Blockchain is Inevitable | Raymond Cheng

Skip to main content<br>Since the 2017 ICO boom, the crypto industry has navigated countless hype cycles and narratives. Through it all, my conviction in one core thesis has only strengthened: Blockchain is inevitable.

This isn't a statement about a particular crypto asset, digital token, or network. I believe this to be true for one simple reason: these are Byzantine Fault Tolerant (BFT) systems.

In other words, these are systems designed to behave reliably and exactly as you'd expect, even if individual servers, nodes, or participants are malicious, lying, or acting arbitrarily. This design principle makes them incredibly robust. They are the "cockroach systems" of the digital world.

We as a society have yet to fully reconcile with the fact that we have now built digital systems that can never truly go away.

In this post, I want to break down the ramifications of that fact—first for the crypto industry, then for the broader tech industry, and finally, for the world at large.

But before we explore the consequences, let's establish a foundation. What is a Byzantine Fault Tolerant system, and where did it come from?

Understanding the "Why": What is Byzantine Fault Tolerance? ​

At the heart of computer science is a field called distributed systems, and at the heart of that field is the concept of fault tolerance .

Machines are inherently faulty.

Devices can fail, power can be shut off, and memory chips can get corrupted.

Natural disasters can happen.

Despite this, we want to build global-scale services that remain reliable over a network. This fundamental property is what drives the entire cloud computing industry and is at the heart of modern data center design. It's what allows us to take the infinite scale of modern software for granted.

From Crash Faults to Malicious Actors ​

Most systems we rely on today, like Google Docs or Gmail, are built on crash-fault assumptions. This basically means we believe the worst thing that can happen to a server is that it can crash and fail. We generally assume that systems will behave honestly and won't lie to you.

When you use a service like Google Docs, your data isn't stored in just one data center; it's stored in many different data centers all over the world. A famous protocol called Paxos—a type of "consensus protocol"—acts like a high-frequency voting system. It allows these data centers to come to an agreement (consensus) on the current state of your document, storing it in a highly replicated, fault-tolerant way so it can recover from any single data center failure. These protocols are called “consensus protocols” and they already form the resilient data foundation of many global applications that we use today.

The Byzantine General's Problem ​

Years ago, computer scientists, most notably Turing Award winner Leslie Lamport, started asking a critical question: What happens if these machines start acting adversarially and actually lie to you?

What if a machine in one data center starts advertising that the state of your document is something other than what it actually is? This is known as the "Byzantine General's Problem."

Over the next few decades, scientists created more reliable and practical BFT protocols. What was once an impractical theoretical result became an actual system that could be built. But for a long time, there remained a critical adoption problem. Most cloud providers and scientists derided this body of work, asking, "Why would we ever assume that our own cloud servers are untrustworthy?"

For many years, these protocols saw little use outside of niche applications, like satellites, where cosmic radiation could flip bits in memory and cause a system to act erroneously.

The Catalyst: Bitcoin and the BFT Boom ​

Then came Bitcoin.

In a post-financial crisis world, there was suddenly a massive interest in building a decentralized, peer-to-peer system that could behave correctly even if a fraction of the network was acting maliciously. We were seeing, in real-time, a cratering of trust in centralized institutions.

This coincided with an acceleration of sophisticated hackers, some state-sponsored, who find ways to steal credentials of insiders to install malicious software on critical infrastructure. Increasingly, we are finding we do need to build systems that are resilient, even if insiders are untrustworthy.

Fast forward to today, and there are countless blockchain networks in operation, providing financial ledgers or computation frameworks that are reliable, robust, and truly global.

The core assumption of many of these systems is different.

Instead of relying on assuming that all servers are 100% honest, we assume that only a certain threshold (e.g., ⅔ or ½, depending on the protocol) will behave honestly.

We then use financial incentives to keep them from behaving maliciously.

These systems are designed to withstand a significantly more sophisticated threat model, and we now...