Market Design Masterclass

Building tokenized products means bridging the gap between theory and practice. Before tackling the practical complexities of tokenomics, you need to have a solid theoretical understanding of economics. 

Our guest for this episode ticks both these boxes. He has a unique way of combining economic theory with practical solutions to real-world problems. Scott Kominers is a professor at Harvard Business School. He is also a part of a16z’s crypto research team, and an advisor to many web3 companies. Over the years, Scott has published insightful research on how tokens can help design markets and build reputation-based systems. His work is invaluable for anyone looking to deeply understand web3. 

We hope you enjoy this conversation as much as we did!

Full Transcript

Sari: Hi everyone. I'm Sari Azout and I'm joined here by Joey DeBruin. This is “Tokens, But How?”, a show where we dive into the messy details of building token based products. If you care about the ownership economy, but have lots of questions, join us. This show is made possible by Every, a writer's collective focused on business, and Seed Club, a DAO that builds and invests in communities. Today, we're excited to be joined by Scott Kominers. Scott is a professor at HBS, a member of the a16z crypto research team, and an advisor to many Web3 companies. Scott also authored, alongside my friend at Jad Esber, one of the most memorable essays I read last year on building reputation based systems. Scott has a way of combining economic theory and analysis into practical solutions to real-world problems. And so with that, Scott, we're really going to try to take advantage of the time you've generously given us today to help us solve some of the challenges we're facing as builders. Welcome, Scott. 

Scott: Thanks so much for having me. This is going to be fun. 

Sari: Yeah. Scott, maybe one place to start would be: what was the a-ha moment for you in crypto? You've been interested in market design for a while. I'm curious, what was the moment you realized that crypto unlocks something unique in this field? 

Scott: That's a great question. I mean, the interesting thing is, you know, because I'm working in market design, and I think a lot about new marketplace mechanisms and new transaction technologies, I've actually been aware of and passively following crypto almost since inception. And I started out as a huge skeptic. Like, I was on record as very skeptical of all of this stuff in large part because the early applications were sort of pushing decentralization for the sake of decentralization in ways that didn't really enhance products or, you know, sort of the design problems they were trying to solve. And decentralization is not costless, right? Like we have centralized institutions for a reason, which is that they're often more effective ways of coordinating effort or interaction. And so, you know, I had sort of written much of the space off, which was clearly a huge mistake. Not just an intellectual mistake, but also a financial mistake, right? Like, you know, I should have updated off of the number of really smart people I knew who were really excited about this thing and thought “I should throw at least a hundred dollars at this technology,” but neither here nor there. But like, what really started to change was when people started showing use cases where this technology was unambiguously more powerful. My friend Christian Catalini brought me, and one of my former grad students who's now a professor at Stanford, into the thinking around the diem cryptocurrency system and the broader idea that we could have decentralized transaction layers that sort of worked in the same way as the internet, but for transactions. This was like super powerful, you know…potential for financial inclusion across a lot of dimensions. Also just ability to reduce transaction costs across the board, right? We have really inefficient systems for moving money across borders, and even for most forms of simple transactions and payment processing, and there's a lot of power in using crypto technology to do that more efficiently.

But also it's a place where the degree of decentralization actually directly contributes to the value of the product. Right? If you're going to have backbone transaction infrastructure in the same way that we have for backbone internet information infrastructure, you sort of need to make sure that it doesn't have a single point of failure or just one piece of institutional support.

And in fact, the decentralization becomes a way of leveraging, you know, institutions in places with strong institutions and extrapolating them to places with weaker ones. And then, maybe one more light bulb moment: So then I started following much more actively. We were involved in this really fascinating market design problem. And I think the newer and much bigger light bulb for me has been with the rise of non fungible tokens and sort of other consumer crypto applications that make it possible to have markets in digital assets that previously existed, but you could never have markets in because you had no way of defining ownership. That’s a big opportunity. Right? Like, why don't you not argue about whether centralized or decentralized transaction infrastructure is the most efficient way to manage remittances, right? And at some level it comes down to institutions, the technology, but like…with non fungible tokens, we sort of immediately started seeing people able to transact in assets that previously were almost impossible to transact in within various sorts of small circumscribed contexts. So like, you know, you could have markets for digital items in a video game, but it had to be within the confines of the video game. You were only trading with other people who were playing this game because you needed a centralized, constructed world in which the trades made sense…that could define who owned which digital item. And similarly, it wasn't that we had never sold digital art before. Of course we were selling digital art, but like, in order to define the unique owner of a major digital artwork, you needed some sort of certifying entity, like a gallery to explicitly declare that the sale had taken place in a way that the community trusted. And the crypto architecture just makes this dramatically easier…and dramatically easier to do in a way that is interoperable, right? Like, you know, a given person outside the contemporary art market doesn't have a way to interface with the ledger of who owns which pieces of digital art. Whereas if you want to enter the crypto art market or even just understand how it works, or even pretend to decide whether you want to participate. All of that is transparent and accessible. And meanwhile, the transactions themselves are very easy to engage in. Right? Cause it's all backed by software.

Joey: Amazing. Yeah. I really liked how you made reference to the internet as this sort of like semi-permission information moving system across the global information moving system. And crypto has maybe like a version of that with, you know, with transactions, because I think it also helps kind of lay out how this isn't just totally a new thing that came out of the blue…came out of left field. That this is a product  of movements that have been in place for a while. And I remember reading a book a while ago, Stephen Johnson, this book “Where Good Ideas Come From,” which is kind of looking at these different eureka moments that people…you know, looking back saying, “oh, this was so amazing.” You know, Einstein's theory of relativity or whatever…and actually proving that these are…well…he called them the adjacent possible. It was opened up by the things that happened just before them. And now it's kind of like, so it's just one step forward. A really important one. You're one of the world's leading experts on marketplace, so I would love to just ask you to kind of lay out…almost like linearly…how do you see the progression of marketplaces that led us to the point where we have these kind of NFT marketplaces, or other kinds of Web3 marketplaces. I think you've already done it a little bit in what you've already talked about, but can you kind of put it in a way where people will understand the narrative arc?

Scott: Sure. So, first of all, we should step back one layer and ask: what do marketplaces do, right? Like what is a marketplace…it's about providing rules and/or infrastructure that address the sources of some market failure…and a market failure is any context in which a market is not creating the maximum value it could. And those can come from many different things. They can come from complexity of preferences, that just makes it hard to figure out who should transact with who. Maybe people want some very, very specific thing or they want to hire some very specific type of employee. And it's very hard to search for the person you're actually looking for. Right? Alternatively, it could be from hidden information, you know, when you're buying…And we also talked about the market for lemons in the context of used cars. Because it's really hard for an average consumer to identify whether a used car is actually in good condition or not. You know, we sort of, apocryphally think of people as masking, which cars on the lot are good or not. And as a result, many fewer used car transactions happen than could otherwise, right? You might be very happy to buy a high quality used car, but because you can't tell which is good or bad, you're not going to go and buy one in the first place. Marketplaces try to address some of these sources of market failure. And in the case of, you know, highly heterogeneous preferences, marketplaces might aggregate information and they might sort of create really nuanced search tools that help people more directly look for what they want. They might even create tools that do the matching themselves. Right? If the marketplace has enough of an understanding of the characteristics of its participants and what their participants are looking for, then maybe it just sort of decides like…you should match with this. You, you know, interview this prospective employee, or they might engage in verification or certification. Right? So a used car marketplace, you know, might repair the cars themselves. That's what Shift does, for example…or they might offer, you know, sort of high-end guarantees that say “look, if the car doesn’t work as described, we're going to insure you against that risk.” And these are all ways of like, you know, enabling people to transact in contexts where there might be issues with trusting quality. Early web marketplaces were mostly just about providing a context for supply and demand to beat without any structure to the transactions at all, right? Craigslist perhaps is the simplest example, right? Like this was literally just a place people could post supply or demand for all types of things, right? Used furniture or apartment rentals or whatever. And the site provided essentially no infrastructure other than some very course categorization that the users themselves chose, and then a communication tool that lets you get in touch with people, which was an email alias. Soon though, marketplaces started realizing that they had high level information about the market that they could use to add value to the transactions themselves. 

They could do things that were more efficient, like providing insurance, right? A marketplace platform is better positioned to ensure individual transactions than the individual participants might be because it has economies of scale and provided the insurance product. And then there's like a next wave of marketplaces which has been building tremendous amounts of infrastructure, right? I used to talk to my class a lot about a company called Gamer Sensei, which is an interesting case. They created a market for e-sports coaching. Sort of like personal training for e-sports. You know, one of the questions that always comes up is like, how do you stop people from disintermediating this platform, right? You're forming a long-term coaching relationship with somebody and they're taking a fee on every transaction. Why wouldn't you search on the platform, find a person to work with, and then both of you go off the platform. And the answer is that they provide a really great suite of tools for the coaches and they guarantee payment processing in a way that the coaches really like. So even if the user says, “look, we're on our seventh lesson, why don't we just take this off platform?” The coach is like, “no, no, no, we've got this great communication platform. The payment processing is all managed here. I've got all my accounts managed at this company.” The people don't want to disintermediate because the platform itself is providing so many services that make the transaction better. With Web3, we're now getting to a layer where the trust certification in marketplaces can often be sort of managed through the Web3 tech rather than through the platform brand and its own sort of private infrastructure. And that means it's much easier to sort of enter and start to become engaged in a lot of these platforms. Right? Before, if you wanted to watch a fundraising campaign or something, you often had to rely on some centralized intermediary to provide the brand and maybe the verification and vetting of you. Now, so long as what you're doing is on chain…and again, there's all these challenges around the relation between the on-chain activity and the off-chain claims, right? Like lots of people are doing these things like rug-pulls where they sell an on-chain asset that behaves exactly as predicted, but the off-chain associations with it sort of fall apart very quickly. 

But to the extent that you can do things that are on-chain or otherwise sort of like, you know…you can generate trust among participants yourself. You maybe don't need marketplace platform brands to be managing the trust process. And then the other thing is there's much more interoperability now, right? Like people are building marketplace layers on top of this like sort of underlying infrastructure layers that are the blockchains. And the fact that the technology is interoperable means you can do much more with it. It's much easier to bootstrap, right? If you're, you know, launching a new NFT platform, you can immediately leverage all of the existing NFT ecosystem in a way that if you were launching a new digital art gallery before in the web two world…like you start off…you have no content. There is nothing there. You have to convince people to bring their content. Here, you can immediately leverage all the content that's there because the underlying technology’s interoperable. 

Sari: Scott, your essay on reputation systems argues that systems should be designed using a two token system, one for signaling reputation and the other for offering liquidity. And so the second case on liquidity would imply that platforms should allow their participants to convert that reputation into some sort of financial reward, presumably. And so I want to take a step back and get your take on why you think that is important. 

Scott: Great. So first of all, for framing: In that article, we were sort of starting from the assumption that the platform actually wants to be able to allow people to convert their reputation into some degree of reward. We're not specifically pushing that you should always do this, but yeah. A lot of Web3 platforms are in one fashion or another trying to do that. They're trying to establish a reputation system that people can also monetize loosely, whether it's somehow directly turning it into cash or turning into like value in the ecosystem that they can trade for other things. But I do think that's often valuable. I'm an economist, right? Like, you know, my inclination is to believe that incentives matter for most types of interactions. And if you want to encourage people to participate in your ecosystem, having some way for them to share in some of the value that is created through their actions is a great way to incentivize it, right? It aligns their incentives more fully with the incentives of the platform. And again, it doesn't have to be money necessarily, but like doing something that gives your most active platform contributors or users a way to share some of the upside that's internalized by the platform, you know…incentivizes them to invest more actively in engaging with the platform and possibly even gives them a desire to build the community around the platform as well. And so in all of those senses, I think giving your users a degree of ownership actually is important. And it's just the simple idea that if you are a part owner in some fashion, you internalize some of the benefits of success, then you're much more strongly incentivized to help create that success. 

Sari: Something I wanted to ask you related to that…The moment that you introduce a financial reward..I always think back to this Dan Arreola quote which is “people will help you to do things except if you pay them to.” I also think back to this anecdote from Sarah Tavel from Benchmark, which said that if Yelp were to pay its contributors, the business model wouldn't really work because you couldn't pay people what they're actually worth for the things that they're willing to do for free. And so if a system like Medium or Twitter, whatever, were to create some sort of token where…maybe I got a token that was worth $30 for my tweets. Could that actually detract from my usage of Twitter? Because I feel that I am worth a lot more than that. And so I guess my question is: Can we get the math to work on those things? 

Scott: I mean, first of all, it certainly varies from context to context, right? There's an altruistic crowd, right? Some people do things because they want to do them for free and actually, like, financial wise that makes them unhappy about it. I worked with Quora on some of their economic design questions. And they introduced a content monetization program and some creators opted out because they create content for Quora because they want to do so for free and for fun and making it financial doesn't address why they're involved. And this is like a general lesson. You have to know your audience and your users, and sort of think hard about what it is that makes them want to participate in your platform and what their underlying motivations are. But at the same time, rewarding people in a way that's consistent with their motivations I think is very powerful. Right? It's not the case that Yelp didn't provide a reward to users, right? They had profiles and they had, like, how many restaurants they'd reviewed. I have friends who had written many, many, many Yelp reviews and they used it as a form of social capital. They would post on their websites, like, “Hey, and if you want to know what restaurants to go to check my Yelp reviews, I've written a thousand of them.” This is a form of social capital. It's like, “Hey, I've been to a lot of restaurants. And like, I've thought really hard about them.” And both of these are things that are, you know, maybe a little humble braggy, but like, we're a form of social reward for the activity they were engaging in. Again, not the only reason people write reviews on Yelp, right? People write reviews because they want to support restaurants…because they're annoyed at certain restaurants, because they just like writing reviews, you know…the process. They like having their own record of the restaurants they've been to. All of these different motivations. I think in many contexts…and I'd argue probably even in these review systems…if you create a way of rewarding people who are active and engaged and participating, you can generate and retain engagement in a way that’s hard to do without any sort of direct incentive, right? Lots of people get into editing Wikipedia or something for a short period of time and then stop. For a while it's interesting, it's entertaining. You're contributing knowledge, but there's very little feedback loop. Right? You don't get much of a direct benefit, even in terms of like, seeing other people learning from your contributions on Wikipedia. Whereas, I like to think if you had a version of Wikipedia where you could discover how much the things that you had written were being read…even if there's no financial component, just like a scoring system…like you see your edits were read 5,000 times this week. That alone I think would be a powerful motivator. And the second you have a motivator like that you can start to shape platform incentives around it. People would start trying to edit the articles that were likely to get more readers, or that they cared the most about. They probably already care about the ones they care the most about being correct, or the ones where they have the most information, but now you would have them tilting those targets on the margin also towards the ones that get more readership that are more important for the platform’s overall role in society. And so I think these can mix very nicely. Like it doesn't have to be that offering rewards is in conflict with people's desire to do things because they enjoy them or because they’re the topics they care about.

Joey: Yeah, it's interesting. I feel like sometimes liquidity is like a bug in these systems, but then it always raises the question for me of like, why token? So for example, like Reddit always has the karma system, which does kind of what you're describing. It's like the more you participate, the more value builtin the kind of network, the more karma points you get from other users and those karma points have utility within the platform as well. You can do more things. You can have more impact. And those are highly desirable. Like if you were to, for example, be able to trade those kinds of karma points on a secondary market, you might open up some revenue stream for highly active contributors but it would come at the cost of also people, theoretically, being able to then buy utility in the system and maybe decreasing the net utility of that system at that point. So basically like internally, this system works super well. As you like, make it something that's inter-operable and has this liquidity to it, it actually gets really complicated. 

Scott: This is why you need a two token structure. If you have a single token that is both the marker of reputation and the thing that is sort of the fungible currency that people are trading, then it sort of fails at both of these roles, right? If it's a really good marker for reputation, then people are going to buy a lot of it. Which means that now if someone's holding it, you don't know whether they earned it or bought it, which means it's actually suddenly no longer a good marker of reputation, which causes the value to crash. And so it's like this funny paradox that like, if you have one asset that is supposed to serve both of these roles, like in equilibrium…it serves neither. And so we argue that your reputation asset really should be like a non-transferable fixed thing attached to you. But to the extent that it's spinning out utility tokens or something that others might want to trade, that's like an emission schedule attached to the reputation asset rather than you trading the reputation asset itself.

Joey: Yeah. Maybe I can just push you on that one a little bit, because I think that the inner designer in me is always like: “We need to make things way simpler rather than more complicated.” And I think the reason that all of these reputation systems work…So whether it's like ‘my Twitter reputation is my number of followers’ or ‘my reputation is my number of karma points.’ And those things have clear utility. So the reputation comes from the fact that you can do something with it. So, you know, it makes sense to me that you would want to build reputation in a way that is attached to you and that has no liquidity. And because that's more of a mark of your position in the network, it's not clear to me why people would care about that reputation, if you don’t use the tokens that are liquid to do stuff. 

Scott: Good question. So it's not, to me, that the reputation asset can't also do things. Right? And even just in terms of real world utility, right? If you're like a top poster on Reddit or Quora or something, the fact that you have this feature is something you might show to friends, or like, you know, using the social capital sense as well. And indeed, right? Like high reputation contributors. Like you might want to give them special powers. You might want to turn them into moderators. You might want to sort of give them additional access or roles in the platform. And all of that, I would still think vests in this nontransferable reputation asset. Exactly as you say. It's the extent that you'd want to additionally give people an opportunity to liquefy. You know, earn off of their reputation within the platform ecosystem. There, it might be that you have a utility token that's like used for sending extra messages or adding cool badges or animations or whatever to your posts or something of the sort. And maybe that's a feature you want to give to everybody, right? Like you don't care whether only the fancy posters or the average posters get to add cute animals to the animations. But you want to restrict the total amount of that so that your platform doesn't become just animations or something of the sort. That's the ideal thing that's moderated through a utility token. And a benefit of having such a token in the system is you could give that to the high reputation people as a liquid component of the reputation reward that they can then use themselves if they want, or they can like, you know, sell to somebody else who wants to add the animations or whatever.

Sari: Something. I want to ask you Scott, because that makes a lot of sense to me, but what gives these liquid tokens value? So in the case of Reddit, there are karma points. But then, based on whatever complex economic design system, you would then issue these liquid tokens. Would they in your mind be tied to the cashflows revenue model of these businesses in any way? Are they governance tokens? Does this only work in a model that is sort of token gated? Uh, because I think, you know, where I stand, the reputation piece actually makes a lot of sense to me. But where I'm still struggling is people assume that creating a token generates value, but the value sort of depends on what gives value to those tokens. And so I'm curious: What do you think would drive value in the case of like, you know, Instagram giving tokens to get the accounts, the most followers or Reddit, in the case of people with the most karma. 

Scott: Okay. So first of all, you highlight a really important high-level point here, which is that a lot of people assume that if you add a token to a system, then suddenly you've added a bunch of value and that's totally wrong, right? Adding a token is not the source of value. The source of value is the things that people do with the token or associate it with. And if you add a token without thinking really hard about the economic design around those associations, you've added something that is like no value added…might be net negative value. If it causes people to lose…people speculate like crazy and then lose confidence in you as a platform architect. So like, you know, that’s a really important point that a token has to have a purpose and a value framework before you can decide, you know, how many of them you should give out. Right? Indeed, the ICO boom and bust was like a good illustration. Just like tokenizing things…even tokenizing things that might be valuable…some degree of ownership or control of startups…if it's not thought through carefully, doesn't actually add value to the architecture.

Um, now all of that said, the other key point is that the use of tokens varies a lot from platform to platform, right? You could imagine a variation of Instagram where all of the advertising and post promotion was driven through a token currency rather than through dollars. Right? So if somebody wanted to boost their posts outside of their circle, they had to pay the internal economic unit, whatever that is. And you could imagine the platform to do that as a way of controlling the amount that people were doing this boosting, right? Like, you know, this is an issue, right? Like people will often create spam ads and stuff, and the platform has to work really hard to make sure that your ad quality is balanced with the payments they're receiving for these ad. You could imagine doing a purely token based version where the shadow cost of posting an ad was sort of in your need to spend this token. As the platform grows, the demand for advertising presumably goes up so there's some downstream value to the tokens as well. That would be a context where it's really natural to think like, okay, we've got some top contributors, they're going to accrue tokens as they contribute. And then maybe they use that to boost their own posts, which raises their position and sort of advances their future contribution, visibility, or, you know, they have the outside option of selling them to others who are trying to bootstrap, and haven't yet built up the on-platform reputation. Similarly, tokens attached to just general levels of engagement…like you think of like Twitch, you know, gives you…I forget what they're called, but you get some points for watching a stream actively…every so often it’ll say like “click here and collect more tokens.” This is in their internal ledger. It's not a decentralized token system, but again, this is a way of sort of maintaining people's contribution. And then it gives them micro rewards. Like if you have these like units of token, you can post some funny message on stream. Everyone will laugh or something like that. In the context of, I guess the Reddit model… my instinct is it’s pretty similar, right? You would use this again to like elevate content or maybe to request specific content or something, I suppose. And again, you can imagine people who have no desire to create content themselves, still having the desire to own the ability to request specific pieces of content. And so they might collect tokens…like sort of a token fundraiser. You know, pull together 10 different people and their tokens together to request some fancy post of some form. And all of those are contexts where you can imagine a stable token ecosystem. Again, if you get the monetary policy, right? You can imagine the stable token ecosystem of people buying and selling and using the tokens all at once. 

Joey: Yeah. So I think that those things all make total sense. And one of our core thesis for this podcast is that there's like a lot of steps between the kind of idea and actually like putting it into practice and really, you know, implementing it as a builder. And I think we were not mentioning a lot of different platforms that, you know, make sense and maybe you have different kinds of models, but maybe we can really go deep on almost like a design exercise. You know, we're going to miss lots of little things, but let's go step by step through an example. I think we were talking before the podcast started about LinkedIn, which is, I think, a fun example. People love to either love or hate LinkedIn or maybe some of both. So it'll be, you know, certainly exciting. And maybe you can describe it a little bit, like how you think implementing a token system within LinkedIn would work and then we can go step by step past that.

Scott: Okay. So LinkedIn is a fun example. There's a lot of information on that platform, but it's very hard to determine its, you know, relative accuracy or importance, you know…cause there's very little trade off, right? You can put whatever you want on your profile. People can like, you know, write whatever skills they want so let's think about the skill system perhaps. So on LinkedIn, you know, if you want to be certified as having some skill. What you do is you ask someone in your network to write a little blurb about why you have this skill or whether you have it. And indeed, LinkedIn sometimes solicits these without people even requesting them. Right? Like, you know, when I open the app, I get like, which of these five people do you think is better at coding in Python, but the problem is there's no real, you know, opportunity costs of saying that one person is good at a skill, you know, versus another. And on top of that, nobody knows how good I am at assessing a skill. Right? So like, if I read an assessment that this person is like an absolutely brilliant economic thinker, I like to think that that has some meaning. And like maybe somebody, if they go and look at my profile and say, oh, he's a professor of economics. Maybe he actually knows what he's talking about. You still know how to benchmark, like, am I overenthusiastic? Am I super cynical about people's abilities? You don't know how to benchmark me, but you at least know that I might know this. On average, if I see one of these skills, I know very little about what it means and I don't have any architecture I can use to interpret what it means. Whereas you could imagine that like, you know, If I were regularly providing these skills assessments, first of all, there'd be some cost to the marginal one, right? Maybe I can only provide one a week or something like this, or I need to pay tokens or currency or something in order to get the right to produce them, which in and of itself already introduced some signaling value, right? That someone has faced a trade off. I'm going to attest to this person's skill rather than some other person's skill…that's meaningful. And then like, maybe like I accrue some reputation in my ability to make these assessments…again, maybe in general, maybe stratified by like skill topic or something, you know, perhaps through other people, like down the line, you know, like rating whether that recommendation was helpful or that information worked right.

Like if I say someone as good at, you know, economic thinking and they later get a job as like a research economist. Maybe I get a little bit of a bump because I tagged this fact that seems to be sort of borne out in their hiring data. And so that's a reputation loop. I know you're going to ask me, so let me just start to say where a utility token could come in. For that second token component…there's like a lot of stuff on LinkedIn. First of all, the platform itself literally wants to control a little bit so it doesn't become too spammy, but like…for example, InMail, right? If you're a recruiter or somebody you can send messages to anybody you want..you could imagine attaching a token cost to that too. So maybe the currency token is used to buy the right to send messages, contacting individuals who haven't opted into like general contact or something. And so now if you're like a really high reputation skills evaluator or whatever, you also gain these tokens that enable you to reach out to people as you wish, or you can trade to other people to reach out. And like, that's a nice loop. First of all, because we actually think these skills are kind of correlated. The value is kind of a line, right? Like if I'm someone who's good at a test of your skills, like I indeed might also want to reach out to people. I might be accruing a currency token that's useful to me. But also to the extent that I'm selling it to others, I'm getting rewarded for my work in making it easier for the people who are trying to contact other people to figure out whom they want to be in touch with. So it's a nice reasonably tight loop and all the different incentives…Again, this is off the cuff, right? This is not like, you know, affordable design. 

Sari: Yeah. One thing I want to touch on is, as I was listening to you, describe all of this stuff--There is no cost to the marginal endorsement. I can think of a lot of Web2 ways to enable that. And, your example of InMail…you could issue credits. A lot of this stuff has been done before. You could say, all right, you can only get five endorsements unless you have some sort of feedback loop that as you predicted…the economist got a job. And so I'm curious, in some ways it seems like getting the monetary policy right, and also liquidity could be a feature, but it could also be a bug here. And so I'm curious if you were advising LinkedIn and they were sort of pushing back because they wanted to do all of this stuff, but it made more sense in Web2. What would be your argument? 

Scott: There is such a thing as a Web2 token. An internal currency in LinkedIn would be a token. Right? You could easily imagine that LinkedIn would implement this with an internal token, and so long as there's a market to buy and sell that token into cash or crypto or whatever…in some sense you have enabled a lot of the features that the Web3 implementation would provide. The Web3 benefit, I think, is much more about interoperability, right? If you wanted to enable people to build on it or offer other uses for this LinkedIn token. Maybe people don't want to use the token to send InMail, but they actually might want to collect it and use it for access to their recruiting fairs or something…or give it out at recruiting fairs. Harvard University holds an annual career fair type thing, and lots of universities do this. Maybe the universities would like to be able to own some units of that token that they could distribute to students to contact like potential employers or something of the sort. Maybe LinkedIn could sell that as a service to universities. But also, if LinkedIn just had this token floating around…in a Web3 interoperable ecosystem, some third-party could build that entire product that's much more optimized for working with universities. They can now be like, you know, acquiring this token, distributing it to their students. And then those students come back and use it on LinkedIn. And this grows the value of LinkedIn’s ecosystem, and does it in a way that is sort of very much leveraging the interoperability of the platform. And so my instinct is more that…It's different if you're a small platform that needs the transaction infrastructure layer as a way of establishing trust in the token’s existence and sustainability, right? People would trust LinkedIn if they said we have a ledger of your InMail or your InCredits, because LinkedIn is an established platform,  just like how we trust airlines with their miles. Nobody argues about whether their mileage balance has disappeared, although we do know they sometimes devalue them. So you don't need the Web3 architecture to create trust in the token’s existence and security and tradeability. But I still think there would be a big add in terms of interoperability and use across different contexts.

Joey: Yeah. I think there are some really interesting points in there. I think in the LinkedIn examples, if you really look at the features…it's very smart. Very specific things that you do as a recruiter are paid and everything else is free. It's certain ways that only recruiters would search for someone… like those things are paid. So it gets very, very precise in terms of how they actually design the product. Another way of thinking about these utility tokens is there is a segment of people that are extracting value from the platform, maybe not adding as much. And can you actually charge for them? 

Another thing that as you were talking, you made me think is…I wrote an article actually saying cost is signals of upscale is why I think these kinds of attestations don't make sense on chain. But there's no reason that LinkedIn couldn't have scarcity to an attestation, right? Like, so you're saying, “Hey, well, if you get only one per week, you get one per month” or whatever like that…that was a design decision that they made to make these things totally open. And similarly to charging someone for InMail, they could have designed an internal currency that you earn and then you can use that to unlock features. So like none of these things require Web3 at all. It seems like the last thing that you mentioned--interoperability--if I create this system for reputation in LinkedIn, or it has liquidity and other people can actually build on top of it and it becomes the sort of ecosystem of professional reputation that everyone is using. So it seems to me like that is the most obviously unique aspect. And I'm curious, how do you think about defensibility? Because in some ways, LinkedIn is so powerful because they own all these systems reputation and you can't get access to that if you're another platform. So if you're advising LinkedIn, outside of the benefits to the whole ecosystem, how do you think about the benefits of building that kind of inter-operable marketplace for reputation from a business perspective?

Scott: That’s a great question. And again, it's not in all cases that the main value out of the Web3 is this interoperability component, but here exactly in this example we've been talking about, I think it is. Web2 platforms have, by and large, been about owning data and locking in users as their principal source of network effects. Your ability to have all the transactions flowing through your platform and have better information than all competitors in terms of what content you're going to serve. These were sources of competitive advantage on Web2. And to some degree they will remain sources of competitive advantage. Web3 is not just going to dismantle all of that, especially not overnight. I still think there's a huge amount of value to producing technologies that are better aligned with your users in the long run. In some funny way, you're giving up surplus in the short. But you're gaining a lot downstream. Jen and I were writing a different article. It's precisely about this. That's like: once you sort of turn your user base into a community, that's actually like invested in your platform and like, you know, forming a coherent, engaged community around that platform. Suddenly now the technical platform can do ultimate gains…you have users who really see themselves as a part of the entity and invest a lot more effort and energy and like growing the value of the platform itself and getting other people to participate. I think in the long run…at least in many contexts…is actually an even stronger formula for success, right? I doubt there are a large number of people who use bank of America because they actively self identify as a bank of America user. Right? They use it because it's an efficient service, you know. Web3 gives us contexts where people use like specific platforms, right? Like, think about UDA swap versus sushi swap, right? Like these are essentially identical platforms and people choose which one to use. They actually have brand preference and identify as one of these users. And that's just in a sort of silly financial exchange context…think about that extrapolated too, like relationship and contact networks. If you want to build in one of these ecosystems, you're going to create much more value. And the ecosystem itself has higher potential value than one where you're sort of constantly being extracted from. 

Sari: What do you think are the implications of that on the size of companies going forward?  Like there's an argument that they could be smaller because everything could be infinitely forked, but there's an argument that they'd be larger because it's interoperable that are aligned. And so do you have a view on, you know, maybe there'll be less venture scale, you know, businesses in the coming years or maybe more…

Scott: It’s very hard to predict. My guess is, because of the way the technology works, we're going to see many more things grow to large scale with smaller teams, because some of the things that teams were actively doing before it can now be done with smart contract type source code. My guess is we're still going to see some very large platforms, but the bargain they strike with their consumers is gonna be different? The potential for vampire attacks, right? The idea that you could move all of your NFT transactions, just from one platform to another, because they all lie on the underlying chain, not actually like on the platform itself. That means that the consumers have much more power relative to the platform than they did in Web2 ecosystems. And it means that the platforms are going to be differentiating on quality of design and user support and rewards and like engagement in the platform. Like they're going to be differentiating things that are much more consumer focused and creator focused rather than sort of trying to just have this ecosystem be so large that it can't collapse…because now the ecosystem can be so large that it can't collapse, but it can still all just like walk away…And so the sharing of surplus is going to have to move and recognition of that. 

Joey: Yeah, totally. It's something that I think about all the time for Backdrop, because we are essentially relying almost exclusively on open data and data on blockchains to build profiles or to facilitate the information that's flowing through projects. And at the same time, it's impossible for us to store everything publicly or put everything on chain. People are always like, “oh, you have these things that should be open and interoperable. And we're always thinking about how to do that, but it's an impossibility that you can build a platform where every single thing is fully open. I think about Link Tree as an example. And I remember listening to a podcast about Link Tree, where this is a sort of a system that helps you showcase all these different links that you have, right? Like your profile here, your newsletter subscription there, you know, whatever. And they are at some point thinking about building their own newsletters. And if they, for example, build their own newsletter system and then make it harder to use than any other inter-operable newsletter system that would basically just tank the user proposition of that product. And so I think about it the same way for us. Like if we can build our own internal tooling, but if we actually like are inhibiting the utility of like an open piece of infrastructure, then we we’re basically tanking the value prop to the user. And so, and I think that's like the dynamic that all of these platforms will play. OpenSea is another good example. Right. They're always skirting with that edge, right? Where they have to do a ton of stuff first party, but if they feel like they are actually like building this moat that will create these walls to interoperability with other platforms, then they could see mass exodus. So I think it's super interesting to see how this will all play out 

Scott: 100%. Totally agreed. And it’s people like you that are going to shape this future. 

Sari: Scott. It strikes me that, you know, there's a lot of promise in this space, but you know, I have a hard time pointing to very sort of transformative projects, DAOs, communities that sort of live up to that promise. Would you agree with that? And if you would, what do you think is keeping us from building that sort of more transformative value? Is it a function of, you know, ‘the next big thing starts looking like a toy?’ Uh, or what are these other complexities at stake here?

Scott: That's a hard question. I certainly don't want to generalize across the complete space of projects because I don't have visibility into all of them. But I mean, I think the simple answer is it's very early days, right? Like this technology is really quite new and we're not even done with the part where we're just taking existing types of transactions and making them more efficient. If you think about the emergence of the early internet, right? At first it was a lot of technical confusion that was not for everybody, but for most people it was so confusing. There was no way to access it. And then it starts, you know, becoming able to do a small number of things, much more efficiently. And then gradually people came up with the transformative innovation and the new business models. Just like a general purpose technology takes much longer to diffuse than a specific purpose technology, right? If you invent a slightly better widget for use in a factory, you know, everyone just switches out their widgets tomorrow, but like here, we're looking at a totally new way of transacting and like owning and creating. I think it seems to me very clear that it is going to transform a lot of our ordinary interactions. And then lie the things that it does on top of that, it's like hard to envision, but like many of those aren't going to come into place until some of these basic ones have happened. Right. Before we see the completely novel use cases and technologies, we're first going to get a really robust way to manage event tickets. Cause we don't have a robust way to manage digital events. So, I guess in my mind, this is already transforming a lot of the interaction forums we already have that are not very efficient in their digital forms. And it's creating abilities to do types of transactions we never had before. And then it will, you know, as it matures and as we see more and more adoption of crypto technology, we'll see it. I'm confident we'll see a deuce of like really cool stuff that we can't even dream of. 

Sari: I didn't want to wrap up, Scott, without asking you something that Joey and I have been discussing on the podcast. Do you think that the most powerful platforms will be Web3 native or will we see a lot of platforms build these sort of two token systems like the one you described on top of platforms’ data that is Web2? 

Scott: I think we will see a bunch of very powerful Web3 native platforms, where the features that Web3 brings are essential to making the platform work or create the most value. And indeed, I think we'll see replacement of some Web2 platforms along those dimensions. I think there's a lot of strong arguments that Web3 creator platforms are gonna do better for all participants involved in the long run then a lot of Web2 creator models do. So I think we will see large scale Web3 platforms. It's also like, at different levels of centralization or decentralization, right? Like decentralized design is not the way to build everything. And then, you know, we'll also still see Web2 platforms in contexts where that technology is right, right? I don't think like, eliminate the entire old model. There will still be some competitive advantage to be had around user aggregation and so forth. There will be constant market pressures to give more portability, more interoperability as people start to see that in more and more of their interactions, right? Like if you're the one platform that doesn't let you take your digital assets with you and use them on all the other platforms, that's going to become a source of competitive threat.

Joey: Yeah, absolutely. That tipping point to interoperability. I think it's so interesting. Well, Scott, I'm sure we could ask you questions for hours on end. So appreciate you coming on the podcast. I think you have such a unique ability to kind of straddle the clear thinking and academic side, but also just building stuff and exploring. It’s going to be so exciting to follow you for the next few decades. So thanks again, so much. 

Scott: Thanks. I'm very honored and I'm not sure I can live up to that close, but it's been a lot of fun chatting and I'm looking forward to watching what you guys build, you know, in the next decades or in Web3 times…that really means just the next like six months.

Sari: You've been wonderful, Scott. Thank you.