The subtle tyranny of blockchain

The past few months have been a new chapter in the development of blockchain technology. The DAO hack triggered the Ethereum hard fork, and the Bitcoin block size debate has not been resolved to this day. All of this has led to dissatisfaction among members of the respective communities, and even disappointed key figures in the cryptocurrency community.

I left the Bitcoin community in 2012 for very similar reasons. Back in 2011, I helped Gavin Andresen design the P2SH function. This design is not very complicated, it is backward compatible, and it provides an important cornerstone for improving the security and performance of Bitcoin.

In the blockchain world, everyone must have the same ideas

Unfortunately, achieving consensus on blockchains has proven to be extremely political. As the price of Bitcoin rises, the number of stakeholders will expand, and money will increasingly dominate technical discussions.

In this context, it is not surprising that what happened to Ethereum recently. As a blockchain grows, with a large and highly vested user base, it becomes increasingly difficult to convince. When combined with time pressures (such as The DAO’s 27-day split creation period), something has to give. There is not enough time to do a safe hard fork for a complex system like Ethereum.

The fundamental difficulty in updating a blockchain is maintaining a shared state. In any protocol, everyone must act in agreement. For a blockchain like Ethereum, everyone must think in agreement. Everyone’s memory (also known as “state” in computer science terms) must evolve together according to the same rules.

Shared state adds a lot of complexity, which has a big impact on developers. The fact that blockchain has been largely ignored by major technology companies but embraced by the financial industry is partly due to the fact that the financial industry has a high tolerance for mysterious and complex systems.

Harmony and consensus are valuable. If we don’t agree on who’s president or how much money we have in our bank accounts, society won’t work. But if harmony is taken to an extreme, it can also cause harm. The dystopia presented in The Lego Movie, where “everything is awesome” is just the surface, is a world of vast diversity and rapid change that doesn’t conform to the established consensus.

So how do we strike a balance between too much or too little consensus?

Project Xanadu and the Web

I hope everyone is familiar with the World Wide Web. But what you may not know is that there is an older project that has been around since the 60s called Xanadu. Not only has Xanadu been around longer, but it also has a much more ambitious feature set than the Web.

But in the end the Web won out for many reasons, but I believe the Web's stateless architecture was the key to its success. Both Xanadu and the Web were decentralized, but the Web was much simpler. All it needed was a minimal protocol, and the simplest data format. Websites didn't need to interact with each other, which meant they could evolve independently of each other, and many features that users cared about could be created just by changing the website or client, rather than waiting for the creators of Xanadu to add a feature.

As active participants in the W3C and IETF, we are always fascinated by how the web evolves. For example, HTTP 2 was implemented by Google under the name "SPDY", and Google controls multiple web servers (Google Search, Gmail, etc.) and clients (Google Chrome). The fact that one corner of the web can be updated and good ideas can eventually spread throughout the system is a crucial reason why the web can keep up with the pace of technological innovation.

A better approach

What can the blockchain industry learn from Xanadu and Web standards? Instead of blindly replacing centralized functions with blockchain, we should consider how to avoid using centralized methods to implement these functions. What we need to build is a stateless protocol like the Web.

To illustrate what I'm talking about, let's think about the example of payments. Bitcoin is an alternative to existing centralized ledgers (like the credit card networks). This is undoubtedly a good idea. But Bitcoin still has a lot of shared rules that participants must agree to. In order to agree to the consensus mechanism, I need to use proof of work, I need to agree to the current currency distribution function. I need to agree to the block size limit. I need to accept the lack of anonymity, etc.

In contrast, by adding a layer of abstraction, the Interledger protocol allows me to choose a ledger with the consensus mechanism, currency, performance characteristics, anonymity, etc. that I like.

"My ledger must be driven by love and rainbows!"

This doesn’t mean that Interledger doesn’t need any protocols, we still need a common data format. However, none of these choices affect me economically or politically, which makes it easier to compromise. And, importantly, we don’t share a global state, so at least our thoughts can be our own again.