Is Bitcoin incentivizing its own centralization?

Thirteen years after Satoshi Nakamoto mined the first Bitcoin Block on January 3rd, 2009, we're faced with the increasing threat of mining centralization on the original blockchain.

The Whitepaper promised peer-to-peer electronic cash, yet the main chain does not lend itself to peer-to-peer payments in its current form. Advances are made with the implementation of lightning, yet, this won't fix the more significant issue of centralization on the base layer.

Before diving into different aspects of mining centralization, and the driving forces, a quick overview of Bitcoin Mining and its history.

How does Mining work?

"Mining" is the process of securing Proof-of-Work blockchains. As implied in the name, blockchains are a chain of blocks filled with transactions. To ensure the chain's integrity, parties on the network have to execute some work (therefore Proof of Work) to add new blocks.

Bitcoin miners compete to solve a cryptographic problem to add new blocks successfully. The fastest one to find the correct answer to the calculation - the so-called "Hash number" will receive a block reward.

One might assume there should be a way to create algorithms that would break the cryptographic riddle, but that's impossible. Bitcoin relies on SHA-256, a mathematical function that turns words and numbers into a string of 256 bits or 64 hexadecimal digits.

While the same input will always generate the same hash, one slight change 
will generate a completely different one. 

Hello World generates:
A591A6D40BF420404A011733CFB7B190D62C65BF0BCDA32B57B277D9AD9F146E

One small change in capitalization (Hello world) and the resulting hash looks entirely different: 
64EC88CA00B268E5BA1A35678A1B5316D212F4F366B2477232534A8AECA37F3C


Bitcoin blocks contain block numbers, previous hash, and a nonce value. The protocol requires the first digit of a hash to be zero. However, with the block number and the last hash as inputs, there is no way of ensuring that. This is why the nonce (a number between 1 and 1 trillion) was added.

Miners' task is to find the correct nonce so that the first x zeros of the hash are zero. Since there is no way to predict the hash, they run through all numbers until finding the right one. Mining difficulty is adjusted by changing how many zeros need to be at the beginning of the hash.

The bitcoin network aims to maintain a block confirmation time of 10 minutes and adjusts difficulty accordingly to how much hash power is on the network.

Hashpower = how many hashes the network can process per second = computing power that is dedicated to the network.

A short history of Bitcoin Mining

When Satoshi Nakamoto mined the first 50 Bitcoin, it was easily done using a Personal Computers' processing unit (CPU). In the early days, anyone could mine Bitcoin using the chip of their computers to solve the hash function.

Back then, Bitcoin was trading well below its current price, making it more of a fun activity for anyone interested in experimenting with this alternative financial currency. With costs for Mining relatively low, early miners had little to lose in the process.

With Bitcoin increasing in popularity, more and more computers joined the network, driving up the network's computing power. The more people joined, the harder it became to mine. Eventually, users started relying on Graphics Processing Units (GPUs) for Mining. GPUs equal several CPUs in parallel computing power and greatly improved mining efficiency. Yet again, more and more people switched to GPUs, further increasing the network's hashrate.

In 2011, Nangeng Zhang, who goes by the nickname Pumpkin Zhang launched the first attempt at specialized mining equipment - the Pumpkin miner. However, it consumed so much energy that it didn't survive its first year in the market.

In 2012 miners faced the first Bitcoin Halving - during which block rewards are cut in half.

Bitcoin Halving: Every 210,000 blocks mined the amount of Bitcoin that miners earn in block reward is slashed in half. This process is encoded in the protocol and happens automatically. It keeps inflation in check and reduces the new supply entering circulation.

The first ASIC (application-specific integrated circuit) miner was developed that same year. The butterfly miner had the computing power of 200 graphics cards without consuming an unreasonable amount of energy. Ever since the first ASIC entered the market, the companies have improved their efficiency and performance.

In tandem with the evolution of mining equipment, we've witnessed the rise of mining pools. The first pool, which continues operating, was slush pool in 2010.

As hashrate kept increasing, it became increasingly unprofitable for individuals to mine by themselves. So they started consolidating their computing power in pools, which enabled them to benefit from economies of scale and a bigger chance at mining blocks due to pools' higher share of the total hash rate.

On January 1st, 2022, the Bitcoin Hashrate hit a new all-time high at 208 EH/s. 4 mining pools together provide roughly 60% of that Hashrate.

Initially, most Hashrate came from unknown sources - individuals mining with their home equipment. Increasingly, pools started dominating.

At the time of writing, Crypto Twitter celebrates a solo miner who managed to find the right Hash with just 126 TH - beating the odds. Solo miners still exist, but it's quite apparent that mining pools provide the majority of the hashing power.

The question then is, how did a network set out to be a decentralized payments network become increasingly centralized?

We will draw on concepts from economics and value theory to answer that question.

Let's start with the potentially least obvious one.

Mining together is less risky

In the early days of Bitcoin, this wasn't the case. Mining with your computer involved minimal risk outside of draining your battery. It didn't require much effort to get started, nor did it come with a financial burden.

As the hashrate increased, the environment changed. First, anyone looking to mine had to buy GPUs, and then once those weren't sufficient anymore, any serious miner had to start investing in ASICs.

One of the most significant benefits for individuals - outside of the practical benefits of not getting involved with the tech - when joining mining pools is their risk-sharing mechanisms. The solo miner above might have won a block reward worth $260,000, but that isn't the norm. He beat the 1 in 10,000 chance of finding a block per day with that hashrate. And there is no telling how long this individual had been mining for.

In the worst case, individual miners can spend vast resources without ever validating a block - the most significant risk is not getting paid. In short, mining is risky. That's why miners have an incentive to minimize their risk.

Joining a mining pool immediately provides a more stable hashrate for miners in the pool and increases the likelihood of validating blocks tremendously. In "Decentralized Mining in Centralized Pools" Cong, He, and Li 2018 argue that a miner in a pool is twice as likely to receive payouts as a solo miner. Even though they end up sharing rewards, miners benefit from pools providing more stable cash flow.

The authors further analyzed the risk-adjusted payoffs for small miners joining a big pool and found that joining a pool can change their risk-adjusted payoff by more than 85% (p.9). That's a considerable improvement adding context to why solo miners would join pools - even if it meant that the network becomes less decentralized.

After all, miners are in it for the money, so the philosophy of decentralization might not make much difference to them. And all the calculations point towards them being better off when mining in a pool.

A different way to look at miners being off in a pool comes from value theory.

Are mining units in pools more valuable?

This is a question answered with yes by Nikos Leonardos, Stefano Leonardos, and Georgios Pilouras in their research paper"Oceanic Games: Centralization Risks and Incentives in Blockchain Mining" (2021).

To explain what they mean, we will start with the value theory of Oceanic Games. It's a theory developed in 1978 by John Willard Milnor and Lloyd Shapley for "voting games in which a sizable fraction of the total vote is controlled by a few major players and the rest is distributed among a continuous infinity of individually insignificant minor players."(p.1). Because the individual players lack coordination and cohesion, the authors decided to call them the Ocean, which gives the theory its name.

In the original paper on oceanic games, the authors analyze outcomes of a weighted majority game in which major players hold different amounts of the voting power.

If we look at the Hashrate distribution, it's clear that there is a set of "few major players", and an ocean of individuals.

Nikos Leonardos, Stefano Leonardos, and Georgios Pilouras set out to use the theory of Oceanic games to draw conclusions on the value of mining resources per miner and per unit of resource. Interestingly, their concept can also be used to look at PoS chains. The resource in question would then be native protocol tokens and not computing power.

The notion of value might imply this is about monetary value, and income does play a role. However, it's also about control. In a PoW blockchain, when one entity holds 51% of all the computing power, each resource unit is more valuable than if an entity just owned 49%. A slight percentage difference, deciding if an entity has control over the network or not.

In the current scenario where pools are already big players, they have significant resources that attract players from the ocean. When an individual player joins one of the mining pools, according to Oceanic Games theory, the value of his mining unit will increase. The increase in value aligns with the above discussion on the risk-adjusted payoffs from joining a pool.

You might wonder, what if initially in a network most of the resources were controlled by oceanic players - as was the case for Bitcoin? Unfortunately, the authors' findings aren't encouraging. As soon as a few oceanic players start working together and crystallizing out of the ocean, the value per unit is significantly higher than any single entity. On the other hand, oceanic players' value per unit of resource decreases the higher their network share. Yet another incentive for merging and forming coalitions.

As soon as a larger player (miner) accumulates 40% of the hash power, individuals have an incentive to join them, to continue having a chance. In most cases analyzed by the authors, individual miners are better off coordinating their actions and acting as a single entity.

The collaboration and consolidation among miners have created a situation for Bitcoin miners resembling oligopolistic markets in traditional economic theory. In an oligopolistic market, a few companies dominate, setting the general trend and prices. Because of that structure, it's hard to impossible for new firms to enter the market unless they have significant funding.

It's reasonable to say that starting a successful mining pool today will come at a high up-front cost.

While the previous theories were all concerned with how miners themselves benefit from joining a pool and what might drive them, we'll conclude by arguing that it was unavoidable.

Centralization is in the protocol

Bitcoin Mining is competitive. Proof-of-Work in Bitcoin means that an entity's rewards earned over time are proportional to their share of the overall computing power. It also includes that rewards are fixed. Regardless of hashpower, the maximum a miner can earn for every block is 6.25 BTC.

Humoud Alsabah and Agostino Capponi (2019) outline the consequences of these features. They came up with a game-theoretical model to analyze companies' behavior in the context of Bitcoin mining. For the model, they split the mining competition into two phases:

  • Companies invest in R&D
  • Companies compete for block rewards

Companies will compete to create more efficient mining equipment to increase their margin and win against the competition in the initial phase. While some might argue that technological spillover (other companies' getting hold of a competitor's research results) could at least mean less capital expenditure, it still results in everyone having better chips - consequently, the hashrate of the network increases.

Suppose each company decides to compete by itself. In that case, they won't capture the potential surplus secured through R&D expenditure because if everyone has great chips, all you end up doing is compete for the fixed block rewards throwing more and more computing power at it. Working with others is the only way to secure an edge in that scenario. Sounds familiar?

The authors conclude: "Our analysis shows that the two properties of the PoW protocol mentioned above are also sufficient to drive the Bitcoin mining industry towards centralization, against the fundamental principles behind the design of blockchain and cryptocurrencies." (p.24).

Where does that leave us?

When discussing mining centralization in the sense that hashing power is under the control of pools, some Bitcoin enthusiasts will counter that a network of decentralized individual miners holds up these pools. The argument is that these individuals could change the mining pool (ironically, they never say that the individual could just go back to mining alone) at any instance.

It sounds easy in theory, but in practice, solo miners will often purchase mining equipment from a pool provider and let them host it in the pool's facility. It's usually a lot more complex than just changing a few buttons, and as mentioned, individuals decrease their risk mining in pools, increase their earning potential. Bitcoin's reward structure still drives centralization even with an initial decentralized setup.

What will happen once block rewards stop entirely is anyone's guess.

Ergo Minima

At Minima, we've applied lessons learned from Bitcoin's history to ensure that everyone is equal on our network. Instead of relying on a set of miners to validate one's transaction, every user validates their own transactions by running a complete node on their phone.

80% of the world's population owns a phone, and Minima is small enough to fit on IoT devices. When everyone is in control, no one entity can take over. On Minima, users collaborate, they don't compete. And there is no monetary gain to be made from increasing one's computing power on Minima. All the factors driving Bitcoins' centralization are non-existent on our network.

Minima puts decentralization first, to achieve what Blockchain was intended to achieve: empowering freedom.