Funding the Commons | Impact Certificates and Impact certificates and Impact Markets

[Music] [Applause] [Music] my next speaker owen cotton-barrett is a research associate in the future of humanity institute at the university of oxford focusing on decisions in the face of great uncertainty he's an effective altruist director of the research at the center for effective altruism and holder of a d field in mathematics owen will be speaking to us about the about impact certificates and impact markets the benefit of markets and altruistic impact and what's needed to achieve them welcome owen uh hi i'm really happy to be here um that screen check okay yes great yeah um so i i'm gonna talk about uh impact certificates and impact markets i think that many of the original uh ideas here originally traced back to uh paul christiano and dennis drescher's recently been helping me uh work out some things so this is a kind of relatively uh theory talk about how we could be reconceptualizing uh impact and how um the doing of good things uh we relate to that and what types of things even can be tradable um but i i think there might be a lot of value on the table by getting a good system set up here so what what are we even talking about uh roughly i'm thinking of impact is managing to change in the supply of a public or maybe private good um uh there's a couple of modifiers here uh if you ask change what does it that's supposed to be change relative to the idea is that there's some kind of counterfactual world history where you didn't take a particular action um mostly i think that we can not bother too much with this and we just have this intuitive notion of like oh you know what what do we mean by by impact and uh that'll flow through okay but i think it's interesting to uh be able to bottom out what we mean uh uh if we need to and then i also think that i'm regarding it as the responsibility for that change and that uh the key notion is that um you like do something good and then the the impact flows back to the person responsible um so that's the basic setup uh and then maybe the kind of like an interesting question is well if this is a thing that belongs to people often for things that belong to people we assign people property rights over them um and they can say oh well you know i want to give this away i want to assign this to to somebody else and then you can start building markets on top of that um and we can ask the question for impact as we've defined it here should we be able to uh uh give assignation rights and have people say actually uh this impact which was coming to me i am passing on uh to another party um and if we do then that could be the first step towards allowing markets in impact um uh like impact is this slightly rarefied uh abstract notion and i think that there are several possible benefits of having markets and they're kind of the usual benefits but transferred over to a new context i think that several of these benefits we can get through uh mechanisms which don't make any reference to uh uh like rarefied impact like i think that i've put here is the first benefit exposed evaluation of a charitable work or other other work that's aiming uh to be good um and prizes are a normal way of doing that but i i think that uh the reconceptualization can help uh to spot places where we can uh port over other market mechanisms and and there can be something useful about it so yeah the normal story for why x post evaluation is better is that uh you you have more information like we you just get to see more oh did that turn out well um uh versus if you're uh restricted to funding things uh beforehand then you have to go on the basis of do people have a plausible story and private information is like bites harder in that case uh i think that yeah so this can allow for better integration of diffused information where um somewhere before kind of the final price for some set uh the place allow whichever individuals have pertinent information as to uh the eventual value of something to get that information into the market uh by buying or selling and [Music] the markets can be a good way of integrating that information and then i think another potentially big deal is that it can incentivize excellence if you have a system where uh markets can give big rewards to people who do things that are really good then it incentivizes people to look for hey how could we do something that's great um and i think that uh uh just humans often incentivized by uh uh things like uh getting recognition but also getting rich and uh there's a lot of energy that goes into um oh how can we make a for-profit company that's that's gonna uh uh make us rich and i think that if we could get that um same innovation and energy going into uh how can we do things where a for-profit model doesn't work but we still want to be really ambitious and aim for uh big changes in the world then uh this is something where markets and impact can be pretty helpful um so the basic i like conceit uh which i'm gonna illustrate through to a metaphor here and so the idea is like rather than just accepting the impact from a good action that you've done and and you in this case could be a person or it could be an organization uh or maybe it can be some other kind of uh entity but you you uh like intercept with a bottle um and you say no i'm not taking the impact for this instead i'm going to put it into storage and now it's a transferable unit which i can go and sell um and yeah so you can think of this like an nft for a particular good deed that has been done um and the metaphor with the bottles uh the key step here is that owners can later open the bottle um uh kind of un uh meaning that now whoever uh gets to uh drink it or whatever uh gets the impact that was originally bottled and so um if you have tokens then at this point they would become untradeable they would record uh who uh was attributed the uh associated impact um and maybe if people often think of nfts as like a market in fine art this might be sort of more like a market in fine wine uh ultimately uh the value of things is propped up by people wanting to uh consume them at some future point um so that's the basic idea and then there's lots of reasons why this is pretty hard to get set up uh one is that impact attribution is often implicit and it's complicated you have a thing which a bunch of people have collaborated on and it's just not clear who gets which amount of credit um and and it's a mess to try and sort out and that means that it's a mess for anybody to try and uh sell the associated impact because the buyers don't really know what they're getting um i also think that right now we don't have good ways of recording anything there um and like the recording thing is we can deal with and then finally you know nobody cares we don't have people who are saying oh yeah i really want to buy into uh these impact markets there isn't the demand for it um so one other concept is i want to draw a distinction between impact certificates which are the things i was representing by bottles tokens uh denoting the rights to attribute the impact associated with some particular specified actions and impact shares which entitle uh the owners to share a future stream of impact uh uh certificates from whatever their own entity is uh i mean impact shares are just kind of shares uh but in in the impact context i just wanted to draw this distinction because i think that sometimes when i've heard people uh talking about ideas here uh the two can be blurred and i think that conceptually they're distinct conceptually they're both useful and i expect that if we ever do get to the point where we're building mature ecosystems around these we will be using both so what would be needed to set up impact markets i think there's uh maybe three central things there's uh bookkeeping uh we want some way of uh recording okay who owns the impact associated with particular uh actions um uh such that people can uh trust uh however that bookkeeping process is done um i i think that uh kind of connected to the bookkeeping is you want some basic way to validate that the the records uh of the things that are being written down line up with real world impact that they are uh supposed to represent um uh you don't want somebody just claiming hey i i did this amazing thing um uh when uh and and getting paid for it when uh uh it's kind of if essentially just fortunate um i think that the second is needing social buying you need buy-in from lots of different parties you want buy-in from whichever bodies are potentially doing the uh good deeds which would be bottled and issued as certificates um uh you need buy-in from uh marketplaces to facilitate uh the uh uh purchase of these um uh but i think and and maybe buying from if you want to get benefits of speculators and markets who maybe want the buy-ins uh from uh speculators but i think that the key thing is from the ultimate buyers uh those people who are uh have an interest in eventually consuming um uh the bottled impact and so here i might think of uh this as substituting in for when people currently give money to uh charities or other uh good uh causes um and saying well hey maybe they would prefer to buy uh impact if you're giving money to a charity it i think that the social contract is such that implicitly you are buying some impact from the charity you're saying well i uh want to give this money because i trust that it will be used for good things um and this is saying well there's a different setup where you could uh still uh buy the good deeds but you get to find out exactly what you've bought uh rather than doing it uh all on trust that something uh good will happen and also if these things get implemented legally i think that the point of consumption of these certificate should be tax deductible uh there's a bunch of questions of how you can actually set that up but that's the stage which is a very tightly analogous to making donations um and then there's a bunch of infrastructure uh that one would want to have in a mature ecosystem uh helping to support this i think maybe the most key elements here the ones three certificate auditors you want um some bodies which do the work of investigating um certificates to check out and say hey yeah this is legit the thing which it is claiming happened really did happen um because if that is left to each individual buyer to do uh then that presents a massive amount of overhead on the market and uh is likely to uh make uh just yeah purchases unwilling to enter and maybe you also early on want to have uh certificate issuers which help to facilitate that process it's also possible that you could set it up in such a way uh that people's self-issue you might want to combine uh three and four here you might want to have some uh relatively trusted bodies which arise as uh hey we will issue certificates and uh uh we will develop a reputation which people get trust in that we do a good job of auditing the things that we issue certificates for and therefore any certificate issued by us can be trusted um but that could also uh just live as a layer on uh on top of this i'm imagining that a mature market would have competition between multiple different auditors and uh trying to audit things efficiently without spending too much resources but also to a high enough standard that it establishes a high degree of trust and then there's a question of okay well if we would eventually need all of these things what is the critical path uh towards setting up markets here um and of the three bookkeeping social buying and infrastructure i really think that social buy-in is uh uh the key aspect uh we want people using this and then that will the more that there's demand and people using it the more that will drive everything else uh uh being built to meet that demand but i think that um uh bookkeeping and uh some aspects of infrastructure are helpful for facilitating uh that social buy-in if there's a clean implementation if people feel like oh yeah okay uh it it's it's very obvious what's happening and i feel trust in uh the technical way that's been set this has been set up then that can facilitate the social buy-in but i don't think that it's the only part of social buying which matters um i outlined earlier why i am excited about impact markets what i think that the the big upsides here are i also think that there are uh a number of potential downsides and i'm just gonna highlight a handful which seem to me like the biggest issues um one is uh that at least on the a naive implementation of this uh impact markets could incentivize likely harmful charities and the reason is that there's an asymmetry where if a charity uh does a bunch of work i mean sorry i'm i'm saying charity but it could just be uh actions uh which uh somebody is doing where they're hoping to make uh money by selling the impact um yeah if somebody does something that's very risky and that there's uh some chance of getting uh extremely bad outcomes and some chance of getting extremely good outcomes um the value of the associated certificates is generally capped below by zero um it's due to know zero is what you get if nobody in the market is willing to pay for this um and this could mean that uh if there's a high tail upside that people are incentivized to take those actions uh even if the expected value of the thing is strongly negative now this isn't a problem which is unique to impact markets uh in general it's a question of negative externalities and there's a question of how society should coordinate to uh try to tax negative externalities or otherwise uh incentivize people to uh not create them um but i do think that it can bite the issue can buy it particularly cleanly here and so it's worth as some extra attention yeah it's worth pointing out that um you might think well uh markets generally let people send a price signal uh that they they think something is overvalued by shorting uh a stock uh but because the eventual resolution here can't go below zero that doesn't provide a direct way for um markets to say that they think that something is uh actually strongly net negative uh so this could be this could be worked on um and uh maybe there are some uh [Music] interesting innovative uh solutions which could work here this could be regarded as a paul christiani pointed this out to me part of the general problem uh that impact markets uh as we're currently conceiving them don't solve the problem of moral trade where you uh meaning that you fail to reach pareto superior outcomes uh which everybody would um uh prefer uh because they're nobody's kind of first preferred option uh by trade um i think that there's also issues about changing the incentives uh otherwise on um uh people pursuing good work and it might push people towards things which are easier to uh turn into a impact that can be sold um or easier to uh or to put more work into marketing and explaining why this thing that you did in the past was really good of course this is to some extent a duel of issues which already come up for charities where they can be incentivized to pursue work that they think uh will uh uh sell well and they'll be able to persuade people to give them money to do rather than for having done um and uh to put a lot of money into fundraising for um uh uh uh their future work so there are traps there it's not um i don't have a particular reason to think that they are larger than the current issues that uh could be worth for the investigation and in general there's a kind of like hesitance of well this could be a large shake up in how uh charitable work is funded and uh some people have been thinking about that some but uh we definitely haven't worked through all of the implications of that and maybe there are unknown effects of how it affects uh of the impacts on on the ecosystem uh of moving to this system so i'm this one makes me particularly interested in uh experiments which have some ability to say okay no let's let's close up shop uh this uh seems like it's not going uh in a in a productive direction um uh but there are complications of how that interacts with incentivizing trust from people okay and so then in the last part of the talk i want to talk about what uh we're calling virici 0.1 um the numbering system and name is to to suggest that this really is uh kind of early stage thinking um and the idea here is that if we want functional markets and we want people uh having a lot of trust in those markets i think that that uh depends on people having a good understanding of how the markets work um and if i go and i try to buy a stock on an exchange i don't need to have gone and looked into all of the legal contracts that are supporting the particular um uh type of stock that i am buying and to check and i i can just have like some generalized trust in the exchange and uh uh that the thing is regulated in some way which uh uh s stops them selling uh 50 times as much stalkers actually exists etc um and i think that there are other ways well if we just set set up common expectations of how things are going to work that could be pretty helpful and so this is my current best guess uh for the social norms uh that we should try to set up as the rules of markets uh impact markets if we were doing experiments today to establish impact markets and uh you know i'm sure there'll be mistakes in here but i really think it is worth trying it if we're making attempts to start with our best guesses and so there are a few different properties the first is verifiability that impact has occurred uh uh i think that it is helpful if there is gatekeeping around the market uh which makes it clear to market entrance that uh things which are on sale there have passed some level of auditing uh that they really happened um i think that it's also really important to have clear ownership of uh rights and not to try and sell impact in cases where there's too much ambiguity over ownership rights i think that we want some kind of consumption mechanism uh this equivalent of uh unbottling uh of the impact treating it like fine wine rather than fine art um and i think that we want to set things up in such a way that they are incremental and that this is acceptable to existing uh interests in in the markets um and so i'm not gonna say a bit more about each of these four uh and yeah they they don't yet address most of the risks i mentioned before um uh these are more uh rules where i think creating common knowledge around this uh would help give the best shot of uh creating healthy markets which a lot of people are actually happy buying into um yeah i mean the basic issue with verifiability is that it's hard to facilitate trust and buyers if people can't verify whether the impact has happened um and uh as a matter of ui and making the process relatively seamless for buyers i think it's helpful if there is some relatively trusted zone where they can operate in and they can say well now i'm just going to look at what the things are claiming rather than thinking anytime i need to make a purchase here i i need to uh go and do a massive amount of uh due diligence and there are various different ways that this could be set up it could be um that one has a marketplace and one accredits various different auditors and then uh buyers coming in can do filters on i only want to see things which have been approved by uh this kind of auditor um yeah so the issue of rights i think is interesting i think that there are lots of cases where uh say i work on a project with someone um and then we say and then then we're like oh well that did well and now i want to go and sell the impact uh for my work on this this can create adversarial dynamics uh and incentives between the people who were the collaborators because now they're each incentivized to produce a public narrative uh for why their contributions were particularly important and other people's contributions uh were relatively unimportant and i think that that could be um pretty unhealthy and destructive and so i think that we uh want to avoid having that and often i think that the way to do this will be via explicit agreements among collaborators about uh how impact will be shared um similar to the advice for uh uh startup founders to uh agree a share of equity very early on uh in in the process so um society has to grapple that with this uh uh in in various cases and there are um a bunch of situations where uh ownership already ends up uh uh pretty clear and like society has allocated a standard answer to this if somebody owns a company uh 100 then presumably they own uh all of the uh impact certificates that that company is generating um similarly for copyright and if there is uh impact associated uh uh uh coming from the uh the copyrighted material um by default i think that should go to the copyright owner um i think that one can specify a clear explicit divisions uh and in general if you're doing something by yourself uh it's uh pretty simple relatively unclear cases include donating to a charity and so uh a challenge for getting impact markets off the ground is that i think that right now charities often couldn't uh in good faith issue uh certificates of impact because there is significant ambiguity as to how much the rights for the good deeds that they have done lies with the charity rather than the donors to the charity if they have some kind of um uh explicit expectation setting with the donors uh uh then i think we'll clear that up um i think uh you get a similar thing with um people who are employed by a charity where uh i think that there's an implicit social contract that uh they are doing that in part because they regard their work as contributing to the good that the charity is doing uh but i think it's ambiguous as to how big their contribution is and just lots of other activities where there's a collaboration uh but there's no been no explicit discussion about divisional credit um then consumption uh i was pointing to earlier i think that this has a couple of different benefits one is that it helps to provide social proof um that there are ultimate buyers uh interested uh and these buyers are about removing tokens from circulation and so i think that this should uh support trust in underlying value of uh tokens because it builds up a track record um and less than fears of bubbles i also think that it helps with uh the the final property of being incremental because it provides continuity with the current practice um in the current practice uh implicitly when somebody is doing something good they're just consuming uh the impact immediately and this is saying hey there's another operation that they could take uh where they say no we're going to hold off on that um and bottle it and put it into a certificate and then maybe that can be sold but if they don't do that uh uh we're just in the status quo one um and i think that that continuity is helpful because if you imagined uh something without a property like this then there could be uh a bunch of people saying hey does this impact markets thing mean that now lots of charities need to be paid again for the work they did didn't we already give them money for that uh ahead of time and i think that that could create bad feeling and could uh make parties uh who like the parties would be giving money here uh uh somewhat disinclined to engage with the process for fear of being exploited um yeah so this is the the basic principle behind incremental i think that at every stage we want to uh set things up so there's just incentives for uh people to uh come in and start using this uh and that nobody is directly losing out as a result yeah in order to avoid creating uh kind of entrenched opposition uh uh to the system i think that uh this can be totally i think this is totally achievable um and i think that uh it's normally the good pathway to to go for if uh you want a new economic technology to to to take over um okay so uh i'll wrap up that uh just to as as and then we can take a few questions and the final comment is this yeah i kind of hopeful that impact markets uh could be a big thing and could transform um how a lot of analysis is done and a lot of incentives in the ecosystem of people trying to do good things and i'm really interested to to see how they develop and i i think that this kind of conference is great because the people at conferences like there so exactly the right people to be going and building experiments i think so and for a comprehensive look at markets and altruistic impact that acknowledges the risks as well as the upsides and we have some time for questions and i know there are a few in the chat if you don't mind i'll surface them here the first question is from actually another one of our forthcoming presenters andrew kritsch and he asks oh and i know you've thought about chapter value to reward counterfactual um impacts like making a good attempt that falls short of someone else's i think having people engage in public discourse about chaplain value attributions could really help reward people who almost made it and thereby incentivize more attempts to do good could you comment on whether something like that seems good or what challenges it would raise yeah okay great question um i think that the main challenge uh uh that i'm worried about is this thing that i alluded to about uh trying to sell impact when uh uh there's unclear ownership of the rights i think that it could create adversarial uh incentives for people to uh overstate their contributions and understate uh those of others um however i i see i see like the prize that you're pointing at and i i think maybe there could be something good there i feel more like oh this is worth more thought not i know how to make this work um i'm in the moment mildly pessimistic about uh getting it to work in a straightforward way but there are lots of uh more complex implementations which might dodge uh many or more all of the issues thank you and i'll i'll ask another question this one's from one oh and do you think bottling and consuming is required to make impact certificates work why can't consumption be a time bandwidth oriented i.e the current holder of the certificate like with art markets yeah so i do think that there i i think that this is a a pretty significant bifurcation in how you want to conceive of and set up impact markets i think that the uh world where you say you treat it like the art markets and you say uh the current holder is deriving value makes some sense but it has it loses significant amounts of cleanness relative to the uh one where it uh which has consumption and it i think it may lose significant amounts of the benefit so some of the cleanness that it loses is uh you can no longer just look at something and say oh how good was uh this as as a way of informing uh how much you should care about uh buying it you sort of want to look at well how much of the value has already depreciated um and uh you're going to be baking in some expectations of well how much uh society uh seems to care about this thing but it reduces the ability to say well we want to to use this as a kind of like consequentialist uh a mechanism um i think that it uh also probably makes it significantly harder to set up in a way that gives you anything like tax deductibility um uh uh for buying these things i think that uh uh for my my understandings of like the principles of how tax deductibility is supposed to work uh not anchored on uh the laws of any any one country the idea is when you give money to a thing for which you do not receive like material benefit that's the point at which you should be given the tax deduction and i think that uh the fine art model of uh impact certificates never has such a moment um and therefore it is hard to uh uh work out how the tax deduction should work i think that the uh version with consumption does have such a moment and that's better and then uh i also think that the fine art model makes it harder to uh incentivize profit motivated uh speculators and analysts to enter into the market um because uh you ex uh in this model you expect by default uh most certificates to depreciate with time maybe there'll be some which go up in value because people just realize how great the thing was so you can still get some of this but essentially there will be a significant tax on anyone who is uh trying to enter into this uh from the perspective of well uh i'm not gonna be the person ultimately like buying these and wanting to hold the impact but i can add good information to the system and i i can earn a profit for doing that and so i'll i'll come and uh help things out and and those people are just a bit more pushed out of the system which means that i expect to get uh uh less of the benefits of relatively efficient markets uh out of that all of that said you totally can do a version uh without consumption um uh uh there's also the uh the issue of maybe you're more likely to just get bubbles um uh in in the version without consumption uh because there's nothing uh uh propping up the uh uh uh the price of things um but the yeah the original proposal for certificates of impact uh uh by paul christiano and catcher grace uh didn't have consumption uh i i feel less optimistic about it doing really good things but maybe i'm wrong thank you for ending on a very humble note there i think we have time for one more question just a quick one from evan miazono oh and do you know of anyone building the market you described are you planning to and what support would be most helpful um i am personally not planning to but i know uh uh uh dennis uh dresher who uh uh helps me with some of the thinking about the uh the vuichi standards has been uh thinking about what would be needed to get a fairly minimal uh implementation of this uh set up on top of uh nfts uh uh in some existing uh chains uh i yeah i am actually i haven't touched base with him recently enough to know what the current status of that is i think maybe we were thinking oh we should publish the blog post on virici and we haven't quite got that great well we would love to keep uh apprised of this research and there's some excitement in the chat about this idea and um now it's time so i'd like to thank owen again for his time [Music] [Applause] [Music] you