybody hit the cap the tree gens ended up getting 32 000 matched out of that pool from only 0.2 K so 200 something dollars worth of donations steered that right so quadrat funding scales in this really interesting way where you know of the smaller Community right only ten thousand dollars donated ten thousand you know sort of votes cast if you will from a much smaller pool of people still resulted in this very large outcome an

output so you have to look at that from the sort of like what is the the value elicitation of this group does this really match right it's not about just raw numbers but the actual sort of quality of the votes in the community that's participating um top line numbers 300K donated in total with 100 1.25 uh uh million in matching funds so that's just like a quick one so okay let's get back into collective intelligence here so as you're looking at this and you're thinking about designing a quadratic funding round or any kind of collective decision making there's a handful of things that you can you can play with like what are the sort of values in this this system you can adjust so um one of the most important things is

value elicitation so what is it you're you're trying to fund I kind of think of this as a fitness function so when the round closes how much did it match your expectations um one of the things we've we've decided is to group these things around themes right so you say open source the attack open source we've sort of defined those as communities that will likely have opinions about uh where the best funding opportunities are for the public goods that they want you might have something different right so you might have new projects versus sustaining projects versus something else entirely um though where you can play with these things are currently are sort of around configuration so those matching caps as I mentioned you want something where

someone could run away with a huge amount of the matching pool or do you want to cap it at some sort of Maximum output and then kind of go down to the next level and reallocate that so sort of more people sort of get a share of that pie it's not necessary by any means to set it up that way and as you saw with the climate round we had a very high matching cap and one project did get a huge amount of that that matching pool it's only like 10 of that matching people um maybe you also want to like think about the grantee criteria does this does this grantee the project or person have to be part of your community you're going to be anybody and how are you going to Define that so what are the acceptance criteria for participating um and then also like how long you're

gonna keep these applications open for is it is it continuous is it time limited to two weeks or something like that and then how long is the voting period And this is again when you're really sort of I'm kind of anchoring you on the get coin sort of view of the world as a quadratic funding round but there's other flavors we'll get into too which go beyond this next one I think is the most sort of interesting one is also ballot presentation when you think about you go to explore.getcoin.com and you pick your grants you know how are you making that decision right now it's very rudimentary right our last round is a beta round you know it was just really just proving the decentralized stack um kind of adds some more uh scale but you have some real options here and if you look back at what we were doing

in the past we were surfacing Things based on popularity so who was you know raising the most money in that round at that time or giving people estimates of their their matching which would sort of encourage you to donate to projects that were already sort of successful in the round your dollar would steer ten dollars versus a brand new project we hadn't received very many donations your dollar might only steer two dollars or something like this and and sort of encouraging various behaviors right so we were we were creating some kind of mechanism that was leading people towards very like certain kinds of outcomes right which was driving people towards uh generally sort of a popularity or an ability for a project to you know um muster its Community uh to uh to to drive their their project to the top

we've also experimented with um Community staking so projects could have kind of an open sort of address on a staking contract that people could come in and stake uh to indicate their sort of conviction or conviction votings another term you'll hear for this uh that this is a known good project you should uh probably donate it to it too as well as various kind of curation things we've had Collections and various techniques in the past that let people signal like this collection of three or four projects is legitimate and I support them um and you have people also like you know Carl and others who do some analysis that say these projects are very active get coin a GitHub profiles I'm going to donate to them over the ones that have less activity which is a form of an impact metric right so should you be surfacing that to your voters is

it important currently in our system it's a random sort we have not gone into this design space but we anticipate doing that but this is something that is now possible for every Community to start experimenting with and seeing what kind of results you can drive we also serve as past outcomes is it useful to know that this grant has raised you know ten thousand dollars in the past round or a hundred thousand dollars like the lifetime value that this this community has received is another form of sort of past performance which again kind of gets at that sort of collective filtering is it a useful metric to know that or should everyone kind of walk in and evaluate a grant on the merits of what they're reading in The Proposal um also has the permission to vote so we've used Bitcoin passport to ensure

that we're not having civils attacking our system but you could also see uh communities oh shoot I got to speed up I missed the time um permission to vote so do you need to hold a badge or do you need a badge holder of some kind to participate um and then lastly is the access to matching so I'm going to speed up how much time do I have I'm out of time I missed it entirely okay I'm going to show a few slides so how can you do this allo protocols or permissionless protocol to uh to innovate on this some interesting things people have built with endowment so just using our protocol without thinking about the the front ends they built Universal impact pools which is steering uh matching pools based on past donations to other um non-profits you got those other flash funding of the matching pools is

something that's super modular built where anybody can be raising funds and delivering them through an allo protocol and then you have hyper search which is pulling the on-chain data and trying to issue sort of certificates of impact that can then be potentially you know moved around or used later on Grant stack is the client of aloe protocol this is the new sort of open front end that we've we've built it is self-service you can Define these parameters as I mentioned for your rounds and Define the participation rules and you have the fulls life cycle for that so my question to you and my encouragement with my last second here is how can you design your Capital allocation for collective intelligence and I encourage you to please look at Grand stack and aloe and come talk to me

afterwards we'd love to hear what you guys come up with okay thank you [Applause]