 Okay, cool. So I'm going to be giving a presentation today on crypto economics and science, but I do want it to be really collaborative and interactive and really open source the content. So I'm not gonna give you all of the content and I would like people like our CAD-CAD crew and Paul and anyone else in the audience as well that has been maybe speaking today or involved in other blockchain or science project to help out with the content. So what we're really gonna cover is kind of what is crypto economics, what is token engineering and a little bit about curation markets, TCRs, TCBs and FTEs, stablecoins. Who, if you know what, just to give me a quick idea, who know what token curated registries are? Who, if you have no idea what a token curated registry is? Okay, some of you just don't like raising your hands. It's token bonding curves. I put the C on the B the wrong way around. Okay, cool. Who of you are building projects using token bonding curves? All right. I know that there are some projects that are currently sitting here that are using them, so I'm gonna get you guys to give some input in on it and those of you that don't know what it is, I would like you to ask all of the questions. Other than that, yeah, all right. So to give a quick intro to myself, my name is Devin. I'm originally from South Africa, Johannesburg, but I've been in Cape Town for about seven years after that and I've been living in Switzerland for the past year now. I studied a B-com in IS at UCT, so a little bit of computers, a little bit of business, a little bit of everything in between, and then after that, my previous life prior to blockchain, I was actually a Google Trekker, a project manager for Google Trek in South Africa, so I spent about a year hiking around the country, collecting 360-degree street view imagery of all of the hiking trails, so that was quite cool. Prior to that, I worked at a bank, so I mean it's not as great. A bank, yeah. So I did my Simpson corporates, and then, gratefully, after that, I founded a blockchain software development studio with Paul and out of that, spun molecules, so it's been quite a while, I think, changed in careers a hectic few years, but it's been really exciting. In addition to that, I also run a monthly curation markets call, so it's an online global call. The next one's actually this Wednesday, if you guys want to join. We got Billy Renicamp from the Clover's Network and Mike Ilias from Ideas Cartel who's going to be presenting. Anyone's welcome to join, anyone's welcome to apply to present, and it's a really cool collection of people sharing ideas, sharing knowledge, and it's great way to build up your knowledge base. Thanks, Sinka. I also run blockchain community calls, so that's a monthly one, it's for more introductory-level people, so any projects can also apply to present on them. It's a more entry-level audience of understanding, so we basically cover sometimes blockchain 101, but it's really interactive as well. It's a great starting place, really friendly and supportive community. I recently started up the Blockchain and Pharma Network in Basel in Switzerland, so I don't know if you guys have got it by now, but a lot of my work actually revolves around community-building, ecosystem design, working with people, creating communities of people that want to share, collaborate, and be together and are interested in the same thing, because that's a really important part of building out a technology, which I'll actually show you a little bit about that later as well. I run, I'm a part of the Future of Females, which is an international girl empowerment, female empowerment, female entrepreneurs network. There's about 80,000 people globally in this network, so I started up the one in Berlin in 2017, I think, and now I've started up the one in Switzerland, but there are in about 20 other countries as well. I also run Ethereum meetups across Africa. Actually, not anymore, I recently just backed down from that. I ran them in 15 African countries, and it was a global community base of about 15,000 people. So the community was insane, the hunger to learn in Africa was powerful, it was very inspiring, I'm sure. If you guys are interested in any of that, I'll talk to you about it later. And then last year, oh, this year actually, wow. I ran ETH Cape Town, which is the Ethereum hackathon. First time I was on the African continent. It was amazing. We had, I think, close to 200 hackers, 33 submissions. 66% of them were South Africans, so there was a lot of involvement from the community about that. Kind of what I've been involved in. So now what I'd really like to do today is get the involvement of this community here, talking to one another, figuring out a little bit more about one another, and seeing where we can grow our strengths here. Yeah, so a little bit of an introduction to molecule. I was gonna go into this a little bit more, but Paul's already given a really good presentation on it. So for those of you that didn't hear, molecule is an open-source ecosystem to incentivize decentralized research and development in chemical compounds and drugs. If you have any questions on this, I'm gonna let you go straight to Paul. But if he didn't mention, which I'm sure he did, we're actually launching our office soon, which is the Molecule Catalyst. It is crowdfunding scientific research, huesing token funding curves. And we're actually launching with a really cool project as well. So it's the huge sort of micro-dosing silo-siven for creativity and mood enhancements. So it's a really exciting study that we're launching with. It's with Rotem Petranca and Thomas Anderson from the University of Toronto, Psychedelic. It's a long acronym, hold on. Come and ask me afterwards. But yeah, so crypto requires an understanding not just of the tech, not just of how to develop, not just how to code, but it's actually a number of things. It's economics, it's the business, it's the cryptography, finance, cybersecurity, monetary policy, securities law, psychology, sociology, geopolitics, and history. It's never just one thing. There are so many things that influence our decisions and how we act and what we do. And this tweet was actually, it's quite a while ago, it was in 5th of March, 2018. So I mean, it still stands true. And I think if anything, it's even more now that you have to consider when you're building systems like this because they are complex systems. So crypto economics, not to blow your brains, but it is cryptography and economics. So it is the study of economic interaction in adversarial environments. The designing systems like those decentralized systems for your peer networks, we have to, we are often faced with the challenge of there's going to be bad actors in the system. And we kind of need to account for that. So using cryptography and economics, economics on steroids, we get to use the power of economics, but we get to secure it with cryptography. And a lot of blockchain technology runs on the principles of these crypto economics, some of which we're going to touch on now, and we're going to look at new ways to incentivize changes in our behavior. I think one of the most important things to consider with these systems, with crypto economics, with blockchain, with trying to incentivize behavior changes, is that a lot of this is still, it's experiments. This is not the way to do it, it's not the blueprint. We're not going to say this is how you do it and this is going to be the result. It's going to be very varied and we have to factor in that change and how much we don't know. So crypto economics, why do we need it? I mentioned this earlier, in decentralized peer-to-peer systems with no centralized authority, we must assume that there will be bad actors looking to disrupt the system. So we use crypto economics to create a robust decentralized peer-to-peer network that thrive over time despite adversaries attempting to disrupt them. In summary, cryptography, economics. We secure the networks and we incentivize the actors. But also keeping in mind that there are a number of different things that are going into this. And what we'll cover today is token engineering, just the definition of it, we're not going to dive into too much token engineering, but we're going to dive into curation markets, token curated registries, token bonding curves, and then a little bit about economic modeling simulations with CAD-CAD. And the way that we're going to do this is through a fishbowl type style. Yay, so that means you guys actually have to speak back to me now. Do we have a microphone that floats around and would it be possible to turn the chairs to face each other? It can be a little bit dynamic. Yeah? It doesn't matter. I mean, we have a microphone style. Okay, who of you work in blockchain projects? Yeah, and we can just give it a while. Please raise your hands and attract with me. Yeah. Who of you work in science? All right, okay, cool. Okay, yeah. Who work in both? Three, four, four? Five? I'll take it. So the way that this fishbowl panel works, basically, is I'll bring up a topic and I'll give you maybe a little bit of an introduction around it. I'm gonna pick two people. I'm gonna pick these two at the front because they're here already. And they are gonna be my core panelists, but then the other two seats, basically, they're gonna robble. So if you guys have a question to ask, you're part of the panel then as well. And somebody out there has to answer the question. So generally we have the panel up here. Do you want, should I put the panel up here? Yeah. Can you just bring four chairs up here? Yeah, yeah, of course. Cool. Yeah, cool. I don't think he realized I'm talking to him. Yeah, yeah, yeah. You need some beer then I'll put you in here, alright? Yeah. Thanks for wearing with me and rearranging all the furniture. Yeah. Cool, so can we do, you guys probably been, not just four, four is fine. Yeah, okay. And then what I'm gonna do is I'm gonna bring up a topic and then I want at least one volunteer to come up and take a seat and answer the question. Or if you wanna ask a question as well, you have to take a seat and sit up on the stage while you get it answered. Cool, so our first topic is token engineering 101. And what I have here is systems and mechanism design plus software engineering, token engineering. It's rigorous design analysis and verification of systems assisted by tools that reconcile theory with practice, too. Engineering is a discipline or responsibility being ethically and professionally accountable to creations and number four consists of ethical choices. Crypto economics are considered a building block of token engineering. But token engineering is not necessarily token economics. It is a building block. So I would like to ask an open question to you guys. Who doesn't understand what token engineering is? What is it token? The question I've taken, probably there. What is a token or what is ethical? Ethical, okay. What is ethical is more muscly, but we should be started by asking what is a token. Does anyone want to answer the question as well? You're welcome to come up. Okay, I can give my interpretation. I think in the context that we have a token, I mean, a token is just an entry in a database, basically. It's an entry in a smart contract that allocates that token to a specific person. It's actually really, really simple, but from a database perspective, but I think the most common conception that we all have is that a token confers some type of right. Yeah. So I generally, in the context of blockchain and crypto systems, associate token with an atomic unit of information that is sort of provable through some sort of on-chain process or through some cryptographic process. So it may not necessarily be something that you can transfer, but it's something that you can verify to be true. And it may generally confer a right or at the very least prove the existence of something. So it's a basically, I think the easiest way to describe it would be an atomic unit of information, which is cryptographically verifiable. Please have drawn up. There is definitely is not a single definition, so. Okay, so I think that the token is the most defining part of it is that it is a kind of digital information that is owned in a non-custodial way. So, and actually it is transferable. So this is for me the definition of a token that was not possible before these inventions of decentralized blockchains to own a digital token. Like you have Euro on the bank, but it's custodial. The bank or whoever runs the database is in control. So it's ownership, but not possession. And tokens can be possessed in the way that they are non-custodial. And this is in my view is the most important part of it. So the one thing I will note is that that is a sort of subset. So of the definition where it's sort of verifiable state information like we were discussing. So if you take a broad definition, you can get something that encapsulates things that are not inherently transferable, but that definition is in fact created in order to include but not exclusively include the definition that you're giving. Which again isn't to say that it's the ultimate definition, but there is sort of narrower and broader definition. So you're talking about for example, a verifiable claim or something like this, which is a token. Okay, okay. I think if it's a more broader thing. So I think what you're referring to is absolutely correct. I think it's a feature of a token. Like there's a self sovereignty that like we can own it and it can be transferred on a network without a custodian. I think in the future we'll see many, many sort of token based systems that still have custodians. Because for example, if I now tokenize my house or like a piece of real estate and the tokens are self-solving and I lose them and this is a question that often comes up. Once you start to tokenizing assets and then you lose those assets, like is it like too bad that you just lost your house or like a part of it? Because you lost it. I mean, that's one of the huge problems with crypto in general is that like the custodianship of private keys. And yeah, but I think that takes us down a completely different route at all. So token engineering. So actually this is a good segue to token engineering though, because part of the reason that I gave the definition I gave about system state and atomic units of information is that when you are engineering systems, there may very well be aspects of those systems that don't meet the definition of a sort of transferable asset or a specific instance of a transferable right, but actually just more generally represent state information of a system, which is mutated by some transactions or some activity by the users and that the token engineering sort of sub-discipline is a bit more interested in the sort of system design with the crypto economic primitives as tools. I agree. No, I just want to add that I think a lot of what we think token engineering has become as a discipline is much more like system design. And then the token takes a defining point in that system. And I think, I mean the whole point of token engineering I think emerged, so specifically, is the guy from Ocean still here? No. But a massive shout out to Ocean and specifically to Ocean in Berlin, I think in my view, I think they really kick-started a massive conversation around token engineering, which has brought this space a huge step forward. And I think really token engineering emerged out of partially out of the ICO boom, which like the earliest definition, I think that trend gave around token engineering was like it was ethical design. If you're an engineer and you build a bridge, you want to make sure that that bridge is sound and stable so it doesn't collapse. And I think a lot of the early token economies that were launched through ICOs, there wasn't even, it was like a design for a bridge, but like completely unimplementable that would just coax people into like funding the bridge, but the bridge never even got built. And so I think token engineering emerged out of that, really trying to build sound bridges. Yeah, actually, so to kind of go along this vein, one, thank you for making the sort of ethical considerations in the engineering ethics, which is actually has a long history in its own right, forefront in your definition, but going back to sort of talking to Trent and the original coinage of the term token engineering, one of the things that was interesting in a very early conversation I had with him about this was the choice of the term token engineering and its relation to electrical engineering, where in electrical engineering, there's sort of these units of information in the form of electricity and electrons flowing through systems. You use them to design circuits, you use them to design control systems, you use them to make stuff, but actually the unit of information is in the flow of electrons through those systems, and then there was this sort of mini debate about what to call it, and it kind of was defended as token engineering on the grounds that the tokens themselves are actually those atomic units of information, which is why it has a sort of analogy to other engineering fields where the sort of thing that's being named is the thing that carries the information in the system. Thank you. You just explained, I'm sorry, didn't you just explain cut, cut through the back door somehow to us? It's like because doesn't it like, so my understanding of the system is that it basically simulates business or like economies by like having single units flying through a system. I think that's later in your agenda, so I think I'm gonna try to keep on track, but we will talk about modeling. Yeah, we're definitely gonna touch on economic modeling and simulation so I'll have to have a little bit later. But just to round up the token engineering talk, is there is a big movement as well, starting up specifically this year, I've noticed they had their first token engineering hackathon I think a few weeks back, which is really exciting to see, there's token engineering global, and we're really starting to see more of an uprise than people actually putting conscious thought into how they're designing their systems now, as opposed to maybe the ICO craze. So the next topic that we're gonna talk on is curation markets. And I have here curation markets, reduce information asymmetry in the market through the usage of novels, skin in the game signals generated through the use of tokenized crypto economic incentive games. Does everyone agree with this statement? And if not, do you wanna come up? Or if you do even, do you wanna elaborate on why I've used this particular quote? Or anyone on stage right now? So I think maybe if we go more into like the, I think the most common forms of curation markets today in the engineering space are token curated registries and token binding curves. Yeah, so the ones that I have got here already are those two exactly. And then I said, so just because we needed to cover it quick to maybe understand those are fungible tokens and non fungible tokens. So NFT is like crypto kitties. Does everyone understand the distinction? If you don't, just raise your hand. Sorry, if you guys are in the light there. Who doesn't know what an NFT is? Okay, so from my understanding, a fungible token is one that can very easily be exchanged with another one. So like an ERC-20 token, I can swap one ether for another piece of ether. It doesn't really matter which one I have. But when it's a non fungible token, something very specific and unique like a crypto kitty, those, you can't really trade them one for another without it not being the same one. Just, sorry, I'm not really good at explaining, but I mean, you're welcome to. So what's, you're pretty close there. So what's in the name basically is the concept of fungibility. So the great thing about money is that it's fungible. So I can take maybe one year and it's divisible into like its parts, which would then be 100 cents. And so we can price goods and services at like varying degree based on the fungibility of money. Non-fungible tokens are non-fungible goods, but they're definition not fungible. So a house would technically be a non-fungible like asset or an artwork. And so this is what crypto kitty I think really pioneered was the awareness of, hey, we can use non-fungible tokens which are not divisible by nature and apply them to all sorts of assets, whereas most EOC20 tokens are divisible up to 20 decimal places or most crypto currencies are divisible. So that's the whole definition of dividing a Bitcoin into like up to one Satoshi is essentially the concept of fungible. So I would add that a non-fungible token is often characterized by its unique metadata or by its unique characteristics. So the reason your house is non-fungible is because I can't just swap it for another house, not even another one on your street. Whereas in the case of crypto kitties again, one is not really in any way meaningfully substitutable for another, but we do see a rise in fractionalization of non-fungible assets, which is to say you could take your house and actually fractionalize ownership in it at which point you might have sort of fungible slices of a non-fungible asset. So things get interesting and complex fast when you start combining these concepts. So when we talk about non-fungibility, I tend to emphasize the uniqueness characteristic and that generally is defined in terms of some metadata, if not, again, going back to the house. The data itself, the house is different from another one. So that's where I would put the emphasis for NFTs. Thank you both. Other crypto economic parameters that we're not gonna touch on today are prediction markets and stable coins, but I'm not gonna dive too deep into that. I think there's, are there any others from crypto economic parameters that are missing? Besides the fall that I mentioned. I think the main thing is that the vast majority of them are likely undiscovered because in reality, what we're describing are sort of unique mathematical patterns that result in some predictable, not predictable behavior, so much as predictable properties associated with those things. And because of how new this is, I think the vast majority of them are yet to be determined. Come on up. Yeah. Yeah, I won't ask you, just a one question. When you speak up on the picture. All right, I hate being a spotlight. Anyway, what exactly is a crypto economic primitive? Who actually coined this term? And what are its boundaries? So how do we know? Wow. I'm gonna, I don't know who coined it. It is in heavy enough use now that it shows up, I guess, without people realizing it. But I think I'm gonna argue that a crypto economic primitive is essentially a design pattern in the same way that like, you know, I won't, I don't, like think of like a chip in a circuit board that has a certain set of patterns. Like it doesn't guarantee anything at the higher level, but it's sort of a combination of, in this case, equations, but a combination of elements that it works the way that it works every time that you use it. And in a system like a, you know, crypto economic, you know, system, we for the vast majority of the time we don't actually know what we're gonna get when we start combining them together. But a bonding curve, for example, is one that I've worked with quite a bit and Paul has worked with. And there they have logical rules that are assertable. Like it's essentially a conservation law of sorts. And so when we use that primitive, we know that the conservation law is held. We don't actually know anything else. And when we start combining it with other things, we might get completely non-trivial sort of behavioral circuits with completely non-trivial properties, but we don't lose that sort of primitive conservation law. So anytime that you're building with something that has sort of assertable properties, I would say that that's acceptable to call a primitive. So smart contract is not a crypto economic primitive. I would not call it one, no. So it's like a more atom-like unit of a molecule than this is a kind of molecule, right? I would add one thing. So I think crypto economic primitives, like proof of work, I think as a design system, I would say is a crypto economic primitive. Proof of stake, in my opinion, would also be a primitive. Primitives are, I think, could be building blocks that enable completely new behavior. And as bonding curve started coming along, people were like, okay, this could be a new primitive. Although then I'd also argue like, you see 20s, that's more of a standard. It doesn't actually enable new behavior. I think we also have to recognize that we're in a very weak language space. So this field is so new that their terms being thrown out, coined, being like remapped. And in fact, I wouldn't put too much credibility on any particular definition in its current state. This is actually one of the reasons why we need a sort of growing, I would say, academic and research collaboration community because the standardization of terms is actually something that often emerges through a kind of research, even an inward-facing research process. Thank you. Thank you. I think communication is a big one. But moving ahead. So curation markets, this is basically saying the same thing. But market participants put their money where their mouth is. They say value and attention into the markets which they believe will be more valuable and the market's currency is a proxy for the attention. Early adopters are very often rewarded for their early attention to the market as the value increases. And this we can see through token bonding codes and are quite strongly. Any comments on these four points? It's, or any questions on these four points? So I think that in the case of curation markets, one of the ways I'd like to frame them is in terms of collaborative learning. So a well-engineered curation market is actually has the human's participating providing signal and the market itself is just an estimator of sorts. So there's some hidden private information on the part of the individuals and there's an algorithm combined with some reward structures which is job is to facilitate the estimation of that underlying hidden information. So in that way it's actually much more closely related to AI and online learning than it is to, I mean cryptography or even economics. Why? Because actually designing estimators is something that is actually predicated on optimization theory and sort of the, this is more than I wanna try to go into here. But in short, the idea is that you're designing a system where you are defining rewards and the assumptions predicated are that the agents will use their private information to take an action that moves in a certain direction, say increasing their potential returns. And as a result then if you only take as a prior assumption that they're gonna leverage some private information to try to profit along that vector, then what you're doing is trying to get them to reveal some preference or expose some information and that the purpose of the curation market being to curate is then to essentially coordinate that large number of private signals into a global estimate which is actually how sort of online and decentralized machine learning approaches are framed. So from a mathematics and from a design perspective it's much more closely related to AI even though what we're seeing is an emergent sort of behavioral economic phenomenon. So you would kind of go from design in the optimization and estimation paradigm to observing what people actually do and sort of measuring and studying and ideally feed that back on future designs. But it's engineering to behavioral economics back to engineering, ad infinum. So it's basically about, do I get you right? That's the way to make this a workable and a profitable in a sense system is to invent a way to analyze a crowd of these signals, right? A big decentralized entity of very, like, well, very large entity which the AI would do better than any kind of a human being or sort of. I think it's a way that I would more frame it is that it's sort of a cybernetic AI in the sense that people are not actually exposing their private signals. They are, it's a sort of thing where they're acting in a certain way from which you might infer their private signals but even still you're not even necessarily inferring their private signals. You're actually only inferring an aggregate estimate that is hidden in all of their private signals. It doesn't even mean that people have the answer. It means that through the highly divergent private signals of all of the participants that you hope to estimate some underlying true state of the world. And that framing is the framing of online learning and when I think about the curation markets I generally think about them in terms of these online learning problems because ultimately that's the goal that curation markets are trying to accomplish. It's still not clear whether we've effectively done that with any of the curation markets that exist today but in a sort of what is the objective of the curation market? Well, it's to estimate some hidden facts from a large number of actions of individuals who are taking those actions predicated on their private information. Frank, maybe you played devil's advocate a little bit. How do you think reddit communities or Instagram or the stock market are different than the systems that we're building in terms of curation? Especially stock market, which is exactly the same. I don't actually think that they're very different. I think all markets should be characterized as estimators but not all estimators are actually good at estimating. The thing that they're supposed to estimate. So take the stock market, for example. The financial incentives are not necessarily aligned with estimating the true state, so to speak. They're actually aligned with, they are incentivizing you to estimate the future estimate of the other players of the game. And that doesn't necessarily converge to an estimate of the state we're interested in. It actually converges to some shelling points which lead to some really weird dynamics that we actually observe. So ideally if we were designing new kinds of markets they would be designed from first principles to estimate the thing that we actually want to estimate rather than say incentivizing people to estimate each other's future estimates. But DCR is exactly about estimating the other's future estimate. Yeah, so that was the reason why I said it's not immediately clear to me that we've done a good job of this yet in this community, just that we're moving in a direction where we have the equipment to potentially design better decentralized estimators. By the way, in the prediction markets, like noses are more into the way you are talking about than the creation. A little bit, but most prediction markets, if they don't have a real observation of a relatively objective outcome, you still have essentially Keynesian beauty contest problems where people, so in a Keynesian beauty contest it's a like economics, like canonical example where people actually are incentivized to vote for what they think is going to be the outcome, not what they actually think. And so this is a broad problem in any prediction market that is predicated on something where there isn't a pretty objective measure of the outcome. Would it make you guys more comfortable if we had the stage lower down? How do we encourage more participation here? You don't have the light in your face. Yeah, I was worrying about you guys. This way it's less scary. So there's also this notion of semantic or linguistic arbitrage as well, which pops up on prediction markets. So the question gets framed in as objective a way as possible, but sometimes people find ways to pose, deliberately pose ambiguous questions, frame them ambiguously, and then take advantage of the asymmetry in the various planes of context, which these we later interpreted. So yeah, it settles all the way down. Well, it's sort of interesting, right, because you see people finding ways to turn what are supposed to be complete contracts into incomplete contracts by doing this sort of linguistic technique. And I think that this sort of reopens a lot of the questions about what is and is not objective and the extent to which putting something on a blockchain makes it objective. Because as long as there's a natural language component, there's always a subjective interpretation of that claim or of that outcome. Now you're reminding me of the Arizona statute, which Angela Ross likes to quote and talk about, which I can't recite from memory, but basically they're trying to give this definition of a blockchain in regional legalese of the various chambers of law in the United States. And this thing is so watered down to be completely meaningless, you know, it can be tokenized or a tokenless and decentralized or permissioned or free and can be secured by proof of work or not and there can be tokens or not. And so then you start to get into that linguistic swamp where I still think we're there. I don't think we have workable definitions of any of these things. You might be able to convince someone that your database is a blockchain based on that definition. Sure, yeah, there's also quite famous social networking outfits that have got projects that they call blockchains that might not be blockchains if you... Doesn't it still use cryptography? Oh, yeah, that's the definition. I'm a fan of minimum viable definitions, using the absolute minimum definition and then qualifying and specifying it there. Because in a situation where the terminology isn't mature or settled, people can be talking past each other. I mean, have you been to Twitter? That is structurally set up for people that do not use the same framings and meanings of words arguing past each other until they realize they agree. I'm gonna move ahead to the next one, but you're welcome to stay here because you participate. So the next one is we're gonna jump to token curated registries. So these are TCRs and basically what they are is a list. It's a fancy, decentralized list that is curated by the token holders, token holders. And this can be from anything, from names to hashes to records. Examples are like white lists, black lists, lists of the best universities, lists of the best science projects. And the holders of the token holders stake their tokens to perform an action, either adding something to the list, challenging something on the list, removing something from the list. And there's a whole bunch of different types of lists. So you can have a normal list, you can have a weighted list, you can have a list within a list. So there's a lot of different ways that you could do these registries. And generally the token holders will vote on whether or not to accept a new listing onto the list, the registry. How it works basically. So if for example, everyone here in this room wants us to start a list and we are all token holders and we're gonna vote on something science. We are gonna vote on the best science projects of blockchain for science 2019. And we have all now got the daily list of the top 10 that we all like and we've kind of voted on them and agreed on them. But then somebody comes in from, what science do you guys not like? Astrology. All right, then astrology tries to submit that their project was the best science project and they now stake maybe 50 of their tokens to try and get onto this best science project list. And everyone on our list, we kind of have a vested interest in making our list the best list because then it creates more value for itself. And it's also, it can't be then infiltrated by other like, let's say we want somebody who wants to bribe us to get on the list or they wanna have an unfair listing. And now we can also challenge this person who's trying to get on the list and say like, nope, you are actually one of the best science projects and we're not gonna let you on the list. He loses all of his stake tokens trying to get onto the list and we get the satisfaction of having him on. Exactly all like answers to science questions or to hypothesis, right? So this could be on this list, right? Yep. I suppose do you wanna elaborate a little bit more on your- No, no, I mean, we can like to have all the things we are talking here about these things because they can be used in science and research for a lot of things, right? To find a solution to hypothesis to the quality of data sets and create new incentive structures, right? Do you know about registered reports? Do you know about this idea of pre-committing to experimental designs and hypotheses prior to embarking on the experimental studies? This is quite a kind of a well thought of idea in certain experimental fields now as a way of like, so the big pivots and changes in experimental design are a bit more, people are more transparent about things like that. I'm just wondering you could maybe use a TCR type system to kind of commit to various goals, experimental goals. Cool, yeah, yeah. Not so cool, no. Why, why, why not? Well, I mean, I don't know if it was missing from discussion but the idea of the skin of the game, right? If I understand the initial concept of the TCR as it was proposed by Mike Golden, right? In the white paper, yeah, and two years ago. Yours, it's kind of more a coordination game than a prediction game. So you're not supposed to tell your opinion of this hypothesis, you're supposed to guess who will, whether the majority of people, of token holders or voters would vote yes or no. They're the thing. So it's more like, it's a weak spot or problematic spot that encourages like her behavior in a sense, guessing what the majority thinks but not expressing your own expertise. The value of what the token or the list or the registry viewers will get out of it. For example, like the example I think that was used in that paper was a Tucker truck, like the list of the best Tucker trucks. The list of the best Tucker trucks in New York was an example of one, being like, people would not want to have a badly put together list because what is the point of having a list of the worst Tucker trucks? It's a strong point, the self-regulating system of this. Yes, it's quite okay but my argument was about exactly the incentive to vote. So in order to, because you have profit in here, you have tokens in here and so you're getting profit if you go to the majority. But yeah. I'm going to try to answer this. This is completely right. If you like look at the system at one second, right? At one moment and freeze it there and then you would create a list and the hypothesis but as a scientist, you are like in the real world, right? And you're producing new results, new insights, maybe new data and then it changes and what the majority will like vote as a right hypothesis or together with you. Yes, it's like a creation market with a strategy. But this year it's a very binary thing that the problem is you are either in or out. Yeah, so I think, I mean you're describing what I brought up before in terms of the Keynesian beauty contest. So you reproduce that challenge and I think that there has to be a separation based on the goals of the system. So a system that is outcome oriented might be very different from a system that is sort of vote outcome oriented. So I think the actual outcome of the research project and what determines whether it was successful, what that's the kind of thing that we would want to be rewarding based on and there's some real open questions going back to the discussion of prediction markets about what is the like arbitrator of the outcome. But I do think that this use case that we're describing makes a lot more sense if any financial incentives that are woven into it or even potential reputational incentives that are woven into it are actually associated with the outcome rather than with the outcome of the vote. And this means that we're dealing with who decides under what conditions and what are their vested interests. We're dealing with this sort of, again this Keynesian beauty contest problem in the sort of using the vote as an indicator of what the quote unquote outcome was. So to be clear what I'm saying is if the outcome is a voted on thing then all of the voters if they're part of that same system also have vested interests. So it creates a very interesting even temporally layered sort of economics question whenever you wire incentives in. The upside is that we're making this a lot more explicit in the past these sort of vested interests were a little bit more hidden. The downside is that now that they're explicit we have to deal with them. This last thing I didn't get quite right. What do you mean by? What do you mean by vested interests of them becoming explicit? So not just the vested interests but in this case so reputationally for example if you have a process where people are say voting for whether something was important or impactful even say giving prizes in an academic setting. So people who are in the process of deciding that actually have non-trivial interactions with other people because you're part of the same sort of network. And so what I feel. The scenes arrangements you mean like. Well not even necessarily so there's behind the scenes arrangements that were like potentially attacks but there's also just like natural biases. There's a reason why for example short for a certain period after you finish your PhD for example your advisor is not gonna be your reviewer. There's so big web of interdependencies people that you collaborate with or that you sort of have site frequently might actually behave differently towards you from people who you don't site frequently or that when you do you actually say well hey like this person did this but I don't agree with that. And then you have explicit voting and you can say hey if I cite you you can support me that's okay that's what you mean right. Yeah so what I'm getting at is that there's a lot of sort of behind the scenes sort of incentives that have existed forever and that while we're dealing with some pretty complex incentive problems the upside is that they are explicit rather than implicit which makes them a little easier to face head on even if they're not immediately clear that we have solutions we make assumptions we make decisions and we say hey this thing is good in this way and weak in that way and we can kind of watch our blind spots but when we don't really understand the assumptions we're making or the sort of economic incentives are hidden then sometimes it's easier to miss that we have blind spots at all. So if I get you right you imply a multi-stage system in which the voting or TCR enabled voting is just one of the stages right. It's not the ultimate sovereigns who or what makes the final decision right. Yeah so I'm not intending to imply a specific design but to go back to what seems sort of comment about the specific process I am saying that it would make sense to have a temporally broken apart process where the first stage which was sort of signaling to support of certain methods, experimental methods, objectives, et cetera was separated from then the outcome and then the notion of the outcome would be a different process from the voting about whether or not it should be supported because there's new information right. You have the actual outcomes was it executed well maybe it didn't have positive results but it provided new information to the scientific community and then build service social service in a sense like a poll. I am saying that it could be a poll. I'm saying it could be a third party adjudicator. I'm saying at the very least you want a separation from the initial signaling that says hey we should do this to hey how did that work out and even if we do that we still have a risk of embedded sort of vested interests. Okay all right. Sorry that was a lot. I hope it was not too much. So I think it's really interesting to be able to dive into a lot of these topics. Do you want to? I would like to criticize one point on this list. Majority wins the vote. So I mean TCRs are nice. They are probably useful for some scenarios, some applications but if the goal is to actually get a good overview on what the social choice of everyone is and to get a consensus about what is the ordering in the list they are probably much better developed voting schemes existing which we know about from research from past decades which could not be applied in the real world in political elections because basically the cost is too high to perform them but thinking about the more advanced voting schemes I'm thinking about things like Schulze method for example which establishes a graph based voting telling method which will give you a very nice ordering of choices not only based on majority not only based on simple voting schemes like instant runoff voting which we have in the US primaries subject to spoiler effect people can win or items on the list can be considered popular in fact they aren't, they are spoilers and I think we should really think more about to integrate all this research from social choice theory and voting theory that has been done over the past decades and integrate them into these nice technologies because we can actually do it now but I'm not seeing it showing up so much yet I concur and I would say that this sort of is the it's the sort of other side of my point earlier about estimation theory so I view the sort of social choice theory as being what is essentially the social science history of the problem of basically inferring a collective choice from not just people's inputs but also a processing of them and then we have sort of on the more technical history this sort of estimation theory and that by revisiting them and ideally together we'll be able to view these as actual inputs being private information from individuals that is theirs to give and sort of algorithms or processes or voting schemes as means of aggregating that information into a social choice and that now that we can essentially program those things in a way that we can sort of trust that the methods will be executed genuinely we get to basically build on the whole history of the sort of voting systems and social choice literature and we may even get to sort of level it up with the sort of theory of decentralized estimation and signal processing. I agree and maybe to add one more point so if you ever work on a TCR and think about social choice theory keep in mind errors theorem and be aware that it has been proven there is no perfect voting system you can't build a perfect voting system so whatever you design whatever you develop whatever you TCR you build you have to choose for a flaw basically. So you will be sure that there is at least one flaw in your voting system but you have to choose the right one according to what you wish for. So funny thing is that this is actually one of their original motivations for why my slides earlier had this discussion about subjective choices of objective measures is because whenever we're entering into designing the systems we're making subjective choices whether it's which flaw you're gonna take or in a more broad sense I tend to frame it in terms of some sort of optimization objective which is to say if you pick an objective and your system is derived from an optimal estimate of something then there's a bunch of things that you chose not to optimize for and so inevitably the algorithm design is going to be a subjective choice even if it's framed as a solving for an objective function and that can be confusing when the term is literally objective function when there's no way to pick one that isn't a subjective choice. You got good hearted. Yeah. So good heart law is one way of framing the idea that anything which becomes an optimization target will itself be manipulated for so any kind of naive metric that you choose will be optimized by people that are looking to gain more arbitrage that system so that's the folly of going too far over to the objective side. If you rely on bare metal objectivity then you may find yourself in the good heart domain but if you're in the subjective domain then it's just your opinion, man. Yeah, if you measure a system you change the system, right? That's the same thing. Yeah, that gets even worse when we talk about observer effects and stuff. Exactly, exactly. And this is what I like about the whole blockchain and crypto economy for science scene because we can very quickly introduce new systems so it becomes very hard for researchers to manipulate them. It's the moment we have one monolithic system, almost one, it's like as you would have an open Google algorithm, page length algorithm and all the web pages could adapt for it, right? But Google is constantly changing it so the web pages can't adapt to it. And this blockchain, you can build very easily and quickly changing still objective, somewhat objective in terms of what you measure. Sorry, I don't want to get into your... Somewhat objective functions. Sorry, sorry, I used it in front of you. So, but at least like provable to the outside of blockchain secure or at least to build it very easily. So it becomes hard for researchers to game it because they have to change every year. And so they might end up doing what's best for science instead of for the system, right? Or... I have a blog post for this, like I wanted. Oh, and let me just take a look. There's just one example which is quite relevant which is the example of Minero, which is a Putes W, A6 resistant GPU, mine crypto currency. And they currently hard fork on schedule twice a year. And every six months now they have to change their mining algorithm because at first it was A6 and then now it looks like it's just really quickly tuned FPGAs. So the substrate is modifiable. So they're playing the game to change the rules but the bit of rule breakers also have flexibility now. So this is kind of like an arms race. Okay, okay. Yeah, yeah, be careful. Okay, good. Yeah. I'm gonna move ahead to token bonding curves in the interest of pizza. Yeah. Yeah. So yeah, our next topic or our next cryptocurrency is token bonding curves. And what I've got here is that token bonding curves serve as an automatic market maker to a token. The bonding curves issue its own tokens through buy and sell functions and where the pricing supply is set by the market demand. Price per token increases as the number of token supplied increases. So I've got this chart. Actually it's from the molecule catalyst landing page. I was gonna ask Paul to dive into it a little bit more. So I'm gonna hand over to anyone in the crowd or anyone on our panel that would like to dive into token bonding curves. I'm gonna proffer a definition at least or a characterization. So in my experience, the best way to characterize a bonding curve is as a conservation law relating the supply to a basically bonded amount of currency. So it doesn't matter what the other currency is. You have a sort of reserve currency that you bond in order to mint a token. When you burn that token, you're able to withdraw from the bonded pool and that what you're actually doing is imposing an equation that says that the amount bonded is equal to an amount of supply and that every transaction, regardless of whether you're bonding or burning, is going to preserve that relation. And while this is not immediately obvious, the implied spot price of this is actually the tangent to the curve in the state space representation, which sounds really math-ate obnoxious, but actually it makes for a much more coherent representation from a systems perspective. And then you're essentially designing that conservation law, the price dynamics, the supply and demand dynamics, and all of the rest come as a consequence of that one assertion. And then when people act on it, they're essentially moving along the curve and ideally finding a point that sort of balances out supply and demand. So functionally becomes an automated market maker, but the shape of that market is actually encoded in one relation between the supply of the token and the quantity of the reserve currency that's currently bonded. I'm picking up so much communication differences between the words that you use and the words that our team uses. So we use words like incentive pools for our reserve. And I mean, there's just an issue that comes, I think, with a lot of these new terms is that we could all be talking about the same thing and do different words, but yes. Well, if it helps, I'm working on trying to write some actual peer review papers related to some of the concepts. So I'm hoping that eventually those things will find their way into a canonical set of definitions. But as we discussed earlier, we're completely lacking them. So the language I'm using is largely derived from efforts to write formal proofs about the properties of the curves that we're using as opposed to, I think, something that's maybe a little bit more directed at describing to a user what their incentives are like. So the task of communicating formal properties and the task of communicating, here's what you're gonna get if you participate in this thing are actually different enough that they almost certainly necessitate some different language patterns. Yeah, very much so. I'm pretty sure when I asked the originally somebody else who's working on token bonding codes, I saw a handle to say anyone else in the audience that's working on a project that uses them. Are you? No. No. I mean, that makes sense. I'll do, yes. Alex. I just like, what's your question? Like ask questions, I will tell you. Sure, so what is your name and what are you working on? My name is Alex, I'm working in the AP.world, so. And I think bonding codes, it's like there's like basically nothing to add to gentlemen's already said, but it's just a good mechanism for illiquid assets, but which becomes inefficient when dynamic of the market changes a lot. And it's basically just an opportunity after that. That's it. Do you feel like they were, oh, you said illiquid assets, but it's very much a way to make them liquid. So for example, we're using what we want to or envision with molecules using it for IP increasing liquid models for IP. And that's something that I think is never be able to be done for. Yeah, so, but some IP are liquid, some are not. So, and if you recognize NIP, like as you said, like fractionalize in the, this asset, some of them might be very, very liquid and very high in high demand. Some of them might be not. And as I said, like I've been discussing this poll already, that sometimes it's useful to have a bonding curve and to like provide liquidity for people who wants to sell, but it's not always, how say, ideal mechanism for doing this. And also I wanted to add like to, okay, that's it. So basically, the difference between market dynamic from the beginning when you introduce a bonding curve and when market involves could become such a big that most of the like shape of the market will be so different from the boundary curve that like you will not be able to efficiently use boundary curve to raise a sell bias, which will be not enough to arbitrage. Prior to moving on, which I think we should probably keep trying to do, I will note that bonding curves are interesting in that they actually only define the sort of rules of the system. So there's actually a really important open question on any particular system about how to initialize them, because even if you have that curve, it's a bunch of points on a line. So where you are when you initialize it has a big effect on sort of even what happens with people interacting with it to sort of even the trajectories that it can follow. So there's a very serious initialization question that has to be answered for any project, even if they have decided on a bonding curve or even designed a particular bonding curve, you when you actually initialize it, it has to be somewhere on the curve. I think also a very interesting discussion you guys started is about a value, like token credit markets, token credit registries, and that you estimate future, estimate, or future estimate, and it's like infinitely. And I think like what's the problem with token credit registries, if you create it with assets like money, it's always gonna be like false and sensitive, almost always, because even what is a value? What is a value? It's always an estimate of our society and always depends on the context, on the dynamics, because money is completely artificial thing, so you just estimate some like, I don't know, vector of values, something like this, and in the dimension of where estimate happens, always changes. I think that that's actually a good segue for us into our next section now, which is why we're just gonna run through that. Yes, of course. I've always been criticizing the current economic system because it disregards the sum of the laws of nature and accepts that nature supplies are unlimited and that our consumers and growth and expansion are the only ideals. So in this kind of context token value, how can we make sure that we don't do the same greedy and fear-based approaches as we did with the current economics and then do something, come up with something more sustainable? So the short answer for me is that that lives at the level of the system that you are representing. So a lot of what we're talking about here is methods, tools, both things that are being tried and being created to represent parts of systems. But in fact, it's still more on the bottom up side of the discussion. I think all of these tools are put into practice in an effort to both represent and steer these sort of social, economic, physical systems and that the challenge actually lives at the side of making sure that you're representing something well and if you are not adequately internalizing externalities, if you're not recognizing where your system has boundaries with the outside world and sort of like what blind spots you get from that or basically the touch points between your system and what it encapsulates and what you know that it doesn't, you're always gonna end up with these kinds of arbitrages of externalities. So it's a little bit less about these tools and more about the taking a systems oriented view of defining what you're even doing with these tools. You mean? And so I think like this is the thing I'm like constantly thinking about like how, because like we're now at the stage with which potentially we can redesign the system like of the society to towards like how to say like, to make it basically better. And I think the reason why like our like society, our current civilization like is not like the best and like to sometimes like it's called radical capitalism and something like this, like too much about money driven and so on, because money is the only universal measurement tool. It's like the only which works for everything. You can almost anything measure with money because like it's a, actually you cannot anything measurement but all of us is measured with money right now. So today unlimited money means unlimited power. Yeah, it's something like this. So in, but it's like, it's not the most efficient measurement mechanism because like money is artificial. And I believe that incentive system works that if you measure something people will chase it. So if something is not measured, it's hard to chase it. So it's hard to chase like quality of your publication. If only quantity of references is considered. So an impact factor is also basically quantitative more than quality of metric. So I believe that the way to go is to measure things which we value, which is closer, like not to have universal like measurement mechanism but have a measurement for healthcare, for education, for scientific contribution and so on. And once we introduce it and once it accurate enough, once it good enough, like it doesn't have to be perfect, it just has to be good enough to how to steer our society to and align it to achieve the goal. Do we have an example or that comes maybe even a little closer rather than others? Sure, it's index of country development something like this. So this once it started to be measured, many countries started to chase it and like not to be like only how to say profit driven but also to consider how well as indication, how well healthcare was in your country and so on. And you have again, you can have this good heart's law which was mentioned already, like the whole metrics craze which is considered to be even more dangerous. If you talk about the good of society than the metrics craze, the craze for quantitative indicators is even a more, I don't know, error prone thing. Guys, I think we're actually talking about science and it's, yeah, we should, and I think if we talk about all these nice little mechanisms and science, there are two important points to be made or to poles, I would say, to opposite, I don't know, attraction, attractive poles of attraction. One is basically these things are in TCR and the creation markets are very formidable, very rigorous stock exchange kind of, or capitalism, economic style tools which are about maximizing profit and getting, well, well, getting profit, right? And another pole is basically what you were very insightful talking about. It's the end goal, it's not the profit, but the getting the information, not information, the accurate prediction to get what the people think, the internal oracle to get it outside, right? And to invent some ways to measure to gorge it, whatever, right? And so probably the second way of the second avenue that these mechanisms open is more interesting for science in my opinion because if you put the things together, like one is an engine for getting profit, the other one is the engine of getting the internalities, right? The internal things outside? Private signals? Private signals, yes. And probably at the intersection of those we have a very nice thing that I would like to talk about what is value? Actually, what is the value here that we are talking about? I mean, in business value is your, how you call it, balance sheet, right? Like either you are in the black or in the red. But here in science, what is value? I mean, it's a very sorry that Paul left us because his molecular is very much like, it's at least it's building a workable definition of value. A value is a potential commercialization, right? Comet of a certain scientific idea of a project, whatever, of a substance, whatever, so. No? Yeah, yeah, yeah, sorry. But it's okay, it's a workable definition, but not all scientific projects are made with the final goal of selling shares from a company that will manufacture a new drug, all right? Nuclear physics, zoology, history, you name it. Psychology, I mean, value is something different. And probably if you could talk about what kind of a value could we insert in the system? So I will comment that I think one of the challenges that we've run into is that there is no, and I don't think there can be one sort of even high dimensional representation of all forms of value that in a sense that it is in itself something that kind of, I mean, in some ways we could say it emerges from its own pursuit, that like the effort to achieve something is what is our best way of measuring whether that thing is valuable and that doesn't limit itself to just money. I would argue that this recognition that all models are inherently incomplete, wrong, or the famous Georgie box quote is all models are wrong, but some are useful, kind of brings us round to, do we have the model that is useful to the ends that we're trying to achieve? And then that becomes a sort of meta iteration. All we can do is sort of decide on some objectives, and then build a system to steer in pursuit of those objectives. And if they no longer line up with our sort of, what we feel like is progress, then we also have to have the ability to retune the metrics so that we're simultaneously benefiting in the short term from everyone sort of gaming the metric because that's what we want until the metric is no longer a good representation of the direction we want to run in, but we need the ability to essentially steer. Like we need a sort of navigation system, not just a propulsion system. But then that's just us, right? At some level that's people steering. Like we can't ultimately, we can't give a computer system the ultimate level of that stack, right? So no matter how much we automate, we're like trying to add layers, but leaving some sort of human decision making and governance for what direction we want to be moving in. Yes, the government see, yes. It's one very important term that you mentioned. So it basically comes down to setting the boundaries of a certain community of scholars, of stakeholders of any kind that set the rules, the goals, the values, and then exactly from this collective, in a sense, decision there would stem the more precise computer-assisted metrics, whatever, right? So I think in order to help wrap this up, I'm gonna make a comment about CAD-CAD since, so my team at Block Science developed this tool, computer-aided design. It's actually complex adaptive dynamics, computer-aided design. It's really a tool for computational social science mixed with engineering design. We have data scientists and control engineers and economists on our team, and what we realized was we wanted to be able to do simulations of the potential consequences of our design choices with sort of more open-facing view. So make some assumptions, then run some experiments, say, okay, well, if I assume that, then this will happen, and then change the assumptions, and run another experiment, maybe change the designs, but ultimately, we would be version controlling not just the designs, but also the assumptions, which is tricky because most of scientific work works that way, but a lot of the sort of economics principles actually take the assumptions as given, and they don't cross-check them. Whenever we don't know something, we should be saying what if A, what if B, what if C? And a good design is one that holds up, even if it's A, B, or C in real life, or maybe E happens, but hopefully E is somewhere in the convex hole of A, B, and C, so the likelihood of the system breaking is small. It's this sort of tool for combining engineering design with data science. If anybody is interested in it, we've got tutorials, and we've got some early examples, and some teams using it for design, including molecule, and excited for other people to be interested in integrating basically computational social science experiments into design workflows for token engineering, or even just for business model innovation. I can take a question or two, but I think we pretty much want to be done. Basically I think it's time, more or less, but we've also just recently done a really good AMA, so ask me anything with the CAD-CAD team, so if you do have any questions around that as well, so it's really good reporting. I think it's available on your side as well. Yes. Cool, yeah, so basically thank you everyone for participating. I think the last thing that I just want to leave you guys with is crypto economics and science, so how to use all of these tools in science now, and what projects are using them, what are the possible things that we could use, how would it change the way that science works and the different incentive models that we can use moving forward, but I'm sure that that's maybe a topic you can take with you into your beer and pizza, so yeah. Thank you everyone so much for joining the workshop, and thank you so much to all of you guys for participating. Everyone in the crowd, it was really interesting, and yeah, that's a wrap. Yeah, thank you very much, and this is cool, thank you. Thank you very much. Thank you.