CEHV’s Blockchain OSI Model Thesis

The Corporate Purpose-Potential Model:

As part of a design for a recent business analysis course I was designing, I put together what I call “The Corpore Purpose-Potential” model that looks at a companies purpose and compares it to its potential valuation range.

Companies that meet this purpose often have a valuation cap within the potential valuation range. When they are at the upper bounds of that range (and sometimes up to 20% beyond it) they are often either expanding into the next range, or overvalued (in backtesting this also worked as a great method to find overvalued public tech companies.)

Companies can move up and down the model freely, and change their purpose over time, but the purpose they fulfill dictates their range potential.

Company Potential Valuation	Purpose	Examples
$10M – $50M	A company that solves a niche problem for a subset of users, such as point-of-sale but only for pizza places, or only doing email mail merges but not the rest of the email newsletter. It is ultimately a feature that could be part of other services but that they are failing to properly address.	Zubtitle, Canny, WP Fusion, Yet Another Mail Merge, SubmitHub
$100M – $500M	A company that solves a problem space and not just a specific problem, but for a niche audience such as independent business operators.	MonetizeMore, GumRoad, ConvertKit
$1B – $7B	A company that solves an entire problem space but for broad consumers across multiple sectors.	Envato, Liquid, BitFury, StockX, NextDoor, Zenefits, Discord
$10B – $150B	A company that builds a platform aimed at solving a specific problem space by leveraging other creators, users and small businesses.	Stripe, Paytm, Instacart, Coinbase, AirBnB, Shopify
$250B – $500B+	A company that is a platform that solves problems for creating off-network platforms for other at-scale operations.	Facebook, Google, Amazon (AWS), Microsoft

This model itself is primarily helpful for those “Application Layer” companies and the superset “Platform Layer” companies, in understanding where their valuation potential is. But, when looking for value in the OSI stack, it is only one of three sectors we find value in.

The Three Sectors of Value:
In the classic OSI stack, I was able to find three operational sectors in which most of the value extraction takes place.

We can divide these into:

• Enabler Sectors
• Standards Sectors
• Interaction Sectors

Enabler Sectors:
Enabler sectors are the necessary evils of the OSI stack. They are the core pieces of any protocol that make communication and interactivity possible, and their needed in order to enable anything higher up in the stack. These are primarily made up of the physicals, data link and network layers. Because these layers are ultimately a gatekeeper to a larger end goal, they create immense value capture.

For example, in the physical, data link and network layers we find companies that own hardware infrastructure and subscription access to the internet. Examples include AT&T ($210B), Verizon ($237B) and China Mobile ($139B).

Standards Sectors:
In the Standards Sectors, revenue is driven by creating a uniform standard of adoption and building ancillary services around that.

One of the best examples of this is companies creating document standards (like Adobe’s PDF) and building a company around them.

But, it can go even further, where an open source standard can be created but the services around it are the value capture. For example, ICANN’s monopoly over domain names that generates revenues of $217M a year.

Interaction Sectors:
Some of the highest value companies however, exist at the application layer of the stack. These are the tech brands and startups that users frequently interact with. Things like Stripe, Shopify or AirBnB are applications built on top of these protocols where users never really need to be aware what protocol they are using.

Then, the absolute highest value companies exist at some superset layer of the application layer, which I refer to as “the platform layer”

• Microsoft – $1.3T
• Amazon  –  $1.2T
• Alphabet  –  $0.9T
• Facebook –   $0.5T
• Alibaba  –  $0.5T
• Tencent –  $0.5T

These businesses in our valuation potential model are the ones who are enabling the creation of other problem solving businesses.

In understanding where this value extraction takes place, and what drives each layer of the OSI technologically, and understanding the different company valuation potentials, we can start to build a sort of hybrid OSI model that can be used to understand the technologies behind a unified purpose blockchain and understand where we need to invest to make this future a possibility.

Each of the 11 layers roughly (although with outliers) ranges from both the lowest to highest levels of abstraction for end-users (running nodes, vs using MetaMask) and the highest potential value extraction (low margin hardware, to high margin fee abstraction on interaction protocols)

Each layer is also improved and expanded in scope by the robustness of layers below it.

For example, the Presentation Layer of the blockchain OSI model is a series of set standards that allow us to build various Application Layer products and Interaction Layer tools because they are open sourced and pre-defined, and do not require a permissioned model which would allow companies to monopolize these tools.

But, the Presentation Layer can only create standards based on:

1. Information that can be Processed.
2. Which required Information that exists.
3. Which requires the ability to Network information.
4. Which requires Data Linking infrastructure.
5. Which requires Physical layers of data collection.

While not all new Presentation Layer standards will require new information from lower in the stack, we exponentially increase the number of Presentation Layer standards we can create when new features are added lower in the stack.

Dissecting the Stack:

Physical Layer:

The physical layer of the stack is where unique physical architecture is created and either connects us directly in a peer-to-peer fashion, or helps capture data from our physical world and process it into data higher in the stack.

Examples here include Helium with their LoFi network, TxTenna with their offline peer-to-peer transaction relay network and FOAM with their physical information processing.

The goal of the physical layer is to provide the infrastructure for connecting to one another and parsing our physical world.

In most cases, the information gathered or connected to here, needs to be processed by the Data Link layer and connected across a network layer before it can move further into the stack. However, many businesses choose to combine their Physical and Data Link layers such as with Helium’s network hubs acting at both layers of the stack.

Data Link Layer:

The Data Link layer can best be thought of as the software of the Physical layer. While it often runs on distinct hardware (mining units for blockchains or hubs/routers for the internet) the goal of the Data Link layer is to provide the connectivity and procedural logic for transmitting data between entities, and in turn catching any errors that take place at the physical layer.

For blockchain’s this happens in miners, validators, and block producers.

Most of our Data Link layer as an industry is actually baked-in to the open source consensus protocols that run the blockchain, but, we can’t overlook the fact that specialized hardware exists and can lend unique traits to the blockchain.

Network Layer:

The Network layer aims to connect any two physical nodes of a network by correctly routing information.

In the blockchain space, we actually have two sub-types of Network layer.

The Micro-Network Layer:

In a classic blockchain sense, our network layer consists of the blockchain protocol we are using (for example, Ethereum) and carrying the data between nodes on the classic Network layer internet protocols (TCP/IP).

In this model, our network is the “Ethereum Network” and there isn’t much room for change or improvement. We’re leveraging an existing tech stack, and a strict standards protocol that must be observed to maintain consensus.

Most people make the mistake of considering this to be the only type of Network layer that exists in the blockchain space.

The Macro-Network Layer:

Given that the core of my thesis is, that all blockchain’s are one type of homogenous evolving network, and not in fact distinct networks, then we must look at the Network layer through a different lens.

Instead, each blockchain itself can also be viewed as a node within our holistic network and the Network layer of our stack are the tools and protocols that connect these nodes together, such as Cosmos’ SDK that aims to bridge blockchain data.

Given these two distinct sub-models of a Network layer within the blockchain OSI model, I think it is fair to presume that the primary purpose of this layer moving forward will be in developing communication between individual chains as we begin to move towards the single network view of the blockchain space.

However, when we talk about connecting nodes to the data, we also need to think of users as nodes within this system. Tools that help to connect us to the communications of the network in a meaningful way fall to this layer. So tools like Infura and API3 that help us to access and digest blockchain information in other environments, the gathering, storing and processing of the information happens at the next layer up, the Information layer, but it is made accessible here in the Network layer.

Information Layer:

The Information layer is the set of standards, procedures and protocols of how we index, store and sort information.

Blockchains are good at storing very specific sets of information, those that are core to their protocol in their non-malleable ledger. But, they are inherently bad at storing other data. From file hosting, to web hosting, structured identity information or private data, there are a lot of key gaps in what information stores well on most standard blockchains. Worse still information on a blockchain is ever growing and difficult to parse and categorize.

The information layer gives us tools like TheGraph for creating standards of indexing and querying information, Sia and Filecoin for storing files and data, and 3Box for creating structured personal data standards.

Processing Layer:

The Processing layer of the OSI stack, is one of the broadest and most diverse in terms of its specific end goals.

If we think of blockchains as one unified network like the internet, then the systems that exist at the Processing layer are the specialized hardware on which we operate. Either the high-end data centers for storing complicated information, or the specialized processing units used in video rendering. The Processing layer is the layer that packs a punch and helps us make specialized or speedy transactions.

Their purpose can range but the Processing layer’s ultimate ability is to extend the feature set of the settlement layer. Many protocols that exist at this layer are robust enough to be their own standalone products, but they add new niche firepower to decentralized systems in the way that they process information. That processing nearly always comes with some trade-off, just like the trade-off in Ethereum’s accessible, decentralized settlement layer, means that it cannot attain the ability to perform some of these niche high-powered processing needs.

This layer cements the concept of a unified homogeneous blockchain network, as no single chain can achieve by itself the same power as these systems combined.

Whether it is the rendering system of Golem, the high speed transactions of Solana, or the on-chain random number generation possible with xDAI, each of these chains add extensive new power to the blockchain ecosystem, which rather than viewing adversarially, should be viewed as new features.

Presentation Layer:

The Presentation layer is one of the most important layers for growing and developing an open system, and yet it is often the most underfunded, least lucrative and most forgotten in any systems model.

The goal of the Presentation layer is to create systematic standards of how we encode and interpret information.

A great example in classic information systems is the JPEG standard.

All information on your computer is a simple collection of 1s and 0s. All programming, all programs and all standards, are just a set of rules we use to abstract or encode this information further.

You can think of it as an encoder ring for encoding secret messages where A = 1, B = 2, C = 3 and so on. The difference with these methods of abstraction, is at each layer we try to make the data more efficient (take up less space or do something quicker) or add new features.

That means our encoding is more complicated. Maybe we still follow the A = 1, B = 2 rules, but we add some new rules, such as “the word And = 27”. So “1, 27, 2” = “A and B” but it takes up less character space so it is more efficient.

The problem with creating rules like this, is that they aren’t efficient if other people (or programs) don’t know how to process that information. So instead of everyone trying to encode and abstract information in their own way, we instead create standards.

These standards are what allow us to know how to interact with data in a specific way. When a computer sees a file named .jpeg, it knows to read all the bytes and translate it to colors and pixels, where as if that same file was named .mp3 it would try and play it as sounds.

In the blockchain, we’re beginning to develop some very important standards at the presentation layer, the most common of which so far are the ERC-20 and ERC-721 standards.

These are standard sets of rules that tell us which features, functions and procedures a token must contain (or in some cases not contain) to be part of the standard. This in turn makes it easy for DApps to support all these different tokens without needing to build custom implementations for each one.

Programs like Uniswap can only exist because most tokens follow the ERC-20 standard (or some derivative of it).

Another important standard we’re seeing created is with ENS names for information lookup and resolution.

The interesting component of the Presentation layer is that it seldom adds new features to the core of the technology stack. It simply enables entirely new sectors by creating a standard set of rules to sort, encode, and structure components that already exist.

Investment in the Presentation layer is crucial for unlocking the potential of our systems.

Software Layer:

The Software layer creates tools that we use to develop, deploy and manage Application or Interaction layer programs, often using existing Presentation layer standards to do so.

To many users of blockchain technology, this layer is a bit opaque, but you can almost think of it as the automated interaction, creation and programmatic deployment of standards created in the Presentation layer.

For example, we have Argent Wallets. Each Argent Wallet is actually a unique smart contract that is created based on the Argent Wallet Factory which deploys a new copy of the wallet code for each user.

The comparison to the old world here, is imaging a piece of software like Microsoft Word. Your copy of Word, which classically came on a CD, was just a clone of the original Microsoft Word program that was deployed on your computer as a local instance. Just like each Argent wallet is your local instance of the main Argent Wallet code.

Application Layer:

The Application layer is where most users interact with actual programs that use collections of the underlying protocols.

These end user applications are the first layer that exist without their primary purpose being the enablement of opportunities above them in the stack.

Instead, the applications hold intrinsic value to the user. This could be a game providing join, or a money market providing financial opportunity. But they are designed for users consumption and not for the adoption, enablement or access of underlying technology.

For example, Axie users don’t really care what goes on beneath the stack, as long as it works.

Interaction Layer:

The Interaction layer are specific software tool sets that allow users to easily interact with applications, or sometimes even lower stack level protocols, usually through both connectivity value propositions (MeaMask helping us connect to Infura and enabling Web3 in the browser) but also through UI/UX value propositions (Zapper making it easy to interact with protocols).

The interaction layer is often where the ownership of audience takes place. Compare for example to Chrome, which would be an interaction style layer for the classic web. Chrome audience users don’t care what goes on in the rest of the stack, and Chrome can direct them to new applications and protocols at a whim. If Chrome decides to change the default services in their interaction experience, the users will go along with that.

Platform Layer:

The Platform layer is one that hasn’t yet been captured in the blockchain space. It will be the platforms that make it trivial for others to build new lower stack solutions, or applications that solve either builder problems or consumer problem opportunities.

We’re starting to see fragments of this exist, such as some of the toolings and standards that OpenZepplin create around smart contracts, or Gnosis Safe’s for treasury and multisig management, and QuikNode for running nodes, but these fractured elements still lack a strong unified platform play, and ultimately lack the abstraction for end user consumption.

Some would argue that in the classic web, the most valuable Platform businesses are Microsoft Azure or Amazon AWS, and I think that is true from a raw revenue perspective.

I think in terms of adding the most value to the internet as a whole, the winner is WordPress, who made it easy for anyone to start a website. That ultimately drove value for the web, demand for websites and in turn demand for the platform hosting services.

It will be interesting to see how we can abstract away the complexities of blockchain creation.

Settlement Layer:

Finally we have the Settlement layer.

In terms of a traditional internet, it’s a bit of an abstract concept.

On the one hand, a settlement layer is a backbone that connects multiple layers together. That’s somewhat akin to the core backbone cables of the internet that exist at physicals layers.

But, more over, a settlement layer is a universal truth, finality and security. That is something that doesn’t exist in the classic internet.

In a unified blockchain space, the reality is that decentralization is a spectrum, and it is one that is counter-posed to both accessibility and affordability.

Ethereum has made it their goal to always be the most accessible and secure of blockchains, with the target for a post-sharding world to be that anyone can affordably run an Ethereum validator on any device.

Other blockchain protocols will prioritize affordability/efficiency by making small trade-offs in how accessible and how decentralized they are.

It’s a necessary trade-off, as even with Ethereum’s sharding and roll-ups, there are still things Ethereum isn’t good at (like file storage, or image processing) and it will always have some upperbounds through-put that is lower than chains who make different pareto efficient trade-offs.

But, it can always act as an accessible, secure, and decentralized settlement layer, as well as a source of ‘universal truth’ in a multi-chain world where other chains may dispute states between themselves rather than between in network validators.

Why Does the Blockchain OSI Matter?