Is the value capture of blockchain at the protocol layer or the application layer?

Author: Beichen, Chain Teahouse

In 2016, Joel Monegro from top venture capital firm USV published "Fat Protocols," a classic crypto paper that compared the differences in technology stacks between the internet and blockchain, leading to a starkly opposite situation in value capture—where the internet is characterized by "thin protocols, fat applications," while blockchain is "fat protocols, thin applications."

Over the past six years, this viewpoint has been largely validated by the market, as the value created and captured at the protocol layer far exceeds that of the applications built on top of them. However, it now seems that the situation is changing—applications like Uniswap, GMX, and AAVE have captured value that far exceeds that of Ethereum and are still growing rapidly.

The crypto market appears to be leaning towards the perspective presented in Multicoin Capital's 2018 paper "Thin Protocols," which argues that applications built on protocols control the end users and are always attempting to siphon off value, leaving protocols powerless in this regard.

I fully agree with USV's fundamental viewpoint on fat protocols, and I also completely endorse Multicoin's thin protocols perspective, as the two are not contradictory at all. They both use the term "Protocols" to describe their respective viewpoints, which actually point to different things, leading to a mere misunderstanding in language.

So, where will the value on the blockchain flow—towards the protocol layer or the application layer?

USV's viewpoint is that while internet protocols (TCP/IP, HTTP, SMTP, etc.) since the 1990s have created immeasurable value, much of it has been created indirectly, with most value captured and re-aggregated at the application layer (mainly data). Therefore, internet investments have been concentrated in applications, while the return on investment in protocol technology has been very low.

Blockchain protocols, however, are entirely different, as blockchain serves as a shared data layer and also monetizes "access," thus avoiding the misfortune of internet protocols (where value is almost entirely siphoned off by the application layer). In fact, it can even reverse the situation, allowing the majority of value to flow towards the protocol layer, resulting in a "fat protocol, thin application" scenario.

I strongly agree with USV's analytical approach, but I have reservations about the conclusion, as that article simplifies the blockchain technology stack directly into protocol layer and application layer. The subsequent controversies stem from this simplification, and we should discuss it based on a more precise layering of the technology stack.

The most classic layering of the internet is the OSI seven-layer model, and there are many types of layering models for blockchain, with a typical one being six layers (the additional bottom layer exists because blockchain itself relies on the internet to function).

In USV's discussion, the "application layer" refers to the application layer in the above diagram, while the "protocol layer" is a collective term for the layers below the application layer. The problem is that the application layer of blockchain cannot be directly compared to the application layer of the internet.

The application layer of the internet siphons off most of the value, but more accurately, it is the end products of the application layer that siphon off the value, rather than the protocols of the application layer (e.g., the HTTP protocol). Therefore, USV's "thin protocols" encompass the entire technology stack of the application layer, while "fat applications" specifically refer to the end products of the application layer.

Although blockchain can also be divided into many layers, the network layer, data layer, consensus layer, and incentive layer are bundled together by public chains, while the contract layer plays a role similar to that of internet application layer protocols, with end applications built on top of it.

Multicoin's thin protocol perspective is illustrated using the decentralized prediction market Augur on Ethereum and the end product AugurCo (a hypothetical product based on Augur) as examples.

Thus, the rebuttal of USV's article regarding protocols in the Multicoin article is largely a misunderstanding of vocabulary—Multicoin's "fat applications, thin protocols" correspond to smart contracts and end applications, while USV's "fat protocols, thin applications" correspond to public chains and smart contracts + end applications.

At this point, the question of who captures the value remains unresolved, but it is undeniable that compared to the internet, public chains will capture a significant portion of the value.

USV was optimistic about investing in public chains in 2016 because there were not many applications at that time, nor many public chains, so most of the value flowed towards the public chains filled with imaginative potential. However, with the development over these years, various applications based on blockchain have begun to flourish, creating real value and revenue (especially in the DeFi sector), and the proportion of value captured by public chains has naturally decreased (though absolute value continues to grow rapidly).

However, at least for the current stage and for a considerable period in the future, Multicoin's "fat applications, thin protocols," or "fat end applications, thin smart contracts" will not be realized.

The current stage of blockchain is still in the wheel-building phase, so people often treat critical components like wheels, chassis, and engines as end products (which is also why the experience is so poor). However, when the key components that achieve the "minimum functionality" in various application fields are perfected, the possibility of "composability" will have room to unfold.

Because blockchain is a shared data layer, these key components are open and can be used infinitely, so later developers can combine these key components and simply optimize them to achieve more refined functionalities (or simply a better experience). This is where the imagination of blockchain "composability" lies, and it is also where the imagination of "cloud-native" in the internet field resides.

However, the result of this will be that, in addition to public chains capturing a portion of the value, those smart contracts that are used infinitely and also monetize "access" will siphon off value, leaving relatively little for end products.

In fact, Multicoin's example of Augur illustrates that at the time, using Augur's smart contracts was free, so value would be captured by other protocols or end applications, while today, the vast majority of smart contracts are charged.

This may raise a question of whether it is unfair to developers of later end products. In fact, from a business perspective, the costs of such end products are also much lower (public chains have solved the high operational costs of internet products, and composability has solved development costs), so the returns are reasonable, unless one develops disruptive smart contracts at the contract layer.

Finally, to summarize briefly: public chains are still worth investing in, but now there are many more application options. Specifically regarding applications, we should focus on those products that can "achieve minimum functionality" at the contract layer (a non-controversial example is Uniswap), rather than expecting them to have complete tools and smooth experiences like internet products.

Because a smooth experience is not the priority issue for blockchain products at this stage (first solve the existing problems), and complete tools are a typical internet mindset, as internet products either become large applications or fail, with almost no middle ground. Blockchain products can focus on specific functionalities and still be fully utilized by composability.