Measuring Decentralization is Hard, Lets Go Shopping

People get the general idea that, “if decentralization is the key property of the blockchain we ought to measure it, so we can compare decentralization across chains.”

You’ll see people try and measure decentralization in two ways:

Failed Attempt #1: People try to measure the number of full nodes running in a geographical area.

Another measure of decentralization is to determine what the node distribution is across countries for Bitcoin and Ethereum.

Balaji S. SrinivasanQuantifying Decentralization

The data shows that the Ethereum nodes are both in the latency space, and also geographically more distributed round the world. Ethereum nodes tend to come from all sorts of places, smaller networks, and homegrown entities, as opposed to Bitcoin nodes, which tend to be located in data centres.

Emin Gün SirerGenesis London Conference

There is no way to prove any node isn’t just a proxy to a single central one.

The number of nodes in the network don’t meaningfully add to network propagation or decentralization. The only full node that matters is your own. Even if the number of full nodes mattered, you can’t count them and you can’t prove where they’re running from.

Failed Attempt #2: People make ‘feel good’ pie charts to show how distributed the hashrate of any given network is.

You can’t prove who is really mining on what pool.

It’s true that centralization of hashrate can increase your risk of censorship on the blockchain. The problem with this is that all identifiable information that people use to measure hashrate on miners is self-reported and can easily be faked.

Decentralization may be the most important feature of Bitcoin, and there is no empirical way to measure it.