In this article, I’ll focus on how highly scalable Blockchains such as the Internet Computer Protocol (ICP or IC) will allow us to use global credible promises to build vastly larger platforms cheaper and faster.
Table of content
- Introduction
- Platforms & Network Effects
- Fractalized Organisations
- Liquidity Mining
- Madness
- Conclusion
Introduction
When I got interested in the ICP, I saw a lot of people describing it as a decentralized cloud. Personally, a new type of cloud network seems to be infinitely less exciting than an insanely scalable Blockchain.
Interestingly the IC currently is something between a cloud and a highly scalable Blockchain. Technically, node shuffling is not implemented yet which means the IC’s security does not scale across sub-nets. Personally, I think we’ll get there. Chain key technology already allows us to add subnets to the system indefinitely. Once they share security I think the IC will become an insane piece of technology.
In my previous article, I explain how Blockchains fundamentally solve problems around collaboration by allowing us to form credible global agreements. What happens if we scale that ability massively? In the following article, I try to find some answers to that question.
Platforms & Network Effects
Platforms generally can be modeled as coordination games as explained here. The default equilibrium for a platform is that nobody is using it. In coordination games, individual actions depend on other people’s actions in a self-reinforcing way. If everyone else adopts a platform you will do so as well and if nobody adopts it you won’t do so either.
Chicken-and-Egg Problem
In between the two equilibria, there is a tipping point where each additional user adopting the platform makes it rational for all others to do so as well. Before we have any users, none of them has incentives to join the platform in the first place. That problem is what people refer to as the chicken-and-egg problem of platforms. Once the tipping point is reached network effects typically cause the platform to become very large and essentially a monopoly.
Penguin Problem
When we try to reach the tipping point we face another problem. I already feel a bit ridiculous obsessing over these topics to such an extent, so you can imagine how I felt when I found out that management researchers call it the “penguin problem”.
We can think of it in the following way; even if we can incentivize people to adopt the platform early on, we need to incentivize a sufficient number of early users simultaneously, or else we face large additional costs. For example, if we roll out an incentivization program that pays Uber drivers to offer rides then we need to do so globally in order to kickstart consumer adoption. If our fleet is not big enough to trigger consumer adoption we will have to pay for the entire fleet ourselves. This sounds unrealistic for Uber as people would already use it in some cities, right? That is absolutely true and I’ll come to this, but if there are none of these early “local” use cases, then we’d face the described scenario. Each additional delay in bringing all required drivers to the platform can cost us a fortune. That is the penguin problem.
Sides
Researchers typically differentiate platforms by their number of “sides”. Uber for example has two sides, riders, and drivers. We commonly relate the number of sides to how many different user interfaces a platform has. When there are network effects across these sides we call that cross-side network effects and same-side network effects when they occur within one side of the platform. Common assumptions around these effects are that each additional user adds value to the whole network by offering an additional “potential” to connect.
Local Dense Networks
The confusing thing is that this is not the end of the story nor is it entirely true. In reality, we know that some potential “connections” on these platforms are highly unlikely to ever matter to us. It is rather that specific users create value for specific other users. That means if we look at it from the perspective of adoption incentives, we need to understand, which exact user is creating value for which exact other user. We could refer to each user as a “node” and could refer to valuable connections between them as “edges”.
If we knew which users create value for each other we could try to engineer a chain reaction where we could just focus on bringing the exact right users to our platforms sequentially.
As a platform orchestrator individual relationships might stay hidden from us but when we “zoom-out” typically we can observe larger patterns. Uber for example is certainly aware that drivers and riders of the same city mostly create value for each other.
When a specific local area adopts the platform, let’s refer to that as an “infection”. Infected areas can cause “spillover” effects and thus infect further areas. If enough people of an infected city regularly commute to another one, say for work, then they might start to infect that city by using Uber there. The more exposure that nearby structure has the more likely it is to be infected.
Global Network Effects
While dense local networks can facilitate the adoption of a platform, global network effects allow it to stay dominant. Uber for example does not have strong global network effects while Facebook does.
Fractalized Organisations
The major web2 platforms are a direct result of an increase in computational resources and connectivity. This allowed us to 1) build a larger variety of platforms and 2) have richer interactions on them. Now that the computational power of Blockchains can scale infinitely on the IC we can expect the same to occur on platforms running fully on-chain. As a result, I expect the amount of these dense local clusters of relationships for a given platform to become vastly higher. I think we can make use of these structures to massively improve how we bootstrap platforms.
Let’s dig into that by briefly summarising the findings of my previous article; Humans have an inherent ability to innovate, yet fail to collaborate. Collaboration is achieved through over-collateralized, and thus credible, agreements. Network effects and the presence of a community can make an entrepreneur’s promises credible. Blockchains allow us for the first time to have credible global and inexpensive agreements.
Based on these thoughts I developed a hypothesis on what that might allow us to build. I reached a major point of confusion when I realized that what I had come up with was essentially one thing: franchising.
Franchise the damn thing! Franchise, Franchise, Franchise!
Bare with me.
Fractalized Franchising
I believe current management literature has only partially comprehended how credible agreements affect business models. I think digital platforms were up until now very limited in their ability to make credible promises. That has caused them to look vastly different compared to traditional businesses and has thus distorted our perspective on them. I believe this will change.
Franchising is simply one of many ways how firms leverage people’s individual or say “local” knowledge/assets to build something of greater value. We can now directly apply the same ideas to digital platforms. A restaurant concept is simply something of value that others can reutilize. The Uber application can be used across different cities and is thus equivalent. Similar to platforms, restaurant franchises typically contain scalable parts. For example, the number of standardized deep fryers, various other input goods, or the brand are parts of Mcdonald's that are subject to economies of scale.
What are the motivations behind franchising?
Instead of running a franchise operation an entrepreneur could simply raise money from a VC and begin to open up fast-food restaurants around the world. In exchange, of course, the VC wants to get a stake in the business. The VC is mostly interested in the scalable part of it as owning a bunch of restaurants in highly competitive markets is not extremely attractive. This implies that the entrepreneur has to make that part work by opening up a large number of restaurants. This is hard to pull off, therefore risky, and will result in a bad deal for the entrepreneur.
A further problem is that the entrepreneur does not know where it makes sense to open up restaurants. He grew up in a particular city and has limited knowledge about other cities, countries, or cultures. Will the restaurant concept work well in South Korea? No clue.
Franchising solves 2 main problems. 1) It allows leveraging local knowledge of a diverse set of people. 2) It allows raising resources by selling the unscalable parts of a business in order to fund the parts that scale well. All without selling any equity.
Local Knowledge
Due to local knowledge, it can be fully rational for franchisees to open up a restaurant locally even if they don’t expect Mcdonald’s to be a global success. Maybe they know people in a particular neighborhood would love these burgers, or maybe they know they could get an underpriced location.
This is equivalent to us searching for early platform use-cases. Franchising drastically improves this by allowing people to self-select to market and monetize the concept locally. If enough early use cases are found that way, that alone can be enough to bootstrap the entire business.
Things get insane when you consider that the scalable part of the business plays into the local business’s profitability. If Mcdonald's becomes a large success the local businesses typically profit as well. That is very related to the penguin problem. If the local entrepreneur’s expectation is that Mcdonald’s will become enormous very quickly, then they are more incentivized to open up a restaurant and thus contribute towards making that a reality.
Selling Non-Scalable Assets to Fund Scalable Assets
Typical franchising agreements require the franchisee to pay an initial fee and to purchase inputs goods from the franchisor over time. In return, the franchisor allows the franchisee to pocket large parts of the revenue from each restaurant. That allows the entrepreneur to grow the scalable part of the business by selling the non-scalable parts without losing any equity.
In the end, there are clear similarities between franchising and bootstrapping platforms. In both cases, we want to bring our entire organization to a point where we benefit exponentially from its adoption. However, typical platforms still raise money from VCs to solve the chicken-and-egg problem. Why is that and what is the missing piece?
Inexpensive, global agreements.
Enter Global Agreements
Agreements in the legal system are neither cheap nor global. They have to be tweaked for each country to work and they are expensive to set up. That means we can only gradually grow the franchise empire. The profits franchisees expect to make are therefore lower which further slows the growth of the organization. This is in essence the penguin problem.
Global franchising agreements on the other side could cause an explosive stimulation that makes it more and more rational for each franchisee to open up a restaurant at the same time, which in return fulfills everyone’s expectations.
At its core, a franchising agreement simply grants someone the right to use and monetize a concept locally. What I basically propose is that we bring this idea to the digital world. What if we credibly allow entrepreneurs to profit from bootstrapping local markets? We can structure such a promise so that we can still get large parts of the upside.
We could simply build an application on the IC, let’s say an Uber clone, and then allow people that bootstrap local markets to profit from doing so. In the simplest version of this, the main smart contract would just not ever be changed and would therefore establish trust in the agreement. People around the world could then self-select to bring IC Uber to their city and profit massively from doing so. The platform owners would profit from the spillover effects. Of course, we want to change the contract over time, I’ll come to this later.
2 things make this insanely useful;
- Local does not mean physically local. We can think of a subreddit as a local structure. That means the diversity of these structures is only limited by people’s imagination.
- This happens all at once globally.
If there are enough local use cases rational to pursue without the platform having to be a large success, then that itself increases the likelihood of global adoption due to spillover effects. The higher the expectations are that the platform will be a global success the more incentives entrepreneurs have to contribute. The ability to do this globally can exponentially benefit our ability to bootstrap platforms. While there might not be enough early use cases within a given jurisdiction it is far more likely that there are globally. I’m saying if that is the case, then a global agreement creates a huge amount of value as the penguin problem gets solved.
Credible, global promises change everything.
Competition
In franchising restaurants typically get a local monopoly over specific areas. If that was not the case, marketing in a specific area could be free-rided by the competition which would eat up incentives to open the restaurant in the first place. Promising a temporary local monopoly could therefore further stimulate the growth of the platform.
Incentives
All user interactions can now be on-chain. The local entrepreneurs, therefore, know with absolute certainty that they cannot be cheated. That is very similar to how the franchisees are in control of selling burgers. The franchisor cannot pretend that they did not sell any burgers. This is not the case for web2 platforms. Uber would have full control over all transactions that occur on the platform. They could simply lie or change the way they count “revenue” to cheat local entrepreneurs out of their rewards.
Things get more complicated if we actually want to dynamically change the core application and how it gets monetized. This is obviously essential for any startup. The credibility of the promise to reward early contributors will be affected by that. During the growth phase of the platform, I expect the mother organization to stick to an initial agreement even if they had all means to modify it. Changing the agreement would lead to an erosion of trust and would therefore reduce contribution incentives. Once the tipping point is reached these incentives change. Mcdonald's for example started to buy back franchises pressuring them to revoke their license. This can be solved in two ways;
- By giving voting rights to the local entrepreneurs.
- By allowing the entrepreneurs to provide additional value to their local communities.
If third parties provide heterogenous value to platform users, researchers tend to call that platform an ecosystem. If the value created by the third parties is high enough I think they can credible threaten to leave the platform. Such a power balance could also enable collaboration. I have to think more about this particular aspect though.
Someone I personally know very well has built a company for 30 years distributing enterprise software to the German market. The license he was granted during that entire time was revocable within 180 days. He told me that his strategy to reduce his risk of being screwed over by the mother organization was simply to become irreplaceable. That would be equivalent to the local entrepreneurs adding to their value proposition.
Fractalized DAOs
Earlier we said that local structures are sort of fractals. The more you “zoom in” the more structures you see. That implies local entrepreneurs could further incentivize sub-entrepreneurs to contribute towards bootstrapping even smaller markets. We’d essentially build a fractalized global DAO franchising empire.
Such structures actually indeed exist in the context of traditional franchising. Local incentivization on the lowest level could get extremely diverse and targeted to a specific community or use case. Note that this particular community could still be global in terms of their physical location, which means we’d still need global agreements to incentivize them.
Here lies a nice intuition, or a different way of looking at it;
If Ethereum allowed us to reward very homogenous behavior credibly (providing liquidity or investement), the IC will allow us to scale the diversity of these incentives to an insane extent.
Each entrepreneur could customize incentives for sub-structures of local markets. This is truly what a cost reduction in deploying smart contracts leads to; much more of them, enabling much more fine-grained collaboration. The crazy part is that each sub-DAO could launch its own token. The cheaper smart contracts get to deploy the more fine-grained credible incentives will become as lower and lower levels of the organization will make use of them.
Cumulative Effect
In the previous article, I described how decentralized platforms are likely to become much larger than centralized ones. Ridiculously this further adds to the explosive cocktail of incentives to adopt them, as it increases the expected profit that local entrepreneurs make from spillover effects. All of that multiplies. How much more value would such platforms create relative to the cost of bootstrapping them?
Let’s say platforms that have decentralized governance are likely to get 10x bigger than their centralized counterparts. The fact that they become bigger increases adoption incentives which reduces coordination costs by let’s say a factor of 5x. Solving the penguin problem feels like the most insane thing of them all so I’ll go nuts and say this makes platforms 100x cheaper to bootstrap. Additionally, we can leverage the creativity of people around the world to find early use-cases to bootstrap for us, let’s give that a 5x.
If all these effects multiply and if the numbers are even remotely accurate guesses (no clue), then it implies an efficiency gain in bootstrapping platforms by a factor of 25'000.
I know that sounds absolutely mad. Maybe I’m missing something, I’m certainly open to any suggestions. But at the same time, we have to realize that we can prevent collaboration failures for the first at an insane scale.
Open Problems
I have so far not wrapped my head around how monetization would work for certain markets where locality is hard to determine. In the case of an on-chain Facebook, it would be much harder to determine which entrepreneur should get rewarded compared to an on-chain Uber.
Liquidity Mining
This thing above is NOT liquidity mining. We did not sell any equity so far. We merely sold specific assets. As if this is not crazy enough yet, I think liquidity mining will also be improved by insane degrees. It will not be called liquidity mining anymore, it won’t be about liquidity. It will, however, still be about incentivizing contributions with equity.
We now not only have vastly more on-chain interactions that we can incentivize but also the ability to specifically increase incentives for specific local entrepreneurs. For example, if someone has already built a subreddit for web developers maybe we can use liquidity mining incentives to further stimulate the adoption of other coding-related subreddits. Targeted liquidity mining could be used as a means of stimulation to help with the infection of local structures that are already heavily exposed to infected areas. Doing so once again plays into everyone’s expectations.
Dynamic Reward Rules
So far the actions that DAOs incentivize are often around some form of staking tokens. The nice thing is that this is sybil resistant as you cannot provide an undesired way of staking. If we were to reward users for posting pictures we have to make sure to get the kinds of pictures we want. Any hard-coded reward rule is likely not to result in desirable outcomes.
What could be a better option is to set aside a given amount of tokens to reward a sub-set of local entrepreneurs. When doing so, the mother DAO would not be in possession of the tokens anymore, but could only choose to whom to distribute them. Stage-wise the entrepreneurs would get rewarded based on a dynamically changing reward rule that they themselves can modify. Locally this would set incentives that the rewards have to benefit the entire mother organization otherwise the mother DAO will decide against future rewards. If the set of entrepreneurs is large enough, each would have incentives to propose rule changes over time to prevent others from exploiting the program. This works as the effect of stopping others from exploiting the system benefits them more than exploiting it themselves. There could be several such programs running at the same time dynamically adjusting incentives. This entire idea could also be applied to each level of the organization.
Madness
If this would all work, what will the world look like?
As platforms would get bigger and bigger, the local sub-structures will gain more and more network effects. On every level of the organization, communities will organize as DAOs, issue tokens, and run liquidity mining programs. So I think it will be a world with plenty of tokens ;). As said before incentives will therefore scale-out in the sense that they will become more diverse and targeted.
Secondary markets will gain in importance, as raising money will be often used to continuously incentivize people, meaning a higher token price will have potentially more direct implications on the real world. As launching token contracts will be nearly free, we’ll literally see millions if not billions of very small markets. Capitalism will change as a result. Investors will decide how money is poured on various small wheels making them spin in the context of a large overall machine. Diversity of thought will be more needed than ever and possibly break the VC monopolies. Investors and influencers will merge by becoming informative agents that point markets towards value creation opportunities.
Local knowledge, as well as trust between people, will allow anybody not only to invest money but their own actions to bootstrap large platforms. It will be much closer to how capitalism should work, rewarding diversity of thought and actions rather than making a closed, privileged, Standford VC industry rich.
Conclusion
This whole article is way too long, so I won’t summarise it fully. All in all the key essence is that global agreements allow us to nurture individual entrepreneurial efforts to bootstrap large platforms much more effectively.
I’ll end with some additional thoughts:
- The Dfinity Foundation already employs an insane number of cryptographers and computer scientists. Why not add some economists?
- Standards around enabling such incentive structures are needed in the long run. It should be both easy and secure for developers to credibly deploy promises around local monetization.
- Shared security across subnets makes all the difference between the IC being a new type of cloud network or true madness.
- We should start to explore if the NNS could be used to construct relational contracts.
All these thoughts might or might not make sense but I hope at least that I inspire someone in the ecosystem to explore ideas around incentive and organizational design.
I’d be really glad to work on more concrete projects in the near future or to talk to more people about either the IC or incentive design. DM me on Twitter if you want to talk about these things.
The next article I’ll write will be around something I want to build. In case someone is out there looking to work on something really cool, let me know.
If I’m only remotely right about this the IC could be one of the most incredible inventions ever created. And we have not even talked about how we can turn it into a global democracy and bootstrap a Balaji-style network state (I like dramatic ends).
Exciting times, complete addiction.
If you appreciate the article follow me on Twitter.
