Delegation and Centralization

Decentralization is a desired property of blockchain networks. It implies more economic freedom for those who are are running it, those who want to join it, and those who are using it. Moreover, it makes them much harder to kill–-think about Bittorrent. We tend to favor systems that are more likely to survive. De/centralization has been explored thoroughly in an article by Vitalik, and I recommend reading it before moving on below.

Blockchain networks consist of agents that perform certain tasks for others, such as creating blocks and including users' transactions. In Proof of Work, they are called miners, and in Proof of Stake, they are called validators.

Delegation is a recurring concept in blockchain networks. From Wikipedia:

Delegation is the assignment of any authority to another person to carry out specific activities.

Any agent can assign authority to any other agent. Agents who assign authority are called delegators, whereas the agents who receive the assignment are called delegates. Below are some examples.

• Bitcoin mining pools: Most mining pools are operated by a centralized entity, called the pool operator, which is responsible for tasks such as operating member hardware, tracking contributions, and distributing block rewards fairly. Miners delegate these tasks to the pool operator.
• Proof of Stake delegation: In Proof of Stake, regular users can stake tokens without becoming a validator themselves. Users choose a validator and delegate their tokens, that is to say, the validator is given the right to stake the tokens. Having more tokens, the validator earns more rewards, takes a cut, and returns the surplus to the delegator.

By its nature, delegation causes these networks to become centralized. It creates and assigns roles to otherwise undifferentiated agents. If the network becomes overall more efficient as the number of delegates decreases, then we should expect all authority to be concentrated in the hands of the few. For example, variance of miner revenue is minimized when a single mining pool serves the whole market.

In this post, I will argue that centralization does not hurt freedom in blockchains as long as

• delegation is permissionless,
• doesn't have any barriers to entry or exit,
• and the services provided by delegates are reproducible.

That is because it is not centralization which hurts freedom, but the ability to monopolize/oligopolize authority. As long as above conditions are satisfied, a network can be centralized and have economic freedom at the same time. Centralization is thought to bring great evil, because it allows those in power to restrict freedom. However, blockchain is the one place we can prevent that, just because the technology allows us to do so. By setting the rules of the protocol appropriately, we can create a system where libertarian actors are rewarded and totalitarian actors are eliminated by creative destruction. In such a system, a delegate would be deterred from acting like a monopolist, because they can be easily replaced by another one providing identical services. This deterrence is the key to making centralized blockchains work.

Elaborating on the examples will make the point more clear.

Bitcoin mining pools

Since Bitcoin's inception, competition in mining has greatly increased the difficulty and the duration required to mine a block. Solo miners incurred costs continuously, but block rewards became too infrequent and unpredictable for a viable operation. Miners soon discovered that pooling their computational resources together decreased the variance in revenue and provided stability.

Today, mining pools are crucial for Bitcoin's operation. They are the reason how Bitcoin scaled to its current size. A lot of research was done from very early on to yield the fairest and most optimal distribution of rewards, such as Meni Rosenfeld's 2011 paper on different types of reward distribution.

Pools are not part of the Bitcoin protocol, that is to say, the rules of participating in a certain pool does not directly affect how Bitcoin works. A pool can exist as long as people are incentivized to join and stay.

To understand how mining pools relate to delegation, we need to explain which authority is assigned to whom. Most mining pools are operated by a centralized entity, called the pool operator, which is responsible for tasks such as operating member hardware, tracking contributions, and distributing block rewards fairly. In this case, miners delegate said tasks to the pool operator. It is in its best interest to perform these tasks as good as possible, because the pool operator realizes they are just an intermediary, and an expendable one at that. Members can tolerate only so many mishaps before deciding to leave, and join one of the many competing pools. Therefore, pool operators have a direct incentive not to misuse their granted authority.

If you already have heard that Bitcoin is a "decentralized" cryptocurrency, the above fact may pose a contradiction. You may know that the majority (>51%) of current hashrate comes from a handful of pools. Bitmain, a Chinese ASIC manufacturer, operates mining pools that amount to 50% of the total hashrate, and many are concerned that this may lead to a centralization of power that could bring Bitcoin's demise. But as I explained above, new mining pools can be created in a completely permissionless manner without any barriers to entry. This deters Bitmain from abusing its power.

In fact, Bitmain directly owns and controls only a small fraction of Bitcoin's total hashpower, despite operating a near-majority of it. Bitmain supports transparency regarding this issue, and discloses its hashrate every couple of months. Unfortunately, they don't keep history of previous announcements, so I gleaned the values from pages crawled by the Wayback Machine. Dividing them by the total hashrate gives us exactly how much of the total hashpower Bitmain operated in the last 2 years:

Date	Bitmain controlled hashrate (PH/s)	Total Bitcoin hashrate (PH/s)	Percentage of Bitmain controlled hashpower
July 22, 2018	1692	46852	3.61%
October 9, 2018	2339	48177	4.86%
March 5, 2019	1692	44973	3.76%
May 7, 2019	237	42312	0.56%
June 12, 2019	422	48578	0.87%
October 9, 2019	285	87533	0.33%
December 5, 2019	930	92816	1.00%

They apparently cut back considerably in 2019. While we should take this data with a grain of salt (because it's not possible to exactly verify their claims), we can expect Bitmain to stick to a certain level of professionalism. A misdeed that hurt the network would not stay hidden for too long, and it would be a matter of time before miners swarmed to an honest pool.

Proof of Stake Delegation

TBD