3 mental models for crypto infrastructure | Ali Yahya

Transcript

[Music] thank you so I have found for some reason it's sometimes tricky to have strong opinions about how to build infrastructure because for whatever reason it pisses people off and you would think that it would be dull but you would think that something about the technical details would just you know not cause people to go out into all-out flame Wars that are like heavily religious but that's the way it is in crypto for whatever reason so instead of telling you prescriptively how I think that you should go and build infrastructure in the space I'm going to

give you three mental models for how to think about this stuff and then you can go and use them and draw your own conclusions from them and then go pick fights and Flame Wars and so on uh of your own of your own choosing so the first mental model is the is the model of the narrow waste of any protocol Tech stack so here's a motivating question for you how do you build a protocol that takes over the world can anyone hear name the most successful protocol in all of the history of computer science and I'll give you a hint it's a network of Networks IP is the internet there's no better

example of this mental model than the protocols that are at the heart of the internet and that is TCP which stands for transport control protocol and IP which stands for Internet Protocol and these protocols they go hand in hand but for our purposes I'm just going to focus on IP and I'm going to use the term IP interchangeably with TCP and I'll give you a little bit of the history I'll tell you how it is that they have been so damn successful and then I'll use IP as a driving case study for this mental model and then throughout the rest of the presentation if you think about it it is truly mind-blowing that the same protocol design of Ip took us from literally nothing to over 20 billion connected

devices it has survived throughout the course of 40 years throughout the period of fastest technological change in computer history and throughout all of that time the tech um for transporting bids from point A to point B has improved something like a million x and so it's truly mind-blowing that this same protocol design has survived throughout all of that change and has remained future proof and evolvable with Minor Details here and there like where there have been changes but by and large is the same protocol it was envisioned back in the 70s when it was initially designed and throughout all

this time the the ideas behind TCP and IP have have remained remained true so how and why has IP been so successful so there are many many reasons and there's actually a very good book called designing an Internet by David Clark that I would highly highly recommend but the most important reason is this concept of the narrow waste the first thing that IP did is that it create a you it created a unifying layer um for the entire fragmented world of computer networking a key goal for its design from the very start was was to

enable any networking technology no matter what it was based on to power the networking needs for any application on the other side so we look at this diagram there's TCP and IP in the middle but in the bottom there's an infinite diversity of different kinds of networking Technologies uh whether they be based on cable or fiber or radio or whatever or what have you and then on top we've got an infinite diversity of different kinds of applications that require different kinds of networking support every single packet that's sent through the internet flows through IP which is just this simple interface that creates a subtraction between the application

layer above and the hardware layer below so before IP existed each application had to deal with all of the messy details of the various networking Technologies um which essentially made for networks that were completely fragmented it was a world where each Network had to be tailor built for the needs of each application and it meant conversely that each application had to build its own network had to build its own networking protocol and had to provision its own providers of bandwidth and Hardware in order to actually be able to communicate it also meant that applications as a result were just fully unable to

interoperate with one another so the upshot of all of that was that networking really was hardly a thing it was just a play thing of researchers and academics and in contrast IP acts as a kind of aggregator it collapses the entire networking world into a single standard and by doing that it does three main things the first is as we've seen it allows for any request that comes from the demand side those are applications on top that need to move bits from point A to point B to be serviced by any provider regardless of the technology that they're using as long as they have the capability of moving bits from point A to point B the second thing it did is that conversely it allowed for all of the

hardware providers and and bandwidth providers on the supply side to span uh the totality of the of the addressable Market of applications that require connectivity and they can do that by supporting only a single protocol as opposed to having to work in a pairwise way with all of the various different applications that may need that may need networking and the third thing that it did which may be the most important is that it decoupled the application layer from the hardware layer such that both of those layers could now evolve independently of one another there's actually a really good quote from David Clark in that book that I mentioned which is that interfaces are

constraints that deconstrain and what he means by that is that an interface is a constraint because it means it now means that both sides spanning that interface have to support that interface and have to conform to it so it constrains in that sense but it deconstrains because now that they're doing that they can no longer they're able to evolve independently without having to care uh about what the other side across the interface is doing so that was the first order consequence of IP but the second order of consequences are economic in nature and I think they're actually even more interesting so by connecting the disparate worlds of

hardware and software uh on top the simple interface of Ip created a powerful flywheel of reinforcing feedback that is what ultimately led to IP taking over the world and the way that it worked is that as more bandwidth providers came into the network uh below that created more available more available bandwidth for uh application developers to build applications to use that bandwidth and as those developers built applications that were useful to users there was more demand for bandwidth which created an incentive for further for more more bandwidth providers to come into the network and so then we've gone full

circle and that that essentially is a runaway that's what basically created The Runaway chain reaction that just led to IP gaining the adoption that it did and eventually it eventually became the internet as we know today so next what is it that makes this narrow waste quote unquote narrow the most fundamental property of Ip which is the one that you would read about in any textbook about computer networking is that it was radically unopinionated unopinionated it's like the keyword and what does that mean it just means that the protocol simply does not care it is unopinated to how applications on top and how so Hardware providers and and

bandwidth providers below use the protocol it tries to maximize their freedom and that is the defining property of the internet and that is the reason uh that it has come to be known as the narrow waste of the Internet it's just a single protocol through which all traffic flows no matter what kind of traffic and no matter the technology um and that is ultimately the reason that IP has had the reach that it that it has it is the reason that it has been flexible and evolvable throughout 40 Years of technical of technological change um is the reason that it's Global and and sort of so totally uh kind of spanning as it as it has as it has been and by the way at the time at which

um IP was being developed there were a number of other competing standards like ATM and xns which were actually more complex and had more features but as a result they were less flexible and they they were less unopinionated and constrained the way in which users could use those protocols a little bit more and so in the end IP and its radical minimalism and its radical and opinionatedness was was what ended up winning so that that is the history what are the implications for new protocols today so let's bring this back to crypto to what all of you guys are doing or to what people in the space are doing so first I want us to step back and observe that what a narrow waste

does is that it allows a multi-sided market to emerge what is a multi-sided market my favorite definition is that a multi-sided market is a kind of common ground that creates value by enabling the direct economic interaction between multiple kinds of participants so that's a bit of a mouthful so it's a common ground it creates value by enabling the direct economic interaction between multiple kinds of participants and there's lots of examples so IP is an example because it's a common ground that creates value by enabling service providers and Hardware vendors to interact with application Developers

and there are other examples from maybe traditional Tech an operating system like Microsoft Windows or Mac OS 10 or iOS that's also a protocol that that allows for a multi-sided market to emerge with developers application developers on one side and then users on the other side something like uber like a ride sharing service and similar like there's writers on one side and there's drivers on the other and I would argue the multi-sided market is the core template of success for any protocol which maybe uh contrarian claim but that is what a protocol is born to do it's born to connect right it's born to bridge worlds between different kinds of

participants so let me give you a few examples from our own world namely from the crypto world so unit swap you all know uniswap is a decentralized Exchange again two sides there's liquidity providers on one side they're the market makers and then there are Traders Market takers on the other side compounds similarly there's lenders on one side borrowers on the other and those are both D5 protocols but here's a few other examples sound.xyz it's a decentralized music streaming platform there's artists on one side there are listeners fans on the other side forecaster it's a decentralized social network there are uh just content creators uh creative people on one side and there's users or to sort of people who just use the

platform on the other they all create a common ground for different kinds of participants to come together and economically interact with one another but the best example of them all is blockchains themselves like I contend that down the line we're going to be able to look back and there will be such a thing as the narrow waste of blockchain computing so by now you're all very familiar with the high level architecture of a blockchain and this is what it looks like in the most natural place for a narrow waste to emerge is the consensus mechanism at the very center the consensus mechanism creates a

multi-sided market that connects participants across all layers but most importantly it connects validators um below who provide computational resources and provide security with application developers above who build applications deploy smart contracts onto the blockchain and in the same way that IP created a unified address space for all of the participants in the on the internet a consensus mechanism creates a unifying a unified Computing substrate for for people building below and above and in the same way that IP allows you to not care about the messy details of the linked layer technology and you just have to sort of support the interface on

top a consensus mechanism allows you to not have to care about the specifics of validator Hardware when you are deploying a smart contract onto a blockchain and then finally as with IP consensus sets up a similar two-sided Network effect it's a similar flywheel the more validators come into the network they provide more security they provide better Computing functionality for application developers to come on top to build useful applications which create more demand and then drive value to the Token which creates a stronger incentive for more validators to provide more security and computational resources to the network and around the fly where we go right we have the same type of self reinforcing feedback loop so what can we learn from this

comparison between blockchains and could the consensus mechanism within a blockchain and TCP as we saw one of the most important factors that drove the success of Ip was that it was unopinionated and today there are countless of different blockchains out there that are running every kind of experiment imaginable and they span the entire Spectrum on one hand of the spectrum you have blockchains that are highly opinionated highly vertically integrated that insist in controlling everything from the peer-to-peer Network to the consensus mechanism to the compute layer above that and the kind of instruction sets that you're able to use and therefore the kinds of programming languages that are supported to the other side of the spectrum there are there are blockchains that are fully

minimal and and want to express no opinion about the kind of peer-to-peer Network layer that you might use or the kinds of programs that you may be able to to build on top the kinds of programming languages that you might be able to use they try to be radically unopinionated and there's everything in the middle as well and just to give you two polemical examples Bitcoin is a highly opinionated blockchain even though it's a minimal blockchain it's very opinionated because it prescriptively insists that you must build any programs that run on top of it in Bitcoin script which is a limited programming language that isn't even turn and complete on the other hand ethereum which is actually a more complex blockchain is more unopinated because it gives you a more expressive programming language with

which you can build smart contracts you can build more things you have more freedom as a user than you would be if you're a user of Bitcoin so the jury is still out but it'll be interesting to see if a similar story plays out uh where the unopinionated protocols lead to a kind of narrow waste and decouple things effectively such that you have the flywheel uh effect kick in and maybe those maybe those are the blockchains that emerge as dominant as it did for the internet and by the way ethereum is somewhere in the middle I think there are there are radically more unapinated blockchains on the other side and it's unclear as to how far you should actually go because there are

trade-offs involved and making something more unapinated does sacrifice some degree of control and so you lose something as a result so this is why this is a mental model as opposed to a prescriptive like solution this is the way that you that you should do it so takeaways if you're building a protocol I would encourage you to think of the ways in which your protocol allows for a multi-sided market to emerge who are the participants how much Freedom do those participants have how opinionated is your protocol does your protocol have the elements that might lead it to becoming a narrow waste I think those are the questions that you should be asking and that might inform your design in the direction that you take

all right so second mental model there is a closely related concept to unopinionatedness and that concept is modularity so these two are are often related and they often coincide which which means that they're often conflated it's easy to think that they're actually the same thing but they're not they're different things and then and they're actually I'll show that it's actually those these two things are actually independent from one another so the driving question for this mental model is how modular should my protocol be and we'll we'll see what all of this means a little bit to get started let's go through just simple definitions what does it mean for a protocol to be on opinionated it means that it

maximizes a user's freedom in the way that they can use the protocol or build on top of the protocol that's what it means for it to be unapinated what does it mean for a protocol to be modular on the other hand it means that the protocol or the architecture that the protocol leads to can be broken down into into Atomic sub-components that are self-sufficient sometimes a protocol being modular leads to it being unopinated because the modularity sometimes leads to Greater flexibility but that's not always the case and so I'll give examples where actually in order to become more unopinionated you actually have to

become less modular so you see how these two things are different I should also say just for clarification's sake that the internet is an example of a protocol and stack that's both it is a both an unopinated protocol design but it's also modular because it was designed to be built in layers where each of the layers is independent from all of the other layers and there's clean dividing abstractions and interfaces in between each of the layers so here's a diagram that helps illustrate how these two things are separate there's two axes on the y-axis we've got unopinionatedness which I've just abbreviated as user freedom and then on the x-axis we've got modularity

and so we talked about two examples before Bitcoin and ethereum so you see how Bitcoin is not very modular because it's fairly vertically integrated it bundles consensus within a specific Computing environment that prescribes a specific programming language so it's vertically integrated in that sense uh and it also doesn't really give you that much user Freedom it's not that unopinated because again it prescribes a specific programming language it doesn't give you the freedom to maybe build something in a program in a turing complete programming language the way that other blockchains do so ethereum is further off to the right um more modular I think it like for example the the roll-up Centric vision of ethereum encourages uh the building of a blockchain stack in layers where you

have ethereum layer one and then there's l2s that maybe provide greater performance or provide additional feature features so it's a more modular vision for how a blockchain should work and it's also more unopinionated even though it's more complex because it has a more expressive programming language so it doesn't have an opinion as to what kinds of things you should be able to build on top of ethereum uh some of the other examples here are Aptos and sui which are highly vertically integrated because the team that are the teams that are building those those blockchains are building everything end to end and they've they've built every layer throughout the stack but they they also are fairly

unopinionated in that they have like highly expressive programming languages um and then off to the right there's Celestia which is a vision for a hyper modular uh very minimalist type of architecture where Celestia is just a data availability layer and then all of the other things get built as modules around it and so here you can actually see how how Celestia is more modular uh than ethereum because it's just the availability and compute should be provided by a different module um but it's less unopinated and it's it's less unapinated it's more opinionated because it prescribes that no smart contracts should ever run on top of it right that it should just be

used for data availability there's no programming language it makes a strong opinion about what you can and cannot do on top of Celestia so it's more modular less unapinated this is this is again the way in which these two things are different TCP IP all the way up to the top right it's just both it's a very modular architecture it's also very unapinated so this hopefully illustrates maybe the distinction between these two concepts there's a very good book it's called the innovator solution it was written by Clay Christensen in the early 2000s and there are two chapters in that book that specifically address this whole topic it was geared primarily for traditional businesses and whether the business should be modular or or

integrated but I contend that a lot of the insights translate over to our world and to crypto into protocols and blockchains so his argument is that it makes sense to be monolithic that is to be integrated which which is to say not modular whenever the technology is not sufficient to meet the user's needs for whatever reason maybe the technology is too nascent maybe it hasn't gotten to the point at which it fully satisfies everyone's desires and the users is therefore always wanting more is that they're left always wanting more I think and I think that that's kind of where blockchains are today but like no one would say like

blockchains today like fully satisfy all of the desires all of the all of the dream needs that we that we all have for what a blockchain should do um and whenever that's the case clay Christensen argues it makes sense to be monolithic but at the point at which the technology does begin to meet the user's needs and maybe even begins to surpass those needs then it makes more sense to begin to modularize and the reason for that is that is that once you modularize you're able to reduce cost because now for any one particular module you can Outsource the work for it to a whole ecosystem of participants that can build the

functionality for that module which creates a kind of commodity commoditization for that module and reduces the overall cost offering of the entire thing it also by the way gives your users and the developers that are building on top of you more flexibility because now you know they can choose to swap swap out one module for a different module so those are some of the advantages of modularity but of course there's cost as well namely that in order to build something in a modular way you you often have to introduce more complexity because you now have to think about all of the edge cases and you have to commit to specific interfaces um perhaps prematurely specific abstractions that divide the different modules that you now no longer

that you now have to respect and you can no longer kind of sort of play with so those are some of the cost of modularizing but there's clear advantages given that it can reduce cost and it can give your users more flexibility so he goes as far as to argue that as soon as the capabilities of the technology begin to surpass your customers needs then at that point if you don't modularize some someone else will someone will build a modular architecture and then they will they will essentially eat your lunch but before you reach that point the advantages on the in favor of of just staying monolithic having control not modularizing and controlling uh the entire architecture yourself

uh so as to be able to optimize across the boundaries of would-be modules and as a result get as close as possible to meeting the user's needs so this here is a diagram from the book from the innovator solution and it may be small and hard to read but but you can see the dotted line is what users expect what users would would want from the technology and then there's two lines that are at a steeper slope the top line is the capabilities that you can get if you're integrated meaning if you're monolithic the line below that those that represents the capabilities that you're able to achieve with a more modular architecture and the fact that it's below to suggest that if you opt

for a modular architecture you have less control so therefore you you can't optimize this aggressively so you're under the line that is vertically integrated which has better capabilities and so it kind of shows that as time goes on the the integrated architecture begins to overshoot it is providing more than what the customer or what the user actually needs or wants and so at that point you're no longer getting anything you're no longer gaining by being integrated and it makes sense then to switch to modular to gain the advantages of modularity while still being above the dotted line which is which is the kind of the the Baseline uh demand from users the needs of users so let's go through an example just to

make this really clear recall the history of mobile so before the iPhone there was this thing called the Blackberry I don't know how many of you remember the Blackberry um but the Blackberry was a very monolithic and vertically integrated device it had a single app really like single important app which was email it was essentially an application specific computer that was tightly controlled by RIM which is a company that made BlackBerry um in order to make sure that it was good enough at that one thing it was good enough at doing email and as a result it was a good product for business customers who just wanted to to use their phone to do email

and at the time doing something more that that was doing something that was more modular and more General and more unopinionated which in this case coincides with being modular would have been too difficult the technology really wasn't there yet and it was only as the technology became good enough that it became possible for Apple to become more modular and in the form of the of the App Store the app store gives gives the the iPhone a little bit more modularity because now you know a bunch of developers can come and write write applications for the iPhone and then that ultimately ended up being the model that that won but it would have been hard to do that at the time and you really had to have the tech Evolved first in order to get that to work

so now all of that is fine and good but there is there is a problem which is that the internet seems to be a glaring counter example of this whole Theory because back when the internet first emerged it was not only not good enough to meet users needs it hardly even worked it was certainly not sufficient and at the time there was also this competing approach known as the information superhighway which was very vertically integrated and it was actually basically this Consortium of all of the big name companies coming together and trying to build this end-to-end experience that worked right that was good and that was their approach

but interestingly IP and the internet ended up winning with their hyper modular approach even though they had started from this place of not being good enough that somehow took over and ended up winning and so why is that in my sense is that when Christensen wrote the book he was not fully aware of just how powerful Network effects can be and in the case of Ip the power of this unopinionated narrow waste was just so great that it just turns these it just turned this whole theory on its head so where does that where does this all leave us like does does this mean that modularity is good or is it bad

and the answer is that it depends there's advantages to modularity and there's disadvantages there's costs to modularizing namely increased complexity and and um having to decide on interfaces and abstractions maybe prematurely um but there's also advantages the fact that you can reduce cost and give your users more flexibility you can create bigger ecosystems because now there's more people who are participating um there are also cases by the way where um modularity confers the benefits of unopinatedness but there's also cases in which that's the opposite where modularity reduces an opinionatedness and so you also have to be mindful this is how this this whole section

interfaces with the first section about the about the narrow waste and unopinatedness so basically the takeaway this whole section is a mental model that can help you think through the complexities of that choice and it'll be Case by case and it'll depend Okay so the third and final mental model the network flywheel this is the final stretch so here's a driving question what role should a token play in my protocol in this section of the narrow waste before we talked about how the best way to think of a protocol is as a as the driver of a multi-sided market

that is what protocols are born to do that is what they do when they're successful but the beautiful runaway Chain Reaction the feedback loop that made a protocol like IP ultimately take over the world has to be bootstrapped somehow and historically all multi-sided markets have suffered from what's known as the cold start problem it's hard to get them started because a ready and willing supply side is unlikely to just spontaneously materialize if there isn't already a demand side on the other side and by the way vice versa like that the reversals as well the man's side is unlikely to emerge if there isn't

already a supply side on the other side in the past this problem has been solved with enormous influxes of capital usually from VCS in the case of something like uber for example where you subsidize one or both sides of the market or from entities like the government for things like IP where initially you know you have to have a lot of external Capital provided largely by the government in order to get the internet to really to really take off so they effectively subsidize one or both sides to really get the flywheel going this is where I believe that this I think actually it's already been shown uh a token can be very interesting

the cold start problem is one of the many problems that a token can help solve so here's a thought experiment for you imagine that Uber had raised less money than it did and so instead of subsidizing both sides of this multi-sided Market with cash the way that they did If instead of doing that they had opted to give drivers on the platform a small amount of equity in Uber the company for every drive that they did it's basically giving them a small amount of ownership an ownership stake in the company could we imagine that having done that

might make some of them realize if this is the early days of uber right it might make some of them realize that that ownership stake might at some day might someday become very valuable right it might be worth a lot of money would that make them more likely to drive for Uber would it make them more loyal towards Uber as opposed to some of its competitors would it have made them more likely to evangelize Uber to other drivers other prospective drivers I would argue that yes this is this is actually a fundamentally more efficient capital structure for multi-sided markets the capital that is needed to get them started doesn't have to come

from external sources like Venture money or the government it can come instead from internal sources namely the participants themselves who can contribute to the network with with labor and you get that by making them owners but making them fractional owners of the platform and you do that in exchange for their contribution so let's bring this back to the internet and to crypto imagine if the internet imagine if tcpip had had a token and if that token had been had been granted to people who helped bootstrap the internet as a network and if that token granted its holders a fractional ownership of the network itself that

maybe gave them rights to some some share of whatever Revenue flowed through the internet would that have made the internet easier to bootstrap without as much government government funding maybe would the network effects of the internet be even stronger than they are today given that you now have this additional Vector for Capital both Financial Capital but then also human capital to enter the network again I would argue very much yes so finally this brings us to the network flywheel it's a mental model that applies especially well to layer 1 layer one blockchains but it can be modified slightly to apply to almost any protocol

you can see you'll be able to see how that's true so the flywheel starts with the founding team and a group of core Developers who come up with a vision for what the protocol should be and with the help of the core development team and maybe external funding from investors that comes in in the form of financial Capital they build the protocol and establish an initial token value the token is valued at something there's some value to the token and now the token exists and it has value it creates an incentive for a

different kind of participant in the case of a layer one blockchain that would be validators to come into the network and provide production capital in the form of computational resources that provide security and functionality to the network such that you now have a base level of platform functionality which encourages third-party developers to come into the network and provide their human capital by building useful applications but then provide utility to end users and those end users eventually begin to form a kind of community that reinforces the original vision

at the heart of the protocol and so we've gone full circle this is how the flywheel spins as the vision for the protocol becomes stronger the token becomes more valuable maybe with the help of additional Financial capital from investors which creates a stronger incentive for validators or other participants to contribute production Capital to the network which improves the functionality of the network encouraging more third-party developers to build applications which provide more utility to end users and again that reinforces the original Vision at the heart of the network so that's the network logo the token in this flywheel plays a critically important role and that is that it's a vehicle of ownership in the protocol itself

and as a result it kind of allows the network to borrow value from the Network's future and if participants in the network believe that the network will be valuable in the future they may be more willing to participate and contribute today if they were able to earn some fractional ownership of that future value so here's the takeaway if a network is designed correctly the various participants at every stage of this flywheel may be able to earn the token in exchange for helping to bootstrap the network and that is essentially what helps what helps the network break through the cold

start problem it's all mediated by the token there are many roles that a token can play but helping bootstrap a protocol's multi-sided market I would argue is one of the one of the most powerful ones and that's that's basically the network flywheel so there you have it three mental models went through the narrow waste the modular advantage or disadvantage and the network flywheel any questions hi Alice thanks for the talk um I have a question about so like based the Regulatory Compliance um changes in the recent months right so especially the example you mentioned about Uber like if they raise from their um like users so how would that would

that make token more of a security and also um I can see like a barrier of you know how do I convince drivers to invest in this token there because like Venture investors they're like professional they look at softwares all the time so how do how do they what do you see barriers here so two questions the first the first question is about regulatory that's a very nuanced conversation and I'm not a lawyer uh but the the overarching argument would be that a network that is fully decentralized where there is there are no asymmetries of information you essentially just have everything out in the open participants can can interact with other participants

as as equals where there is no Insider that has some asymmetric information about the future price of the token if you're able to eliminate all of that and any any asymmetries are gone and you get you get to that through decentralization then you would fail the fourth prong of the Howie test which is that there are material and managerial efforts from others because there are no managerial efforts from others in a world where the network is fully decentralized and so then you can argue that it's not that it's not actually a security but there's a lot more there and I think we're happy to chat online with about about some of the Nuance as for the second questions how do you encourage how do you encourage drivers to quote unquote invest in the network well I was arguing that actually they wouldn't invest they

would just earn the token through work right they would drive for the platform and then as a result maybe they get some cash but then maybe they also gain they also they're also able to earn a fractional ownership uh share in the future of the company Uber and that combination might then shift the way that they behave towards the platform makes them more loyal towards the platform more willing to evangelize it to others um more willing to drive for longer so that's that's essentially the argument and it could be maybe a like I think a monetary investment wouldn't be necessary be more contribute human capital in return for okay for a token basically yep yep okay that's very clear thanks sure

I was pondering two things kind of the last points and thinking about how like drivers like we're just in Ecuador and we're taking a lot of Ubers and like some of the Ubers didn't even want to take credit card because they wanted the liquidity like there and then yeah um something in just like a perspective of someone who's like meteor and you know can't eat in 10 days when the token goes to the moon they need to eat now just kind of like thinking how you know you know in an Ideal World those situations the liquidity would be there um that was one thing but the other thing I was also interested is just kind of get your vision on you know what the IP of blockchains is it is it going to be like Empower us to like code a contract in one language and put it on

one blockchain and you know have uh some ZK or some layer to do something else and we don't think about that at all is it a programming language that needs to like come in and like abstract everything away or kind of like how do you see that that IP like actually come into fruition or like do you see any projects that are working on that like cool uh approaches that are taking to like become that um you know that little middlemen yeah well so let's do the first point certainly I think Uber is an example and I think all the parameters would have to be figured out and that you you do need you need to provide liquidity to participants in the short term so again maybe you do that through some

combination of cash that is liquidity so they can cover their bills but then there's also this component where they they have some ownership in the network so that influences their behavior moving forward it makes it so that it's a longer term relationship as opposed to the very transactional like you drive you drive someone today and you get money today for that drive and that's it and that's the full extent of your relationship with the company as for your second question so again like it's it's hard to be prescriptive and hard to like say like this is the way to do it this is the way that it all work but I I do think ethereum is an interesting example I think it it lands at an interesting region of this of the trade-off space and it's somewhere in the middle in this spectrum from like

vertically integrated and opinionated to um unapinated and completely modular um but I think that that it has made has maybe found a kind of sweet spot because it it does provide you with the flexibility to build all sorts of things with a turing complete programming language and solidity which by the way creates the ability for people to then build layer twos right layer twos are possible on ethereum in a way that really are not possible on a blockchain like Bitcoin and so then that provides a path for ethereum to extend itself and for it to become more powerful uh through different kinds of modules that enhance what the core layer one can do and that could be somewhat analogous to the way that tcpip has evolved where

where you have IP at the center but then TCP is actually a separate module on top of Ip which can be swapped out for UDP as different a different transport level protocol and then on top of that people have built all sorts of different layers and modules and there's like seven layers and you can actually sort of dive deep into into how that whole stack has evolved and it's it's sort of Highly modular and and I think it's driven large part because of the fact that IP itself office so flexible and so maybe the core flexibility of ethereum and the fact that it has initial Network effects in a very large community so far might might lead to it for it to to kind of go down a similar path and for it to be a similar story as IP but we'll see thank you my question is

how do you think about these two token models given the kind of the network flywheel um you know model that you presented particularly in the for example rocket pool where a lot of the value accrues to our eats because it's based on the ethereum staking value but then there's a indirect value accrual to their governance token of you know that token as well so how do you map that to the network flywheel model that you presented well I think that the core like high level view is that the token should in some way encourage a kind of participation that's helpful for the network right and so then you can map that into whatever whatever networks in the case of something like a governance token but the the incentive in that case is to be a

responsible Steward of the network right and that is a form of contribution it's a kind of human capital it's a kind of Labor that a protocol that needs to evolve needs to have and a token that is a governance token could play that role right or it could be that the token is incentivizing a different kind of behavior maybe it's maybe it's providing security and that happens through a behavior like stake where you take the token you lock it up as a kind of skin in the game and you're expected to provide some kind of service to the network and if you don't provide it correctly then you just take a slashed right that's another another kind of incentive structure and there could be many different kinds of incentive structures but the but the overarching goal is that you're incentivizing participants uh along every stage of

that flywheel to contribute to the network and to help bootstrap it and to make sure that it succeeds in the longer term does that mean it's kind of imperative to have both out of the gate so that participants have an understanding of each side of the the value accrual governance and the you know the the more kind of Direct Value accrual one for example I think it probably depends like I don't think it necessarily means that you have to have everything in place from day one it will probably also interplay with the regulatory component where you cannot have a fully tradable token until the Network's fully decentralized and there's a there's a tricky question as to how you get there and so I think it's Case by case in some cases you would want to have all of the incentives uh fully fully available and on chain up front in

other cases maybe you would roll it out over the course of a longer period of time all right I appreciate it thank you so uh you brought up the example on your presentation where you had modularity on the x-axis user freedom on the y-axis and yet all these different networks do you have any sort of framework to determining what applications are better built on more modular platforms or more user Freedom based platforms or protocols I mean anything come to mind on that front for applications or for like for example consumer applications for example like I'm thinking because you brought up the sound.xyz example I collect a bunch of music on sound I'm trying to think like how would sound look like on Aptos for example like and how would you actually approach that mental model determining is sound better

fit for something like ethereum or something like Solana or something like Aptos as an example yeah well I think from the perspective of an application developer what you care is for the infrastructure to meet your needs right to essentially work and so it could be the case that a vertically integrated architecture is more capable of doing that but it could be that because the blockchain that you're building on top of is fully vertically integrated it is more performant or it has like better usability and better is better documented or has like better Primitives that you can build on top of um and if that's the case then maybe you should go for the vertically integrated uh sort of Stack to build on top of but it may be the case that again maybe we're in the realm in which in which the

infrastructure and technology has started to overshoot your needs and maybe it's more performant than it needs to be you don't need the extra performance and I don't think we're in that world at the moment but maybe five years from now we will be and then at that point what you care about more is not more performance but you care about flexibility and then the modular approach gives you that flexibility it gives you the ability to swap out some module for a different one it doesn't lock you into a specific way of doing things and it gives you more freedom it like coincides in this case with it being also more unapinated but then you're assuming that you're in a world where where the the architecture as a whole meets your needs and that's ultimately I think again dependent on what the application requires like something like sound you know might be

different than something like a game where the performance requirements are much higher but that's I think that is the mental model for how to decide those things and it's hard again to be prescriptive and to say this one way and then my next question is are there any examples where tokens have made protocols more opinionated opinionated anything come to mind on that because you brought up the example of like the TCP IP and it didn't have a token I ended up doing really really well but the argument was it could have improved bootstrapping the internet if it did have a token incentivizing the right people to act accordingly that's right um but it was still a success without a token right are there any examples where tokens create protocols more opinionated and actually doesn't have the outcome in

which it was intended to have yeah certainly so in the case of the internet you write it was still successful but we should remember there was enormous amounts of right of influxes of capital primarily from the government to really get it to to come to life are there examples of tokens that have made a protocol more opinionated well there are certainly cases where a token makes to the protocol work less well cases in which the token is not well designed and so then maybe it leads to the overall financialization of the ecosystem so like adding a token maybe to a to a creative ecosystem like maybe a social network or like there are examples like steam way back in the day where introducing a token and introducing Financial incentives

completely warped the behavior and made it so that now is no longer a fun place to be as it is no longer a social network that you want to spend time on and the content become becomes content that's entirely spammy and and like clickbaity and the whole intent behind the content is to try to capture more rewards through more tokens so then that's like an example of the token just steering like basically the incentives are not working the way that they were designed and they're encouraging the participants to not help bootstrap the flywheel and help the network and rather it's incentivizing them to do some other thing which actually is destructive so there are lots of cases like that and it's a very tricky tricky question how do you design the token maybe maybe one potential answer is that the token should be it should be measured like the reward

that you're giving participants should be calibrated so that it doesn't completely warp all Behavior maybe you do that by say locking up the token you you award a participant in the token but it's locked up right they can't immediately cash out for example like that's one one degree of Freedom that you have in a design space and I think we're still exploring all these questions like exactly how should a token be designed for different kinds of applications in order to lead to the desired outcomes cool thank you um thanks so much um wanted to ask a almost related question around uh how you talk really liked how you talked about unopinionated protocols um and then the introduction of tokens as a way to create a business model around this um I was going to ask how tokens play into this notion of unopinionedness

where there are inherently mechanism decisions as well as there is an element of of price and uh like a kind of team behind that or like a community behind that at the at the very least and so how do those two things kind of do you envision them fitting together in the future yeah it's a great question I mean I think that the way to think about this is I think that IP is an example showcases how a protocol that is unapinated has some advantages because it can lead to this kind of narrow waste effect it can lead to a broader sort of addressable Market both above and below the protocol uh and that's a powerful thing but of course there are trade-offs so if you try to maximize unapinatedness to the maximum extent there will be other things that you're giving up because

namely now you can't have an opinion about well the way that people use your protocol and the way that people use the thing that you're building and it might also mean that you can no longer be as deliberate with the architecture of what you're building it may mean that your that you that now you can no longer decide like exactly how your token should work or like how the programming language that you're building that people are supposed to be able to write applications in uh should work and so there is such a thing as taking it too far and so again it's this case by case thing where where it's a trade-off and you have to figure out what is the right level of opinionatedness that your protocol should have knowing that more on opinionatedness is good but you have to then contend with some of the costs

that it comes with so ethereum is an interesting example because it could be ethereum could be more on opinionated but the fact that it isn't maybe allows it to to sort of work well in the ways that it does like it bundles for example it bundles solidity and the evm as part of the core layer one and there are some some advantages to that namely that that it can optimize specifically for the evm uh it can like build in for example pre-compiles for certain things that are very common that people need on top of ethereum and that's it being opinionated because all applications on top of ethereum have to be evm based um so those are the advantages but there's disadvantages what if someone wants to write an application in JavaScript or like rust or something

else well those applications can't readily work on top of ethereum and so that's like an example where like it comes at a cost but that's just that's just the the position on the spectrum that ethereum picked and is it one experiment of many and I think we'll see how it all plays out but it's ultimately adjustment call it's hard it's again hard to be prescriptive awesome thanks so much um how much do you attribute the dominance of Ip to the narrow waste design versus decisions made at the operating system and Hardware layer because I would imagine once routers started to implement IP the defensibility and switching costs were so high that even if it you know more advanced standard had come along it never would be feasible I I think the fundamental driver were the network effects that began to kick in because then I think uh like as soon as you had

some amount of traction and people were using uh computers they were using IP for interoperability and then again as you guys very much as you're saying like as like sort of Hardware vendors and service providers began supporting IP it just becomes this inescapable gravity well where if you want to connect to other people in the world using computer networking technology TCP is the is the network that everybody's using and if you try to use something else well then there will be no one over there there'll be no one to communicate with and so then there's like very little choice that you have you just have to kind of go with it with the thing that's already taking off and I think that is very much this flywheel effect that began to to sort of it just began to Cascade and this runaway chain

reaction that ultimately led to its success and I think it's what triggered actually what you're saying like the hardware developers integrating it and making it making it fully functional yeah it makes perfect sense like the narrow waist design came first and that's what actually made it possible for the hardware events to adopt it I guess so my last question would be um how do you distinguish between the cold start problem versus lack of product Market fit because I would argue that in history there have been examples of products that have achieved um that kind of network effect independent of a token or subsidy by VC capital for example Napster or Facebook before it became a corporate Network yeah I mean it can be I think it can be tricky to to tell us to apart uh a priori but I think there's good signs of

that so for example in the world of crypto if you do build a network where there is a token built into the network to subsidize say the supply side so for example say it's like a storage Network where you're incentivizing storage providers to provide storage to the network or maybe it's uh it's a sort of connectivity Network where you're incentivizing the supply side to provide bandwidth to a group of participants or maybe it's like a normal blockchain where you're incentivizing validators to provide Computing capabilities for application developers on the other side if you do that and the supply side shows up because they're they're going to show up because there's money on the table and no demand side shows up then that's obviously a very clear sign that you

don't have product Market fit at least you don't have broad Market fit on one side of the market like the supply side has showed up and they're willing to do whatever they need to do in order to in order to earn the rewards but uh but the but the demands that hasn't shown up yet so I think that the at the moment where you start to see both sides reinforce one another then I think that's that's when you can say that you actually have power Market fit thank you so much makes a lot of sense appreciate it thank you awesome thanks so much Ellie [Applause] [Music]