Octant’s Journey Towards Sustainable Funding for Web3 Public Goods - Andrzej Regulski

DM DM even it's don't uh so thanks for coming my name is Andre uh I work at G foundation and today I wanted to share our experience of uh building a protocol which is supposed to among other like objectives uh fun web threee public goods um this is uh second presentation following uh art and note this morning uh but I'm sure there will be like some new stuff here so even if you have seen the presentation this morning there will be some uh good stuff uh so uh what can you expect today uh I am an economist so uh I'll start with some uh like econ talk related to what public goods are and some mechanism design and incentives involved in uh this initiative of funding uh public uh public goods uh then I will try to move to this more specific application of public good funding which is the web free space it's kind of similar but there are like specific aspect to that and also uh we have some very cool data from the first year of up operations so you will also see a lot of charts like why I made get another chart as an agenda I thought I would get away with yet another one um okay so uh let's start with some boring definitions uh but nevertheless I think they're kind of kind of useful in this context so in the Legacy world we uh Define public goods as something that is non exploitable basically it means that you cannot keep anyone from consuming the particular good uh as something non R valous uh so basically your consumption of a public good doesn't really have a negative impact on anyone's consumption uh uh there can be some positive externalities involved um you can you can see that the like the the pictograms they show some examples of uh what public so are and uh probably with every example you can see that this definition doesn't really hold up in all of the cases but that's the problem with economic economic definitions nevertheless uh public goods have some interesting properties uh in from the perspective of uh how they are supplied so first very often they are under funded uh they are under supplied and there are some issues with sustainability um and this is especially the case if there is uh no uh if there is no uh nation state that can Levy taxes and basically fund such public of course in many cases such as some basic infrastructure or uh clean environment or a Judiciary System which are like textbook examples of public goods to some extent uh we have a nation state but in many cases such as like digital public goods think about uh Wikipedia or uh uh internet archive uh it heavily relies on users donations which is not a sustainable model of of fun now if we move to like the crypto crypto space uh or web free you can see that uh most of these definition and features more let hold up um the main difference with the Legacy world or digital public goods in um in a broader context is that uh resources to potentially fund public goods are not scarce in uh web free I mean there is a lot of potential streams of funding that could uh be uh directed fund public goods the problem is that we lack good coordination mechanisms to uh accumulate and to coordinate uh various actors to fund certain certain publics of course there are uh some uh ways to go around that for instance by uh using VC funds or or or issuing token but they kind of end but by doing that we kind of end up with the public goods becoming semi-private and this probably shouldn't shouldn't be the case uh and again in uh web free there is no uh nation state which is a good thing in general and there are no taxes which is also a good thing but uh we still end up in this problem that many elements of this like connected tissue of web PR either uh like core ethereum protocol or many open- Source libraries that everyone relies on are underfunded because uh the potential benefits of uh working on on such public goods are not evenly distributed various actors have various motivations to uh to fund such such initiatives uh so the most heavily uh used tool to fund public goods in in web free right now is uh are various funding rounds funding programs uh has any of you participated in a Bitcoin round uh okay do we have any recipients of uh our project who received present of project who received funding from gitcoin or retro PDF or okay so there is there is some um so if we look at uh all of these uh funding uh mechanisms in the web three we can see that there are like four basic components that you need uh for such a program to to run uh so first of course you need funding so you need like a source of capital to to finance certain public goods you need to set some objectives which are what the money should be spent on you need donors and by donors I mean individual donors um to some extent we need donors to accumulate more Capital but but also donors are quite important to Signal what should get funded so individual donations even if there is a very uh generous matching funding then still donors are kind of important to signal and to provide information about what sort of public goods should be should get M and of course you need a distribution mechanis so um here I want to introduce you upand so which is uh like a flagship prod project by Golem Foundation uh in terms of capital Golem Foundation States 100 ,000 actually a bit more than that and uh this staking operation generates uh sustainable yield and this yield is directed to octant uh among other things to fund the matching funding uh for public goods and to distribute rewards to users who participate in up the objectives are basically to express the preferences in terms of public boosts funding of uh of the community of glm token holders uh glm is an erc20 token it was uh created in 2016 um in Golem Network crowdfunding and golm Foundation is a spinoff from goem Network we with the objective of finding new use cases for the for the GM uh the donors in our case these are uh glm lockers so users have to lock uh the glm token and uh after some time they are eligible to receive rewards for participating in in governance and they can either actually decide whether to claim their Rewards or to donate their rewards to public good projects and also we need distribution mechanism uh in our case it was I will uh say a few words about that later but we started with a linear distribution of of matching funding now we are using quadratic PL uh here is a very simple uh uh chart of that shows the flow of value in uh in uh in octant so again we start with uh govern foundations taking uh e uh this generates yield which is then directed to the octant uh octant rewards are distributed to uh users who log their glm and receive proportionally to the amount of glm Lo they receive uh rewards every every quarter they can either withdraw their Rewards or allocate or donate them to public goods projects if they do so then their uh donation is matched with uh some generous matching funding from uh also from the stating operations of of G Foundation uh every quarter uh we have something called allocation window it's a twoe period during which every user is able to First presented with a list of eligible projects in the recent Epoch we had 30 projects that uh users could choose from they basically short list them by clicking this little heart uh it's sort of like a e-commerce basket experience once they do that but uh in another view they can um decide on how what fraction of their individual rewards should be directed to one of the public project also you can see that you can sort of predict uh the amount of matching funding based on the current state of things uh that will be directed to your for your project um so far we uh so far octant uh has uh so so far users of and have uh locked uh 100 5006 million glm which is about less than 16% of the of the token Supply uh to give you some intuition like one glm is about like one polish slot more or less uh so uh or if you are not polish natives uh divide this by four and you will get like more as the USD USD equivalent um we have about 1,000 lockers at the moment uh the entry ticket to join uh octant is 100 glm that's the minimum lock uh so about 25 bucks uh the median lock is about uh 1,000 uh so you can already see that the average like you divide the 56 Million by 1,000 the average is like hundreds of thousands uh glm so uh the distribution of our lockers is a bit uh skewed uh we also have larger ways which is a Big Challenge and I will talk about this later uh over one year we our staking operations resulted in more than 3,000 e of staking rewards uh about 600 of these were uh claimed by users as rewards for participating in the in oon governance uh more than one one 1,100 each was distributed to over 50 Public good projects so this is in today if we take today's exchange rate it's almost $3 million that uh Octon Community distributed to fund various publics um to give you just some intuition about how Octan Compares with other uh funding uh funding mechanisms that are out there in web PR so uh probably the closest initiative that we could compare a with is gitcoin Grants uh the gitcoin grants 20 had a matching funding which was like more more than twice as uh the amount we had in oant Epoch 4 one epoc is premount so basically this is the amount uh we have to multiply by four to get like the but in a way the similar applies to bitcoin grants as well uh if we compare octant with Bitcoin you can also see that oan is a bit more Boutique uh because uh we have only five 30 potential granes uh this we are doing this on purpose we kind of believe that it makes sense for our users to be able to like review the entire list of uh of potential projects uh if uh we would go beyond 30 that would make it really hard like for our community assess what should get get funded uh also you can see that uh the donation the the size of individual donation which again come from Individual rewards that are generated within the system are much smaller than in gitcoin but again this is a consequence of uh Octan being uh uh gatekeep by G by the by the uh by the requirement to log glm so we cannot spontaneously uh allocate in an oont allocation window we are working on that but we are not there we're not there yeah so this also result in The Leverage being way higher in in uh in so the by The Leverage I mean that amount of matching funding that is uh added on top of user Dage uh I have also added uh optimism retro uh funding uh for uh just as a benchmark to show that there are some really huge uh uh Public Funding mechanisms out there which are way larger than po of uh so this is like my statistics uh perspective here but uh just to give you some intuition even though the circumstances are very different uh it turns out that uh the distribution of uh how our users allocate funding is not that much different than in gitcoin or in U in red RF uh for instance uh we the top 25% uh in all cases account for about 40% of the uh of the matching funding distribution oant is a bit special because uh we don't have a very our funding cap is very high so the the most successful project in an octant in the recent octant Epoch was able to Claim about 16% of the of the matching funding which is which wouldn't couldn't have happened let's say in in geek um okay we can look at oite as an experiment as an experimental platform we keep tweaking something in the algorith every every ook and the core of this experiment is to answer the question if uh like users are willing to give up a fraction of their individual rewards uh to fund public goods uh these amounts on the like left hand side you can see that they vary Epoch by Epoch probably Epoch 3 was a bit of an outlier uh I would say that this three to two to 3% uh generosity coefficient fuel uh is probably the equilibrium we are right now in and most likely this is what I would expect in in future in future EPO and by the way it's a it's a good example because uh in developed countries between like three and two and 3% it's exactly the amount uh households are willing to give up for Char even though we don't consider Octan to charity because we kind of believe that our community wants to invest in specific outcomes and funds project that makes sense to them it's an interesting uh comparison also it's worth emphasizing that like more than 90% of our users by users I'm here in my that interact with octant are donors so they decide to allocate at least part of their rewards to uh to to to public goods on the other hand you can also see that uh this generosity or the fraction of individual rewards that is allocated to public goods is way higher in among smaller lockers so by smaller lockers I mean uh those who uh lock less than 20,000 glm for about 5,000 bucks uh and this is also like what brings us to a like a mechanism designed challenge that we faced a couple of months ago so uh we have uh about 90% of uh our users who account only for uh 5% of the total donations and in this linear model of uh funding basically the the 5% also translates to their voting power so uh for for quite some time we kind of believe that this is not a back it's a it's a feature uh but uh there are like two major problems so first of all like biocracy is a is an issue and uh the first one was that uh the results of a funding FR were very sensitive to movements or decisions of a handful of whales uh which if they happen and that's what the case in three if they happened uh just towards the end of the allocation window it just uh turned the table and of the leaderboard and many other users were not even able to to respond to certain Decisions by by the whales and also as I mentioned like a good funding program in my opinion at least it really requires individual donations and especially like smaller ones because uh this is how you gather useful information and signals from from your community and uh the problem with that was that our users like having seen what uh little impact they have with smaller donations they ended up growing this incentivized to to participate in in in funding PR so uh that's why we decided to introduce quadratic funding in U in uh uh EPO 4 I I assume like some of you might might have heard about it but just like a very short introduction so in the SP distribution that we used initially basically your the the share of matching funding that the project receive just depends on the total amount of of funding of individual donations so let's say we have two projects uh each of them uh manages to raise 300 way or doesn't really matter uh the only difference is that project a has three uh individual donors and Project B has uh 300 one big individual one in this case so this is how oant work until EO 3 each of these projects would the same would get the same share of the of the of the mering fund because the to the only thing that mattered was the was the total amount received uh in quadratic funding so this new mechanism that we introduced the total uh the share of uh matching funding a project receives is basically based on the squirt sum of square roots of individual Dons and uh if we replicate the same numerical example so again two projects uh one of them uh with one donor and two of them and the second one with three donors with the same total amount you can see that uh the one with uh more donors basically has a much higher share of of the ma so this is how we envision this and uh we again as an experiment we uh implemented this uh logic in Appo in Epo 4 and yeah the nice thing is that it works pretty well so uh these are like real uh numerical uh it's like real data from uh from the voting in EO 4 uh in the second column you can see that the big lockers so those with uh more than 20,000 glm log their sharing total donations was uh more than 90% and this was would would exactly have been their voting uh Power if we use the linear distribution um what uh changed was that with quadratic funding their voting share is now slightly above 50 and the group which would have only enjoyed about 8% of the Ving power now has almost a half um you can see this working in practice on this um more granular granular granular breakdown of uh of uh specific specific um okay I know that like putting entire spreadsheets in a slide should be uh considered a felony uh but I I I will only do this once uh so here you can see the list of uh projects which receive funding in O Epoch 4 uh and in the last column that's the very interesting one you can see what is the net effect of introducing quadratic planting so if there is like a green uh a green uh and these amounts are in E by the uh the green and the the darker the GRE The Greener the uh the field uh the more benef the more the project benefited from introducing quadratic F uh you can sort of see that this especially applies to those who the those who are the winners in this in this change uh applies to those who projects who manage to accumulate both uh a big amount of individual donors but also the big total amount of B so these things very corate correlate very well with the the magic funding received uh if we applied the linear model uh this would be a straight line only in the uh right hand side chart so it would be just a 100% correlation with the with the number with the total amount of donions uh I think I'll skip this one uh we have at least uh time for uh a couple of questions uh but let me wrap this up with uh uh by returning to this uh like building blocks of uh uh public good funding program so I have sort of hinded that some elements of that are not entirely sustainable and uh by working on the next uh version or the evolution of out we are considering creating something called sustainable public goods funding as a standard an industry standard that could be adopted not only by us but also by other communities um so in terms of funding uh it would sort of make sense to move away from static Capital being uh used on discretionary basis to funding sustainable funding streams such as in our case uh yield from staking operations but we can also think about let's say al2 sequencer fees or many other streams that can uh be directed to public goods funding as for objectives this means moving to not only objectives of a single Community but coordinating objectives across many communities uh thinking more about the outcomes and uh measuring impact as for d as for individual donors uh it would make a lot of sense for uh participants to become financially empowered kind of the way we have been doing this at Outland and moving away from uh being just uh like charity uh charity doners and as for the distribution mechanisms well it again would make a lot of sense for this to become uh multi-level thing where many communities apply their own uh distribution mechanism which are applicable to their particular situation so that's it for me thank you uh make sure to visit our socials uh we have a booth uh downstairs so so down the hallway uh and uh at 10:30 tomorrow there will be a Aram our head of engineering and he will present you with uh a more detailed vision of how we see this uh uh unfolding um I think we have uh space for like a couple of questions so yep I'm going to come over with the mic so the question is recorded for the stream as well uh all right all hello okay are you concerned about um Cil attacks yes well uh we are and it's a it's a huge topic I didn't touch this on purpose because this would result in a so right now we are using uh gitcoin passport uh or passport as a Cil Shield uh it's not the perfect solution and uh there are like really really many challenges with that it's an ongoing discussion and you know our internal calls are usually captured in more than half of the of the time is is about civil resistance so yeah we are thinking about that uh happy to discuss this uh later thank you can you share a little bit more details about the multi-level governance that uh it was on the last slide okay so uh sure so uh we don't yet have it uh we are still working on this new standard um but and we don't yet have a like a full architecture of how this is this is going to look right but long story short it would make sense for various communities such as uh communities of various Els or many other like d DS with uh non- negligible treasuries to be able to come together and decide on what sort of what fraction of their either income or even the treasury themselves itself can be directed to public goods funding and also to enable them to make their own decision which are kind of specific to a particular either or or an ANL to or any other community so we'll be working on that uh it's a it's a huge challenge because uh everyone is sort of incentivized to do their own thing but uh we can also think about certain mechanisms to uh make them Cur and uh we'll see how this UNF Falls okay thank you Andre yeah thank you