7. Funding the Commons Q&A - Lisbon 2022

uh cool so here's some quick updates so in the last funding in the fun in the comments in New York I talked about uh setting these goals for the next year we're about 3 months out from there so I want to do a quick status check on how we're doing as a community so the goals were number one grow the events continue to grow each FTC event Explore More domain specific tracks to create workshops and hackathons we should expand FTC to include um all of these and a enable theorists and Builders to connect and build uh we wanted to set up a fund the common Network Funding structure itself to be able to kind of uh fund lots of groups to be able to work on this uh four we wanted to grow the community fifth we wanted to focus on impact certificates and try to start building the tooling and structures for those six we wanted to uh make a bunch of experiments around impact evaluators um and seven we wanted to start uh thinking about the performance performance evaluations and of mechanisms and structures that we're trying to build uh so first growing the events uh how we doing that hey we ran two events in three months that's huge uh shelling point B there was a pop-up it's a new event format it was super successful 20 plus speakers 300 attendees and the conversation diffused all through Devcon there are a bunch of workshops and 100 people um in joined the hypers Serv Community through that so huge uh success from all the folks working on that and then there was L week 22 lisman FC pop up today here um 15 speakers around 200 attendees or so um with workshops as well and more important than those we now have a plan a 2023 plan um with two main Flagship events next year eight smaller events with partners and 12 local community events in 2023 workshops and hackathons uh so we had a workshop for planning the hyers Ser project I think there's probably been a bunch of other workshops like that that I don't know about around IP nfts and other structures um I'll wait until it comes back adapter fun there we go um there's a uh hypers on impact markets working session in bota around um uh drafting the development and go um drafts of development and go to market road maps and there's a bunch of hackathons where we're going to be uh talking about all this stuff so there's no dedicated hackathon yet for funding the commons but there's a hackathon uh for e listman this weekend there's an fbm hackathon there's EF and web through weekend we're going to make sure that um funding the common is present in all those and help um uh give people uh lots of ideas and what are the kinds of structures they could uh they could use so now FTC Network Funding um we don't have direct progress on this yet uh if you're interested in this ask questions in a moment or come talk to me after um but uh dii we did as a community help run the dii uh Grant matching round and that was um really successful uh we probably should have a grand round like that dedicated to um to this network funding structures so maybe we should like just get one of those things in D sorry in gitcoin um that'll be a good way to get started and there's probably other other structures um and I guess you know uh getcoin is running so like that's a meta uh thing in and of itself uh in terms of growing that Community bringing three new people per event and recommended Three Plus speakers um yeah I think we probably made that with 35 I don't know probably were like somewhere in between I didn't know how to gr this one yet um so I'll check in next time I do this uh for impact certificates there's been a ton of progress here um a lot of attention has formed around uh hypers certs and doing things with hypers certs there's lots of different groups uh starting to work with them uh there's like a project forming there's um tools being built uh there's a team that has been assembled to like go go and build uh a lot of the underlying tooling there's other teams that are being discussed and talked about so this is really turning into into a community quickly um there's a website and documentations there's a vzer prototype working uh I think it's at least in a testnet maybe a mayet uh um there's uh Communications and Outreach there's public talks podcasts blog posts and so on so the word is really getting out there about hypers Sears the hypers Sears idea made it way back into the EA Forum where impact certificates were first described so that's really cool um so a lot of progress here hopefully but however no like proper impact markets yet um we need to get there so but we're in um we're in good trajectory for uh for doing this in a year now impact evaluators there's been uh some progress on this uh as well not as strongly as impact certificates um but but also strong uh there's been a number of experiments uh the network Goods team um uh in the p network run five experiments at least with Falcon and iest I think maybe some other projects uh there's three prototypes tools uh that those need to be open sourced and there's a a few talks explaining what I I are and there's two talks today um this one in a moment and there's um there's impact evaluators for ceramic coming a little bit later today now for performance evaluations how do of these mechanisms themselves uh we do not have progress on this yet in great part because we're just getting started building them um and so that's fine um but this will become more important say in late q1 and Q2 uh next year cool so that that's it for that any questions on this part no questions all right great everyone knows everything sweet great uh let's see now was so the the first thing I wanted to talk so now I'm going to talk a bit about these sets of things um and kind of some progress on on these idea these ideas uh one thing that I think is useful is that I've been working on kind of like formalizing what an innovation network is um I think there's a very specific um construction here of a system to drive uh work across the R&D Chasm uh so some of the work that I've been doing around this is thinking through so I've presented in the past about the Innovation Chasm which is kind of the incentive field in the kind of research side and the incentive field in the market side um and the Gap in between those and so worked on this kind of visualization to help us see that um think about this as a set of projects that are going through um all these systems and think of maybe the size of the um of the circle is like the amount of work that that running these things might take um and with the understanding and idea that there's many different projects that that go on um this is not a linear development so it's very rare that the same team can TR translates everything all the way through um and it's also very rare that it's one single approach it tends to be that it's this Cascade of many different attempts like one result leads to many attempts in the next uh pathway of those some are successful that lead to uh some improvements and and so on and so forth all the way into uh getting to things scaling uh and and so what I think Innovations Innovation networks are extremely useful for is in this area in the Innovation Chasm kind of post research in that translation helping lots of groups um explore this project landscape uh to try and kind of um Orient groups around this development um and help structure many attempts um gather learnings from those uh provide resources for attempts so think of funding structures and reward structures and so on and helps coordinate the work towards the output so you can think of something like B Labs as an innovation Network that happened to be housed within a single Corporation um and now you might think about like the pl Network as um an open network that's oriented towards these you can think of YC as an innovation Network as well though I think it's further down in the pipeline it's much closer to startups it it it's um or it's very close to startups um it needs to it would need to be kind of like much earlier in in R&D uh to be uh I think to be a much stronger Innovation Network um great so um there's been a ton of progress from convergence research and others in setting up uh fros the focused research organizations uh there's been a lot of um several teams that now have um funding and are operating um I think that we need to have kind of like an frro demo day at some point and I think that would be extremely useful um and also a way of and through that connecting those fros with lots of different funders um there's been some Pro progress on structuring these Network funds uh to be able to do Network capital allocation um there's two that I highlight one is the ipfs implementations network fund um which was announced at the ipfs thing and it's going to be announced again in ipfs Camp um there's a significant amount of funding um uh in kind of like the um $5 million range plus um this year loan um dedicated towards funding IP implementations and this is um again uh R&D R&D funding and there's a private content routing fund uh just for helping U privacy and so like this is within our community there's of course the getcoin grant rounds which are are kind of like this and the other big highlight for me is V da I think that V da is not just a blue fund it V da is an example of an FDF I'll get to that in a moment maybe I'll get to that now um uh which has been extremely successful at like doing combination of Grants and starting to um get IP and looking at funding some startups uh and so it's going to be really interesting to see how that develops over time um so an FDF is a fundamental Development Fund um it's a an idea that folks in our community have been thinking about in terms of thinking about a capital capital allocator that sits sits here in the kind of development part um it is sort of kind of like this so this is maybe spanning the whole thing this is not an FDF an FDF would sort of like operate in just a subset of this area and use some of like the the ROI of investment capital in a sense but be able to kind of fund development early on um you can think of PL itself as like a mega FDF but in reality you want to have like something much smaller and and more Concrete in there so think of like a small VC fund or something like that but think of like a small fundamental Development Fund in a field so you could have like a nanotechnology uh fundamental Development Fund or a um uh yeah many other other kinds of fields uh cool so talked about hypers service Anders and one thing that I think needs a lot of our attention is that we now have a lot of The Primitives and um tooling to experiment with um funding structures for open source I think we we uh 5 years ago we did not have all of these tools we have a lot of interest and hopes but we didn't have the tools we now have the tools we now have the capital there's a lot of groups a lot of crypto networks with vast amounts of um of resources um to be able to fund maintainers and developers of Open Source directly from in a kind of network oriented way there's a number of experiments already but none of them have scaled in a very significant way yet um and so I think this is an area that I would like to draw everyone's attention to uh because if we can get this stuff right and we can create a structure where you say couple a GitHub repo to a dow for funding it and there's and people could have like a a highlevel alignment that you know at least 50% or even like 40% of that funding will actually translate into success for that project that's aligned with the goals of the community like that would be a huge breakthrough like um that percentage might even be lower like even if just 20% is efficient um like that might already be very very significant so finding good structures for doing that um in a way that like you know mixing money with open source doesn't destroy the project as like sometimes it does um I think would be extremely extremely useful and and a massive unlock for all of these ecosystems be amazing to be able to just kind of go around GitHub and be like oh that's an important project I want to fund that oh that's an important project I want to fund that and have that acrew um uh resources for the maintainers and developers of those systems um one other area that I think is uh important and valuable to work on is kind of network oriented coordination systems uh so this is like how do you set objectives together as large communities and have like Metro World progress guess against those sometimes in terms of like measurable outcomes like apis and so on in other cases with like um more qualitative key results of saying certain things have been achieved and how how can we kind of adapt the tooling that has been very successful in hierarchical organizations to operate much better in this kind of network oriented environment we have many groups um so it's not like a tree it's more of like a like a dag or like a network of potential goals with many different groups potentially contributing to Pieces um so I think like those kinds of systems are in high uh not exactly demand because I think people don't know that they should demand them it's more like there's a huge need for these and um we don't quite recognize that we need them and we don't quite recognize that we should um make them and so um yeah I think we should have Works around these things and if you have ideas for how to do this if you have observed like good coordination systems working at scale and you're like hey this could actually work really well for any network not just this organization uh give a talk at one of the funding the comments or write a block post or something um and then that might Inspire somebody to go and like run an experiment and try it out uh cool uh any questions on these parts yep toering we've been prompting come talk to a come give a talk of future funding the commons on this topic if anyone makes a proposal on one of these topics and we accept it we will pay for your travel to fund in the Commons so hopefully that's an extra added incentive I think the next one's going to be in Barcelona in February and we'll be like you you've seen the Cadence there will be funding The Commons at a place where you won't you'll want to be so come do like do the work give a give a proposal and then come give a talk yeah awesome so a call for uh Network coordination systems and ways of like establishing objectives and how do you measure progress against those and how to group sign up for taking on those that work and how do you kind of commit to that progress evaluate that progress and and so on and how can we do this across organizations y hi um I have good news and bad news good news is I spent six years at Google and a lot of what I did was just that I was part of the team that built Google's current or are to uh the bad news is that we have there's a lot of stuff that you can do uh about the you know the explicit part of the tool to make it better and easier to to integrate with a bunch of stuff and uh to explain what the meaning of the okrs are and so on but I think that and this is something that I you hear from me all the time you have to be realistic about the extent to which the tool depends on the the community and the the the degree of trust and the the power relationships and dynamics that are actually driving it because when that breaks down it's just a shell and it discredit itself pretty quickly yeah so yeah um there's there's a lot there so I think on the one part there super exciting when I go talk to you later about how do we make that tooling but for these networks um to okrs are a system like work really well in like yeah very high trust um environments um the and yeah they definitely start breaking down when you don't have that or when you don't have a good agreement on whether or not how a k KR or a kpi has been achieved and like that starts like really breaking or how these Cascade um there's another bug here where which has always bothered me about okrs which is that they're not recursive so um an object objective tends to have KRS underneath and a KR itself is not always concretely an objective and so on down and so it shouldn't be but but you need some way of organizing these things and linking them so there might be some other like structure where like these things derive into some other o or or something like that all I'm saying is we need some like recursive structure where you can actually connect these things and so groups working on some part know how this automatically translates into success other places and lowering the lift for humans to have to do that work today there's an enormous human cost in coordination and that needs to come down um and that can be uh it's not going to be completely Sol by tooling but it's going to be supported by tooling um but and this is just one example there's dozens of these kinds of toolings I just tend to use OS because it's very simple and a lot of people um when I said very simple I thought of like many conversations were like so complex but believe me it can get dramatically more complex on okrs um great other questions thoughts yeah's Mike and everybody please say your name so that the community gets to know you if you want to and are comfortable with that of course hi my name is Boris man from Vision uh where's the main space to uh have discussions and uh kind of float early ideas to look for collaborators to try something uh the internet uh I think that's like the closest thing I've got I think uh I think there's a network Goods Discord and is there a discourse Forum maybe there's not discourse and there's a there's a gitcoin discourse and um we should make one for these okay and uh I'll make a quick pitch to see if someone is is looking to do this so on some of that link things I've been thinking about uh my shorthand has been sustain your stack so can we get the pln to use the existing open source repos that they already use uh and commit to funding their uh Upstream that they uh you know some of the stuff has been done before but it feels like it might be useful to like actually do it with some of these new tools so come see me if you want to sustain your open source stack and let's figure out if we can get some people to try it within the pln and I will put it on the internet awesome thank you and I think it's time to kind of try uh some of the ideas that um Source C is trying again um maybe in a different different uh structure different different model um yeah and I think that there's a you things like coape and others are also good good ones to try um other questions thoughts comments Expressions expletives yeah um I haven't say I'm not fully up to speed with with all the jargon so I was slightly confused but um you're talking about science technology and open source all these things seem quite different are you are you uh intending to incentivize all of them using the same structure or um not the same structure um what is the fastest way I can talk about that um so my claim is that um we have extremely inadequate economic systems today um and that most um most of the kind of like most of the progress that we've seen in humanity in the last couple hundred years has been writing on this amazing pathway of improvement that comes from the Scientific Revolution and the Enlightenment and compounding all of that and you know systematically like delivering a lot of um utility across the world um however like it's pretty good however it could be dramatically better and we have so many examples of where this completely breaks down uh you know there's like this great cartoon of like yes the planet got destroyed but for a beautiful moment in time we created a lot of value for shareholders and it's like well clearly that points out that the entire like underlying incentive structure of how Capital works is messed up like there's some broken set of incentive Loops there um that that are yielding this like horrible outcome outside right the fusion never graph here that that I have is another example where like there were these estimates about how much it would take on an annual basis to fund Fusion to like get it within a certain time frame and um the problem was that all those estimates never got delivered and so the actual funding is that black line at the bottom and the there were all these estimates of like when we would get Fusion like we' get in 1990 if we like invested you know in the order of like 9 billion per year is um and you know this is what actually got invested so it's called like Fusion never because it we never got the funding for it until now now there's like a lot of Corporations doing it um with some government funding but it was like not not at the scale that it was required however when you take all of that that Fusion never money and you compare it to the digital home entertainment market today um like our Netflix watch not just Netflix but all of this stuff could have funded Fusion right it's like that's the scale that we're talking about so we're like saying oh you know I'd rather hang out at night than like get Fusion you know 30 years ago um so I don't know yeah not exactly right but I ju Just comment that that the I mean I suppose you will go into this in in the impact assessment things but I mean for example something as complicated as as a racing car like rather than putting out loads of tenur for who can you know give a best gear better gearbox and this is the specification if you do it then you'll get this money they simply offer a price for who comes first in the race it's a much simpler thing other people self organize in order to construct incredibly yeah yeah so so I'm not proposing there's one construction that will just work what I'm saying is that um we need to drastically improve all of the economic systems by in piecewise replacing things and creating new structures so for example uh you the the fact that we have crypto now means that we can encode a lot of these systems not in programs that are hard to implement and and require vast like amounts of coordination around contracts and so on but you can actually deploy a smart contract which is way faster so that means you can like you have this massive lever to um do things much faster so can we kind of use this these like superpowers to try and kind of uh have an impact on this and so the claim is let me skip forward um uh when you think about science and technology as like these two sides of the same thing meaning there's a scientific process of expanding our knowledge and there's the technology process of expanding our capabilities these are two pieces of a larger integrated system that involves tons of people around the world tons of organizations and so on and it's feeding back on itself and this is like the last few hundred years of growth um now there's an incentive problem here where there are two different incentive fields that don't meet and so there's no significant incentive coordination structure in the middle and so that is the bottleneck and what what I'm describing is like we need better incentive structures to close that gap between those incentive Fields ideally eventually make it so that the bottleneck is in science and then like let's F science more but that's not where the problem is like the by my analysis of this situation is like we have a lot of um research happening we have like this exponential number of papers going maybe a lot of them might not be as good as um as it might claim to be but there there's like this issue of everything getting bottleneck between that and the market and that's where I think we need to focus our attention it's like building better systems here and it might mean things like come up with new IP structures come up with new IP markets maybe this regard IP entirely and go into a fully open source structure that has like interconnect the graph of contributions here and do the proper credit assignment so you can flow money from the end all of these things are like different potential things and so the this community is like experimenting with a bunch of these different ideas and then we're kind of like coming together to evaluate them over time seeing what works and then kind of rinsing repeating scaling thanks cool awesome uh I don't know if I did like a good good enough summary but like that's like the um that's like the the program um uh cool any other questions thoughts yeah hi um ma um I really like the fact that you bought the whole IP situation because I think with um kind of like the open source culture and so on the conversations around IP composability have come up quite a lot um and new forms of Licensing um associated with like a certain technique knowledge or whatever that could really obviously drastically uh unlock research but also so many other Market verticals and obviously speed up the solution process um from your perspective uh yeah like exactly the patent light iing side of things like how are you seeing that from a composability perspective I guess composability is more of a longer term outcome that then generates this kind of I guess permissionless collaboration with I different types of Ip um but yeah what's your thoughts it's a great question so um I'm gonna uh at risk at plugging my own talk uh I gave a talk somewhat recently about why fixing patterns is important um to like set my my uh Baseline I tend to come from a perspective that like patents primarily right now are hurting the system more than helping and they're very far away from what they were intended to do originally however there are some Industries where like they're clearly helping and preventing like large scale organizations from uh wrecking and destroying small ones um and like and completely drying up the the the ability for for small innovators to to actually um operate but that's not quite the case in software and so what um what I gave this talk about is like hey progress proceeds at the rate of Science and Technology Innovation um we need safe and fast efficient progress right now Innovation itself is R limited by these incentive structures um we're just like this great superpower now pads are this key Innovation incentive system it was designed as an incentive system and it kind of is that there's an enormous amount you know long history of using uh IP all the way you know 500 bcn before uh there's an enormous amounts of patents um there's good R&D about um the game theory of patents and how they work uh and they're like good conclusions most of which have never been used or tried um you know another example of like things great ideas getting stuck like decades in the pipeline um there's really good analysis about the environment now there's an enormous amount of patents and most of them never get transacted on so there's a guess here that like by cleaning up the patent transaction structure you might end up being able to kind of like um get this like really fast Improvement rate now it's risky because like I'm not the kind of person that is like hey let's have more patents in the world I would prefer if we had less patents in the world but um maybe if we can like remove the exclus like patents give you this like crazy right where you can exclude all of humanity from using a thing and that seems egregious so maybe we can just remove that structure or put a price tag on it that is limited at some of the the resources you've invested in it so if you invested x amount then you can only claim to exclude up to a certain amount and if Humanity wants to buy that from you they always have the right or something like that um like this kind of crazy exclusionary system is like part part of the problem um but for example think of like how Airbnb came into a Marketplace or sorry came into a an environment and created a new Marketplace out of um sort of like missing structures where uh all of the rooms were there uh all of the people were there people were traveling and what really was needed was a different coordination structure and a different transaction structure needed some standards it needed some cleanup of the system it needed a reputation system and like that alone enabled this entire new economy to emerge and so that seems possible here this seems like bunch of the structures are there a bunch of the assets are there maybe what this needs is just like a huge kind of overhaul now I do want to kind of like remind everyone like patents are toxic asset they're like nuclear waste or digital surveillance like um because they create these like extremely exclusionary Provisions um they can be like used very um for you know created a lot of damage now note that that Spike so that was like the patent troll Spike and then you know uh Newegg and Cloud flare and a bunch of others were like absolutely not we're going to like fight all these trolls and look at that you know they Smash down the the um that crazy uh patent litigation Spike and so that's an example of like coalitions to like prevent that kind of stuff work um so yeah so my my guess is like we need some like program to to like actually fix pattern but uh yeah however like I say this with mixed feelings like I think they toxic asset however they are an example of the instrument that was designed to operate here um but maybe we just need new instruments so for example to give you an alternative view instead of patents what we could do is wire up all the credit contributions of all projects we can think of all of the people and all of the organizations involved in these projects make that legible in a in in digital systems like create structures for that and then figure out some rough credit assignment even if we get it wrong by some percentage it's better than nothing Thing Once you have that graph you can then propagate value from the end uh all the way through the graph so that's what uh the source Square project was um uh uh started to to um go and sort of help create many other projects like that have um explored different parts of this credit assignment problem um I think that's a totally alternative way of doing this that could work super well so maybe that's better than than fixing patents but we should try both um all right sorry pan rant over other questions or should we get to impact evaluators just wanted to quickly note that the uh the fixing patents talk slides and all that is now uploaded to protocol's YouTube channel if you search there's a playlist called fixing patents and I think there's 10 talks each about 10 minutes long and uh W is a a 20-minute keynote that was all of the uh the slides he was just showing so it should I think you need protocol ABS in there somewhere sweet yep there it is so you can go and find the playlist there should be the second one from the top fixing patents Workshop there we go lots of awesome talks happen here yeah how you doing hi there my name is River and I've been studying a lot around uh n natural complex adaptive systems and also indigenous knowledge systems and I think pointing to both of those are going to be where a lot of insight's going to come out for this coordination issue um and what we're seeing in some of the indigenous communities is the exploration into living libraries so the library seems to actually be a perfect bridge between research and that production piece and it's the physical um infrastructure of what libraries actually were in the past that created a lot of possibilities but Li don't fund the development anymore like libraries say in like the Library of Alexandria and other places did fund the work of the researchers and so on but these days you can't go to your public library and expect like $10 million in funding for a lab no but that's not not to say that we can't reinvent that and attach patronage to libraries again because if you look at back back in the era there were libraries but there was also patronage of certain individuals through that process as well yeah you could yeah like you definitely could do that um my all I'm saying is I think that's probably harder than just building the patronage system without also having to force it through the libraries you could probably just do that over the internet and just enable that coordination to happen separately and then advertise it at the libraries sure no I'm I'm not even saying using existing Library infrastructure the way that we think about it at the moment but um one of the projects we're exploring with indigenous communities in the Amazon because it's so much knowledge locked up in these indigenous knowledge systems is embedding these living libraries that enable them to record some of their insights and actually make it accessible to the rest of the world as well yeah um and you know they have very sophisticated technology that they've been using for thousands of years it doesn't look like technology to us but this low Tech um is hugely powerful particularly in with regards to organizing and coordination issues um so I think you know if if anyone's interested in looking in some of these aspects I'm happy to share what I've kind of learned along the way from different communities but yeah yeah two thoughts on that one is um information diffusion like I'm a huge fan of all of that whenever it shows up and helping uh store distribute disseminate preserve give open access to all of all this kind of information helping um organize it um the second part is yeah it's amazing what people were able to figure out like um there's a lot of important proces that were figured out early that are like nanot Tech or like um deep Material Science Innovation like a kneeling in like Steel or sorry and making in Metallurgy the the concept of a kneeling of like you have something really hot and you sort of start pulling it out slowly to like let it cool that's to get like the crystal structure of the metal to align perfectly and it's like a nanotech level not quite nanotech but like Material Science um materials technology uh Innovation and like they had they don't have the concepts to understand what was happening but they could able to like figure out the Dynamics around it to make it work so super super deep things were were figured out a long time ago many of which have been lost like yeah sorry other rant not now uh yep not as much of a use useful contribution to this topic ju but I just wanted to take a second and let people know that we're starting a parallel track in the breakout room starting from now till the rest of the day so there's a second set of sessions going on we'll be kicking off with one on data da Primitives and governance mechanisms uh and that'll be running in parallel with hourong session so you can see what's going on uh we'll be kicking off in about 3 minutes there sorry to interrupt thank no that's great um feel very able to if you want to go to that conversation right now feel free to stand up and go there no one will be offended so please like definitely catch that conversation uh this will be recorded uh it's a good translation time to go into impact evaluation time so um let's do that I'll give a minute for folks to uh move over all right cool um impact devel uh so what I'm going to do is I'm going to give a quick refresher intro to what these things are I'm going to give you some examples and then we're going to walk through how to design one of these and then we're going to give you some space to like actually write one of these uh and my goal here is to help you kind of um and help me and help others think more clearly about these structures and how to design them then see we have some interesting stuff that we want to build and then have good candidates for hackathons and so on uh so an impact evaluator is a um an economic mechanism or or a um yeah an economic mechanism that enables a community of Agents to achieve an objective this is a a community of agents that are mutually distrustful they don't have to um distrust each other but the point is like they don't have to trust each other either you want to create an incentive structure that's going to enable the this community to not just cooperate um but ideally cooperate in a super linear way where they're helping each other succeed to achieve the objective faster with like ideally minimal coordination between them uh if we can build these things well they can scale really far so a concrete example of one of these things um my the the cleanest and most simple and highest scale impact in recent years is the Bitcoin block reward the Bitcoin block reward is a very simple uh impact evaluator all it's measuring is the amount of hashes that a particular participant is producing in a unit time and that enables the community of Bitcoin miners to um assemble an enormous amount of computation power to secure the Bitcoin blockchain and dissipate energy and and trade that energy for a um digital representation of that value so it's a way of running a um a market and an exchange in a sense run by the loss of physics that enables participants to trade energy and their access to energy for digital currency and it secures the blockchain at the same time it also has the terrible side effect of being very wasteful in that process and you know consuming an enormous amount of energy that probably shouldn't be shouldn't be happening but it gives you a sense of the power of these things um uh another example of a of an impact evaluator is the idea of an academic conference uh you can think of an academic conference as establishing a periodic evaluation of impact to a field there's an open call people provide submit papers there's a community of reviewers they review the work they pick a set of them and at the end of that they reward the community with a certain amount amount of reputational credit and that credit um is useful to that set of individuals and the community of Agents together is exploring the field and everyone is incented to share the learnings as fast as possible with each other before they get scooped and that causes everyone to kind of like quickly Advance a field uh now of course um both of these impact evaluators have serious problems at the same time they have worked very effectively um now the way that you construct an impact evaluator is by first defining an objective by thinking about who the community of agents that are going to work together think of a measure that's going to track that that's going to help you achieve that objective um for example in in in a bitcoin's case it's like you can easily measure the hash rate um the objective is securing the chain and the objective is giving you this like physics mediated exchange like a way to trade energy into digital currency um and so you can use like the number of hashes in order to to achieve that um there's a set of funds that are pre-allocated to run the impact evalu um this is a way of like pooling a set of rewards now these funds could be a stock meaning a stock here economic term of stock of saying you set aside a bunch of of reward ahead of time or this could be a flow uh so it could be a value flow that is coupled to some structure so you can have a mechanism with a value flow coming through um the block reward is not that the block reward is a is a certain amount of a currency um another example might be the um in the conference for example there's not like limited uh academic reputation um there's maybe it's maybe limited only in bandwidth um and maybe it's one of those currencies that the more you give it out the less valuable it is um but that is a value flow right so the program committee could could choose um has an ongoing flow of reputation and there is you can assemble these things like the measure the funds and a way of um of uh of identifying the community of Agents into this p periodic structure that is going to reward their impact over time so together way can assemble this mechanism and you can think of it kind of like a control theory Loop where you have some sensor you feed that information from that sensor into a controller structure which then kind of like figures out some uh changes to the system and then you then measure what has happened and you feed that back on itself so most you know um things like airplanes and cars and so on work with all kinds of like little structures like these um think of the economy being able to like be being able to be run by a bunch of structures like this and so you could imagine creating economic mechanisms um you know networks of them that work kind of like flying a plane so you can fly a plane with this very complex set of control theory devices and similarly you can think of like navigating a community and navigating an economy with a bunch of these little structures that are automatically kind of sensing where you're going now it's complex and and so on but you know it's possible um cool and yeah so the you don't just have to have a reserve you could have like an intake inflow or something like that you could also imagine this as a valve so you can have some cash flow going and you can think of like this sensor and this evaluation structure as a way of redirecting the value so you have value coming in and you sort of like reorient where it should go and you just couple that in in a system um cool so what do these look like you have some time period uh at certain uh Standard Time intervals you're going to me meure the world you're going to try to understand how well you've done against your kpi you're going to um award you're going to figure out the weighing of the contributions of agents and then you're going to give out some rewards um based on that so for you know you have a Ser series of rounds you measure the world you evaluate um how much that has um what the contribution of each participant was and then you try and give out some rewards based on those weights uh you know measure measurement could be simple you could just kind of like understand kind of the value of the kpi you could like measure each individual agent's contribution um weighing could be as simple as saying great like you can reward proportionally but you don't have to they could be different proportions so you could have things that are like sublinear or superlinear or some other ways of combining these things so you could plug in a different weighing function there um and of course then uh you can also use this evaluation structure to plug in human judgment so you you don't have to do this automatically one of the hard Parts here is you can't evaluate every in these small control loops so instead what if you could call out to um humans and have them uh decide what to do this is the case of the conference paper submission right like you have a program committee uh you have the conference organizers that they say okay great program committee here are the papers like evaluate them give us some output ratings and then we'll take it from there uh and so that's an example of like calling app uh and then you know you have some some rewards uh at the end that get distributed out uh so you can assemble these things into Networks uh so it' be so they kind of look like neurons and so you could you know start plugging them in together and one other thing worth noting is like it doesn't have to be Zero Sum um you could have structures where if you tweak the reward structure uh you could have a super linear um uh outcome thanks to Evan for coming up with this idea um the the idea here is like in a b reward for example you you just have a certain amount that's going to go out and it gets divided evenly uh you can make the super linear but just changing the amount if like if the impact of the kpi is greater you increase the the um uh the total reward and so everyone benefits so that causes collaboration instead of having this Zero Sum fight between say like the Bitcoin Miners and so on you could have a much uh Stronger collaboration between a set of participants to try and achieve the outcome together because everyone gets more if the kpi goes up higher um so you can you know kind of get into um P preferred uh the P preferred quadrant uh now ideally you can get into like fair and like enticing structures where like just everyone is kind of very compelled to collaborate this just kind of like makes sense for everybody to just work together and call each other to collaborate uh that would be an amazing way of like very quickly scaling um one of these things uh great so we kind of cou covered a couple examples anyone have any examples you want to throw out just names of them y wonder if you're aware there a famous result you can't use control theory to stabilize economy cool I am not familiar with the result that's great I'm going to go read that um how definitive is this result I mean like definitive like the second law of Thermodynamics definitive it's actually for ofal banks because governments would optim the situation theil but but what if you don't have central banks if you don't have central banks it's much worse that's why they they give the mon policy to Central Bank and they make it so if you go back to like the airplane example you have a system that automatically trying to like navigate the structure to follow a trajectory yep the Eon so I still haven't been convinced theal thing to do is to say you're going to do a certain thing in the future and then break that commitment over and over again but you what if you can't break the commitment uh in which case that it will work yeah exactly so with cryptographic artifacts we can make commitments that are unbreakable you can pre-commit the the the funding and you can't take it out sure this is this is an additional thing well it's really cool so that you're saying that like theoretically there's a result there that like it's seen as a negative like result that then suddenly like some basic like cryptography can like enable like this leap around this like that could be amazing there's no B you're but you're right maybe not basic but like um like these commitments are like really powerful being able to commit economically is like super super super powerful it's simple like from a conceptual level but like that's really cool we should go write a paper it sounds great um maybe we get another no price strategies game the you can't there's a result in Game Theory uh that uh if you don't if if in a game where where participants don't have theory of mind there's uh there's a certain class of strategies called zero determinance strategies that can dominate any any dominate TI for Tat or any of the the standard uh strategies and the only way to get around it is to have players that have theory of mind and the way that they do it is that they negotiate treaties and they negotiate treaties by building in commitments that that can be broken like uh um what we were talking about is that is that result in variant of scale or is that kind of like on certain levels of complexity of game no that's a very that that applies to to the the class of which prison dma ipd is is a super class yeah cool this is awesome yeah another let's keep the comments going this is like way more way more useful uh Mike wait for Mike uh gri coin's been identifying impactful research and incentivizing it for up to 10 years pretty effectively through a very large network of individuals just another example yep I didn't realize that like the ability to waffle in a commitment was like such a like baked in Assumption about like the economies today like because we didn't is a strategic disadvantage yeah cuz we didn't have computers we didn't have the idea that you could actually have this mechanism that was going to enforce it without like anyone's ability to change it absolutely all these results were yeah yeah um I like that you mentioned judges before and um I've been doing quite a lot of work around land regeneration and one of the evaluation systems that we're looking at and considering much more closely is what we call Birds per minute so when you work with the land depending on the amount of birds singing per minute is an indication of the health of the ecosystem so it's an impact evaluator embedded with the natural ecosystem and we're seeing huge um implications associated with that it's obviously just one indicator but birds are a very good indicator of the resiliency of an ecosystem so I know it's maybe not quite as technical as you know some of the other examples but it is uh very much connected with the natural ecosystem so uh yeah a great impact evaluator great so yeah thank you more birds in systems um was there another one over here or good um wa for Mike yeah just riffing on uh Rivers Comm ments about birds per minute um I believe you guys are into the active inferencing and the free energy principle and so that feels like a very good high level bet trick of course Birds per is an example of that cool thank you um great so I'm going to walk through how to design an i and given that we have three minutes left I'm unfortunately going to have to cut the workshop time but the viewers of this video will be able to uh I leave that as an exercise for The Watcher uh and uh we can have some QA Q&A afterwards so uh cool so let's go through how to think about an impact evat in detail so um oh one other mention about IP and so on um from the conversation earlier mun and V are a good example of like mixing IP uh in this like new new oriented way uh great so I would first suggest you know identify the objective what is the network trying to achieve get super concrete and super clear on like what the goal of your um network is like and here's is like a small Network right so this is just like in the Bitcoin case it's just the Bitcoin um miners or something like that uh so what is that group trying to achieve together get extremely concrete about the objective so in bitcoin's case it might be like securing the network or trading energy for money without having to ask anybody for permission so like that's um that's the objective now you can then after that identify one or more metrics that you could optimize that achieve that objective so you may need if you you have like multiple impact evaluators that are together forming a little Network you might need multiple metrics or create a function that you know bundles these together and you know how you're weighing that function uh because you know you can't optimize for two things at the same time that just you're optimizing one function so just Express that function um you should be aiming for easy to measure things like I don't think these impact evaluators work well at all if this thing is very hard to measure and that might mean computationally expensive to measure it might mean very hard qualitatively for humans to elicit the preferences or elicit the the thinking here or value things like don't aim for things that are hard for humans to Value if you do that like you'll get like very negative responses um aim for things that are extremely easy to measure that you can do quickly and rinse and repeat like this is why kind of conferences ask for like the weak reject uh sorry strong reject weak reject uh weak accept strong accept because hey like it's not perfect and you know definitely reviewer 3 gets the uh gets all the annoying comments and so on but but um it gives you a super super fast way of rinsing and repeating this at scale it's a very simple protocol that everyone understands you can pick it you can go for it and it works reasonably well given like you could try to beat it but it's actually surprisingly hard um so in you know bitcoin's case this might be just uh measure the number of hashes that you're doing in that second committed on top of the last block so that's that's a very precise description but that's that is what the impact evaluator of the block reward is measuring the number of hashes that the network is producing in that moment in time uh from there we get into um identifying the work and group uh identifying the work and the groups doing the work so the work itself is like spending this energy doing these hashes securing the network the group is like the Bitcoin Miners and whatnot um for more complex stuff like this is really important and make sure at each of these stages go back and see if like you got it right meaning like as you identify the work being done you're like oh is that really going to achieve the objective that I have or not or is that going to cause some problems um then think about ways of um measuring that structure easily so for the hash rate that you know you can have some way of like doing that proof of work idea so like counting the zeros in a in a in a in a hash is like a very easy way of measuring the number of hashes that you've produced that that was like the proof of work result from uh Adam back and others before that um the and and ideally identify a few different ways of doing this I I want to express that the Bitcoin impact evaluator is like the most off fiscated way of doing that like you could have done that a much better way and I think this was an accident like just accidentally uh Satoshi and others got into this amazing result I don't think it was like you know Brilliance and covertness I think it was just like accidentally really really good they already knew they had something good but it was extraordinarily better than they expected um then after identifying ways of measuring it easily you have to check the incentives the moment that you start rewarding some activity you will get way more of that precise activity and less of everything else so you're going to get what you measure and reward you're not going to get what you wanted to get so like try and make sure that what you're trying to do is correctly oriented so for example I don't think anybody in the bitco community said let's waste massive amounts of energy around the world and like burn all this coal and like carbonize the planet that was not a goal it just happened as a side effect because they didn't have the legible instruments to express hey like make sure that this is like good energy there's an important conversation that happened before Bitcoin was created where people in financial crypto way like two or three years before Bitcoin were talking about making um greencoin in Iceland greencoin uh and they were talking about creating a cryptocurrency that used the energy of the world and translating into that and um and and they were talking about precisely doing what Bitcoin ended up doing creating this exchange mediated by by by the world and so on um and they were talking about using Renewables and so on and they were using like the K's style like burry gold in the ground and dig it up well do something more use will harness energy and so on so it was not about like let's try and burn a bunch of coal it was like let's actually make make some progress um it just kind of like it wasn't checked and so you got this like really bad outcome um so then after you do all of this and you feel like really good um you're probably going to get stuck here and this check the incentives Loop multiple times as you like refine your IE several times over and over and over again once you are happy with it try describing it in Pros uh make it py think of like what the IP IE rewards think of the metrics it's measuring think of the work in the groups measured describe in a paragraph once you've done done that name it make it really well scoped do not overpromise don't call it like the amazing value IE the fix the world IE make it like super specific like the scaling constant sha 256 impact evaluator which is like the Bitcoin block reward or the scaling storage capacity and Deals IE which is the file coins block reward IE and after that you've got an ie and like you should go on building and try it out and hopefully don't destroy the planet hopefully uh make things really good all right thanks [Applause] w