Bitcoin Mining Decentralization via Cost Analysis

📝 Abstract

Bitcoin mining presents a significant economic incentive for efficient hashing and broadcast of data, both parameters stemming from the Proofs of Work used to advance the network. This incentive has led to the development of Bitcoin specific application specific integrated circuits and centralized mining pools, undermining the decentralized motivations behind Bitcoin’s design. In addition, the imminent block reward halving threatens the profitability of mining at any scale. Some work has been done in formal models for miner profitability, but existing models do not account for conditions such as the pricing of off-peak power and diverse investment strategies regarding sunken costs. There is also a lack of formal study of how the profit model changes as mining scales from the individual to the industrial level. Given the lack of analysis of these conditions, there are alternative models for profitable or net zero mining that operate at smaller, and therefore more desirable, scale.

💡 Analysis

Bitcoin mining presents a significant economic incentive for efficient hashing and broadcast of data, both parameters stemming from the Proofs of Work used to advance the network. This incentive has led to the development of Bitcoin specific application specific integrated circuits and centralized mining pools, undermining the decentralized motivations behind Bitcoin’s design. In addition, the imminent block reward halving threatens the profitability of mining at any scale. Some work has been done in formal models for miner profitability, but existing models do not account for conditions such as the pricing of off-peak power and diverse investment strategies regarding sunken costs. There is also a lack of formal study of how the profit model changes as mining scales from the individual to the industrial level. Given the lack of analysis of these conditions, there are alternative models for profitable or net zero mining that operate at smaller, and therefore more desirable, scale.

📄 Content

1

Bitcoin Mining Decentralization via Cost Analysis Jonathan Harvey-Buschel*, Can Kisagun+ *MIT EECS, +MIT Sloan

Abstract
Bitcoin mining presents a significant economic incentive for efficient hashing and broadcast of data, both parameters stemming from the Proofs-of-Work used to advance the network. This incentive has led to the development of Bitcoin-specific application-specific integrated circuits (ASICs) and centralized mining pools, undermining the decentralized motivations behind Bitcoin’s design. In addition, the imminent block reward halving threatens the profitability of mining at any scale. Some work has been done in formal models for miner profitability, but existing models do not account for conditions such as the pricing of off-peak power and diverse investment strategies regarding sunken costs. There is also a lack of formal study of how the profit model changes as mining scales from the individual to the industrial level. Given the lack of analysis of these conditions, there are alternative models for profitable or net-zero mining that operate at smaller, and therefore more desirable, scale.

1. Introduction

The intent of our work is to improve the health of the Bitcoin network. Some metrics for network health include number of miners on the network, number of operating pools, percentage of hashrate per pool, and total hashrate. Ideally, the network hashrate would comprise of many low-hashrate actors participating in a diverse set of pools with low individual percentages of the network hashrate. As a lower bound for decentralization, no one actor, miner or pool, should control 25% or more of the hashrate [1]. This distribution is dependent on the accessibility and profitability of mining. Bitcoin mining is founded on submitting Proof-of-Work with a difficulty that increases to keep the time to a solution roughly constant. Generating this proof is computationally intensive, and can be modelled as a conversion of electricity to heat [2]. In the Bitcoin case, the economic efficiency of proof generation is determined by electricity costs, hardware efficiency, and operational costs such as cooling. Framed in this way, it is clear that Bitcoin mining is most efficient where it is easiest to dissipate power or cheapest to consume it. These two properties are in conflict; heat dissipation is difficult with high densities of generation, but electricity pricing favors large loads. A third key factor is processing efficiency; as the hardware involved has moved past general- purpose processors to modern ASICs, the cost for miners to stay competitive has increased rapidly. This is caused by increasing engineering costs for new ASICs, as well as growing incentives for miners to keep ASICs they design to themselves as a competitive 2

advantage. The development of this hardware and its impact on the network is clear when observing network hashrate over time:

Figure 1. Network Hashrate over Time [3]

Figure 1 depicts the network hashrate over time. The x axis denotes time, and the y axis denotes network hashrate on a logarithmic scale. Each vertical line in this figure marks the release to the public of a significantly new development in Bitcoin mining hardware. The first line marks cgminer, which allowed enthusiasts to mine Bitcoin using their graphics cards (GPUs) [4]. Prior to this development, mining was limited to then central processing unit (CPU), a processor type common to all computers. GPUs are a popular computer component, but develop more rapidly than CPUs and use much more power. GPUs perform orders of magnitude better than CPUs at mining because they have many parallel cores designed for repeated mathematical operations on changing data, making CPU mining unprofitable. The second mark signals the development of Bitcoin- specific field-programmable gate arrays (FPGAs), which are chips that can be programmed to model any processor design. [5]. FPGAs and ASICs, which represent the last mark on Figure 1, provided further performance by implementing many parallel cores dedicated to only computing SHA256 hashes, the main operation involved in Bitcoin mining. FPGAs and ASICs are both expensive dedicated hardware that most participants in the network would not own outside of the purpose of Bitcoin mining. The resulting hashrate increase made GPU mining irrelevant, again raising the threshold for profitable participation in mining. The hashrate contributed by new hardware made GPU miners a much smaller percentage of the network, and consequently greatly reduced the reward for these miners. The corresponding change in number of miners is shown in Figure 2.

Figure 2. Number of Miners over Time [6]

In Figure 2, the x axis represents the network hashrate on a logarithmic scale, and the size of the circle matches the value on the x axis. The y axis represents the number of miners, which se

This content is AI-processed based on ArXiv data.