Let’s think of Blockchain as a deployment option. That is, one has a program and then needs a personal computer, a server or a cloud to run it on. Or a Blockchain. Ethereum is the archetypal Blockchain-as-a-deployment option, but even Bitcoin can be considered to be a deployment option — albeit one that runs just a single program.
It is deployment with special characteristics. For the purpose of this article I’ll single out the decentralised character of Blockchain deployment. This works out primarily as non-censorship (no one is in a position to bar someone else from running a program). It also adds to availability, of course: execution of a program on a blockchain run by many nodes is not as vulnerable as when it runs on a single machine.
Now I will argue that, with any Blockchain scheme, decentralisation will be a temporary characteristic. I will argue that, in the end, there will be just a single or very few nodes running the scheme.
This is because of two inherent characteristics of a Blockchain scheme:
- it requires quite substantive computing resources;
- its network consists of two types of nodes: end users and miners, where end users need the miners.
This creates a business opportunity for mining. In other words, a Blockchain scheme is a paid deployment scheme.
Now, with most automated solutions, bigger is better. One obvious cause of this is that some costs are constant (such as investing in and maintaining operating knowledge) if the operation scales up, while income increases. This holds for Blockchain mining, too. So a large miner is better off than a small miner.
This creates economic pressure in the direction of ever larger mining operations. Network growth allows an operation to grow; but so do mergers. Hence, we see a centre-seeking commercial force that drives any decentralised Blockchain to a centralised one.
Note that this has nothing to do with the technical features of a particular Blockchain scheme. Neither is the economic force I’ve identified specific for Blockchain. Just think of cloud serving: how many operators are there, really?
One might argue that a scheme of collaboration could be devised that will keep the number of miners above a minimum. No doubt this is possible, but it would require a tightly run ship with stringent, mutual, binding agreements and some agency to enforce them. Such a collaboration doesn’t sound like an open, decentralised situation to me.
So if you really need decentralisation, you’ll need a deployment option that has no business case. As far as I can see, that means you’ll have to bring your own device. In other words, we have to go all the way to truly distributed computing. The network will have to be a network of peers.
In the context of digital cash, this introduces the problem of ‘double spending’. Put more generally: if everyone is allowed to run the administration of his own money, how do we prevent cheating?
For a summary of a solution of that problem I refer to: Why We Need Witnessing For Digital Cash.
The full treatment is given in: “On the Principles of Financial Transactions Secured By Witnesses”.