Some are thin.
And some are fat.
The fat one has
A yellow hat.
Dr. Seuss — One Fish, Two Fish, Red Fish, Blue Fish
Joel Monegro’s fat protocol thesis is one of the most important ideas for understanding cryptocurrency.
Published in 2014, the thesis states that, on the web, value accrued to the application (app) layer: companies like Facebook, Google and Amazon.
While shared protocols like SMTP, TCP/IP and HTTP provided a huge amount of value, they didn’t have any way to capture that value.
This would be like if the creators of SMTP (the email delivery protocol) were wealthier than Google’s founders.
Let’s take a look.
So far, this seems to be true. To date, value seems to accrue more to the cryptocurrency protocol layer: Bitcoin and Ethereum.
However, a number of other analysts have asserted that the fat protocol thesis is misunderstood, lacks nuance or is simply wrong.
There’s three reasons to believe that cryptocurrency protocols will be fat.
One is that in the blockchain stack, data is stored on the blockchain, a distributed database that everyone can hold a copy of, instead of the siloed, walled gardens of the app layer where data accrues on the web.
The core competitive advantage for most software companies is their database, which is why companies like Amazon, Facebook and Google guard their data so closely. The blockchain removes this competitive advantage pushing more of the value capture down the stack to the protocol layer.
As an example, consider how hard it is for a social network to compete with Facebook. Why would a user switch to an new application that’s a ghost town?
By contrast, switching between different cryptocurrency exchanges is easy. They all have access to the same underlying blockchain. If Facebook’s data were on a public blockchain, it would be much easier to fork the blockchain and create a competing social network.
The second is that cryptocurrency protocols have a higher level of awareness about the application layer than their web counterparts because they are more highly structured. SMTP, the thin internet protocol, is highly unstructured and so has a low level of awareness of how it is being used.
For example, when you send an invoice via email, the email can be structured or formatted in any arbitrary way you like as long as the receiver of the email is able to understand your particular style of formatting.
The data structuring is done on the application layer by companies like Paypal, which are then able to extract the value by charging a fee for the value they are providing.
However, because SMTP is an unaware protocol, you could structure the invoice by saying:
“Mail your payment in cash to 123 Alphabet Blvd. Aliceton, KY 12345.”
Most people prefer Paypal because the fee it charges is worth it, relative to the increased speed, reduced risk and more professional impression it gives as opposed to mailing cash, but the unstructured nature of the protocol allows either.
With a fat protocol, the network is aware of important details of the invoice such as the amount, due date, and recipient. The network can then take appropriate actions like sending reminders, debiting accounts and recording transactions to relevant ledgers — actions that were reserved for applications on the web.
Because cryptocurrency protocols are more aware of what is happening, who the parties involved are, and what the economics are, it’s better able to charge for the value being created.
The third reason to believe protocols will be fat is that there is a greater incentive for entrepreneurs to build these protocols because they are able to create tokens, which can let them capture some of the value as usage of the protocol increases.
While much of the market capitalization today is speculation, eventually decentralized applications will need to use networks like Ethereum or NEO for computing power and Filecoin, Sia or STORJ for storage. This usage will increase the value of the tokens held by the founding team.
By contrast, on the web, creators of protocols like SMTP did not capture any value directly.
While these are the three fundamental forces driving protocols towards fatness over the long-term, there is another shorter-term dynamic at play as well right now.
As a token increases in value, such as Bitcoin and Ethereum have over the past year, it draws in speculators, developers and entrepreneurs who purchase tokens and become stakeholders that are financially invested in the protocol itself. They then begin to build applications on top of the network.
When applications begin to show early signs of success, new users are drawn to the protocol, increasing demand for tokens. At the same time, current token holders hold onto their tokens, anticipating they are going to increase in price even further.
This then brings in a whole new wave of speculators and the process begins again.
The dynamic is significant because, at least in the short-term:
“the market cap of the protocol always grows faster than the combined value of the applications built on top, since the success of the application layer drives further speculation at the protocol layer.”
We see this clearly in Bitcoin. In August of 2017, Coinbase, the largest bitcoin application, raised $100 million at a $1.6 billion valuation. At the same time, the value of the Bitcoin protocol was $73.5 billion.
The implication that many have drawn from an investment perspective is that cryptocurrency investors should be more focused on the protocol layer than the application layer.
By buying the bottom of the stack, they argue they are diversifying across everything built on top. It doesn’t matter if Coinbase or some other exchange becomes the biggest player, they will benefit regardless.
However, this ignores another important difference between the blockchain and web stack: unlike web apps, protocols can be forked.
Drop your email here to sign up for my newsletter and get the latest updates.
If you don’t like how a protocol works, you can fork it. Bitcoin Cash and Ethereum Classic are the first of many.
Forking means that the entire protocol, including the database, can be copied by anyone. This allows the creation of a more competitive market with lower barriers to entry.
Because of forking, the incentive of anyone developing a blockchain protocol is to build a protocol that will provide their service at the lowest cost. If they don’t, someone can just fork it and undercut them.
This means as a protocol becomes increasingly fat, it’s increasingly likely that it will be forked into a sub-protocol that delivers a part of the functionality in a more effective way.
For example, if a decentralized file storage protocol gets fat, it could be forked into a new protocol that specializes in a particular type of file storage (say, by industry), which would be able to launch with the value proposition of “we’re exactly like those guys (we literally copied their product) except cheaper and better for your industry’s use case.”
The end result is likely to to be a large number of competing protocols at every layer of the blockchain stack.
This was not possible on the app layer of the web. You can’t fork Facebook and optimize for discovering the best longform articles instead of baby photos and outrage.
So while the protocol layer as a whole may be fat, individual protocols will likely end up being thin due to forking and a competitive market.
You could argue that this doesn’t change the investing thesis of fat protocols because by investing in the base level protocol, you also get exposure to all the forks.
However, there is also the more traditional type of “forking,” where new generations of companies learn from the past one’s failures.
Google did not fork AltaVista, but they were able to learn from its failures in a way that ultimately played a major role in their success. Ditto with Facebook and Myspace, along with many others.
Ethereum is not a formal fork of bitcoin, but clearly incorporated many components of Bitcoin.
Monero and ZCash are also not formal forks of Bitcoin, but can effectively be described as ‘like Bitcoin, but with better privacy and anonymity.’
The open source nature of the cryptocurrency industry makes this even easier. Combine this with interoperability and you can see where the fat protocol thesis might break down from an investment perspective.
An app that previously used bitcoin but preferred the privacy offered by Monero could migrate over and owning bitcoin would give you no exposure to whatever value that app is creating.
So, while cryptocurrency protocols as a whole will be fat, an individual protocol’s fatness seems like it will be highly relative to the value they are creating, not unlike web applications today.
There are (at least) two important implications I’ve been thinking about regarding this fat/thin dynamic.
Sources and Further Reading on the Fat vs. Thin Dynamic
In the traditional economy, the marginal benefits of scale are resilient. Once you get to a certain scale, it becomes very hard for other companies to compete with you.
This advantage was conceptualized by Warren Buffet as a moat. Traditional businesses are able to build large moats, which make them defensible and hard to outcompete.
Businesses based on network effects — like Facebook — have particularly large moats. The more people use Facebook, the larger its moat becomes.
Even when the company ceases to create much additional value and moves into a late-stage cash cow mode, its moat makes it difficult for other companies to compete. Paypal hasn’t done anything innovative in a decade but its moat makes it difficult for other companies to compete and keeps Paypal profitable.
In a world eaten by blockchains, the graph looks different. There is less resilience because of the possibility of forking.
It might not makes sense to fork a protocol that’s worth $50 million, but as it becomes more popular and it’s value increases to $50 billion, it becomes increasingly likely that someone will fork it.
The ease of forking will increase over time, further lowering the barriers to entry in the same way we saw with web products getting easier to build. The same website a company might have paid hundreds of thousands of dollars to build and maintain in the late ’90s can now be built for tens or hundreds of dollars because of projects like WordPress and Squarespace.
So while forking is difficult today, it will become easier over time lowering the scale at which forking is viable.
In effect, this means blockchains seem to have a natural mechanism that prevents monopoly.
Instead of a single, monolithic Facebook, there may be dozens of forked Facebooks, which users choose based on their preferences.
The Blockchain combination of shared open data with an incentive system that prevents a winner-takes-all system means that markets will shift away from platform monopolies and toward users.
Peter Thiel has pointed out that historically there has been very little correlation between value creation and value capture. A business may be able to create a lot of value but not capture it well. For example, most news and investigative reporting is done by text-based media businesses like the Wall St. Journal and New York Times but far more of the value is captured by television news properties.
Other businesses are able to capture a larger portion of the value they create. Software-as-a-service (SaaS) businesses have become so popular in large part due to the fact that the recurring subscription business model is such an effective value capture mechanism.
If you’ve invested a huge amount of resources into setting up a marketing automation suite, then the switching cost is very high because the set-up (lead scoring, tagging, automations) is all stored on the company’s database, not yours.
Even if you find a service that you like better, it has to be much, much better to justify the switching cost.
On the blockchain however, the user would own all (or at least significantly more) of their own data, making it easy to switch service providers.
These reduced switching costs would seem to indicate that value creation and value capture will become more closely correlated.
If a company shifts from value creation towards rent-seeking, users will more easily be able to shift to competing applications. This should incentivize companies to focus more on creating good products, and reduce the amount of resources spent on rent-seeking behavior.
Alternatively, this might just incentivizing creation of protocols which are less forkable so they can retain more of the long-term value created, or ponzi-scheme protocols that can somehow partially capture value of child forks.
I think that while we will see many of these protocols, they are unlikely to win in the long run. All other things being equal, consumers are likely to pick open over closed protocols. In the past, all other things were not equal because there was no effective way for creators of open protocols to capture the value they were creating. With tokenization, there is now a sufficient incentive to build those open protocols — which just wasn’t the case in the early days of the internet.
This also raises the question of how investment returns will be distributed over the life of a company (or protocol).
Though it’s difficult to quantify, I suspect a large amount of the returns generated by larger cap, late stage companies is rent-seeking rather than value creating.
They are able to do this because of the moat they’ve built around their business.
In a blockchain dominated world, those moats will disappear and investment returns would all shift earlier in the company lifecycle.
Forking is a mechanism akin to biological reproduction. It allows more subspecies to be produced more rapidly, which means the blockchain ecosystem could be both more resilient and more efficient than its traditional economy counterpart.
Blockchains allow for what Nick Szabo has called social scalability: high levels of coordination with low levels of centralization.
This is a big deal because traditionally there has been a large trade off between resilience (low levels of coordination and low levels of centralization) and efficiency (high levels of coordination and high levels of centralization).
Biological evolution seems to favor resilience. Cockroaches have been around for 320 million years because they’re really really good at not dying.
There is always redundancy in biological systems which lets it trade off short-term efficiency for long-term resilience. This bias for resilience is why life has existed on Earth for four billion years. It’s also why we are fairly resilient creatures. You don’t need two kidneys, but it’s sure nice to have a backup, just in case.
In most human systems, however, we have opted for short-term efficiency at the cost of long-term resilience.
Consider the financial system today. There is a huge amount of centralization because of (relatively) short-term efficiency gains. The ability to access the global market for a business offers a huge efficiency potential, but requires large amounts of centralization. You need a trusted third party or intermediary that already has global reach to leverage.
Nation-states and their diplomats negotiating bilateral trade agreements, and banks with presences (or partnerships) in all the jurisdictions you want to reach in, offer the coordination it takes to reach global markets, but require a high level of centralization.
Given the choice between going global and going bankrupt, all businesses go global. A business that refuses to make this compromise cuts off their growth potential ensuring that the largest companies will always be the ones that have made the compromise. Refusing to take part in a global market is refusing to remain competitive at the highest levels.
However, the high levels of centralization and interdependency that make this increased efficiency possible also creates the systemic risk conditions for what we saw happen in 2008 with the global financial crisis.
This global, centralized system creates a black swan-prone environment. Companies must take part in this highly coordinated and highly centralized system in order to be efficient enough to remain competitive, but in doing so, they expose themselves to systemic risk, which everyone ended up paying for through government sponsored bailout.
The social scalability enabled by blockchains offers a potential way out. A blockchain ecosystem composed of many thin protocols that are highly interoperable could offer the same, or better, levels of efficiency and coordination as the traditional economy while maintaining a low level of centralization (and thus increased resiliency).
A cryptobank could offer the same access to global markets as existing financial institutions, but with more transparency and less risk. Depositors and shareholders could see the bank’s reserves and lending in real time, eliminating the information asymmetry in the current system.
A cryptobank structured algorithmically could also be self-liquidating. The instant that a cryptobank passed into insolvency, its underlying assets would be automatically disbursed to shareholders and depositors.
Had this been possible in 2008, things would would have been very different. Instead of the losses mounting up and then having to be bailed out, the companies that crossed the threshold into insolvency would have had their assets immediately distributed.
In the short run, cryptocurrency protocols will remain fat with the lower-level protocols capturing the majority of the value. In the long-run, interoperability and forking will make the value captured by a protocol roughly commensurate with the value it is creating whether it is at the bottom, middle or top of the stack.
This same interoperability and ability to fork could substantially change the market dynamics.
For one, the notion of a moat may be less relevant than it was in the traditional economy, moving investment returns further forward in a company’s life cycle.
From a macroeconomic perspective, it could facilitate an even better coordinated global system than we have today while removing much of the systemic risk.
Drop your email here to sign up for my newsletter and get the latest updates.
As with all my essays, the writing is more about exposing my process of thinking. Thoughts, feedback and criticism are welcome on Twitter, in the comments below, or email (my first name at taylorpearson dot me).
Acknowledgements: Thanks to Gary Basin and Tom Howard for reviewing early drafts and providing feedback.