What's so great about decentralization, anyway?

I've been many times a little bit critical about decentralization fanatics who want to decentralize everything as soon as possible.

Here are some quick thoughts about decentralization in the case of blockchains and organizations built on top of on them.

Decentralization of existence

Blockchains are great at existing, at least if we don't count obvious scamchains. It's just a code that runs on several computers around the world. If somebody wants to take down the network, they need to do quite a lot of work for it. If block producers and full nodes are sufficiently decentralized, it is impossible for somebody to attack them successfully at the same time.

The benefit of decentralization of existence is resilience. No matter what happens to some individuals who are taking part in block production – the blockchain replaces them and keeps on going.

This is very valuable feature, for example, for currencies and payment networks. It would suck if a payment network disappeared for some reason.

So if you are thinking some service that should be put on a blockchain, one perspective is to think "how much customers benefit if they can be sure that the service will keep on existing". It is something that a blockchain can provide.

How do you know if decentralization of existence is designed well? You should look the financial incentives. How much the blockchain is paying for block producers? Do full nodes have any incentives? Do the incentives change if block producing gets too centralized? It also matters how the block producers are chosen. Just try to look at the system from every angle and think if it could work in all kinds of different situations.

Can there be too much decentralization? Yes, especially if we think about the economic incentives. To make sure that there are hundreds or thousands of independent block producers and full nodes, we need to pay for them. That is not free. The money has to come from somewhere and usually it's created by inflation. People tend to dislike systems that have too much inflation.

Decentralization of governance

Usually there is some kind of an organization built around the blockchain. Some of the decision making mechanism might be even coded to the blockchain.

Is it a good idea to have as much decentralization as we can in governance, too? Not necessarily. As I wrote in my last post:

According to Coase, firms exist because they lower transaction costs of cooperation. When there is a central party that hires workers and tells them what to do, shit gets done much faster compared to a situation where everybody acts as a contractor and makes individual contracts with everybody else. Transaction costs include non-monetary costs like negotiation cost: it requires time and effort to find suitable business partners and negotiate contracts with them.

The more decentralized is the governance, the more effort there will be needed to achieve consensus. With lots of individuals there is a need to negotiate with everybody, listen to their perspectives and needs, form accetable compromises, and then try to get everybody act based on decisions that were made.

I haven't really seen any good models for really decentralized organizations. Costs of negotiation to form consensus are considerable. Blockchain-based organizations have usually some goals, like making profit, and slow decision making harms achieving those goals. Blockchain markets are global and very little regulated so if an organization wants to thrive, it needs to be able to react fast on the demands of the market forces.

So far we haven't seen much really useful blockchain-based applications. Lots of promises but it has been mostly hype. My guess is that one reason for this is the lack of innovations on organizational level. Blockchains made DAOs possible but people don't know yet how to use a DAO effectively.

Of course we still need some decentralization in governance. If there is only one guy at the top level, the obvious question is: what happens if that guy dies or goes crazy? The governance structure needs to be able replace people. Like in decentralization of existence, decentralization in governance is needed to make sure that the system can work in all situations, including loss of powerful individuals.

Those who can solve the problem of right kind of balance between decentralization and centralization will have great future on blockchain markets. They can dominate by producing services that customers actually need and want to use. They can make visions come true instead of continuous daydreaming.

Timeframe

One thing that seems to be very underestimated is the age of a blockchain system. I've been saying this for a while now: young systems are really different from matured systems.

It's not in anyway shameful to launch a centralized blockchain. It's just very important to make clear that centralization is needed to make sure that the initial development is quick and it goes to the right direction.

Decentralization is a goal for a mature system. When all pieces of the system are tested in the real world, and then redesigned because they are never good in the first try, developers can let it go on its own and become truly decentralized.