After halving, is the Bitcoin network still safe?

Core points

  • Block rewards are currently the main source of income for miners. Halving the reward will cause some miners to quit the network. The sudden reduction in the number of miners in the short term may make the network more vulnerable to security threats such as 51% attacks.
  • The distribution of computing power between different types of mining machines will also have an impact on the security of the network. Old mining machines that can still be purchased in the secondary market (ie, previous generations of mining machines) pose a potential threat to the network.
  • The new technology involving nonce distribution allows us to numerically estimate the computing power provided by certain types of mining machines, including S7 and S9 old mining machines.
  • It now appears that a large number of old S9 miners have been reopened. At present, the mining machines of the Antminer S9 series account for about 32% of the BTC computing power.
  • Although the computing power generated by the old mining machine S9 is far less than 51%, the changes brought by their existence to the network security dynamics are huge.

BTC halved

On May 11, BTC ushered in the third halving in history. This event occurs every 210,000 blocks, or about every 4 years, until the circulation eventually approaches zero.

In this halving, the block reward dropped from 12.5BTC to 6.25BTC. The market experienced significant volatility before the halving, and the market volatility eased after the halving. In general, the impact of this incident on network security is relatively small.

Block rewards are currently the main source of income for miners, so the reduction in rewards will cause some miners to withdraw from the network . In the case of reduced revenue, profit margins are tightened, and less efficient miners may suddenly find that their operations have suffered losses. In the long run, these low-efficiency miners will usually be replaced by high-efficiency miners as the market tends to rebalance. But in the short term, the miners will suddenly reduce network more vulnerable to security threats, 51% of attacks .

This article will focus on the safety impact of halving and the economic impact of running old mining machines. We will use Nonce data to estimate the current share of certain types of mining machines on the network, and discuss how the existence of old mining machines affects the security of BTC.

Halve vs safety

BTC miners obtain income through block rewards and transaction fees. The former is directly affected by the block reward halving, while the latter is not affected . The transaction fee is generally a function of the space requirement of the block, so in the period of block congestion and heavy traffic, the transaction fee tends to surge.

At present, transaction costs account for a small percentage of the total income of miners. In the past five years, only about 4.4% of miners’ income has come from transaction fees.


Source: Coin Metrics Network Data Pro

As BTC block rewards are halved approximately every four years, this will increase transaction fees to incentivize miners to ensure safety on the chain. In the short-term before the halving, transaction costs have soared (approximately 17% of miners’ total revenue), which has increased to levels never seen in nearly a year.

The increase in transaction costs may be magnified by the decrease in computing power after the halving, which is caused by the inefficient miners leaving the network. The decrease in computing power increases the time required to dig out blocks in the short term (the difficulty adjustment is regular), thereby reducing the available block space.

In order to reduce the uncertainty of income, miners usually choose to access the mining pool. The mining pool is a “loose alliance” organized by a mining pool operator. In the mining pool, the miners usually distribute income according to the contribution of computing power. Miners often switch mining pools due to certain factors, the most concerned of which is the fees charged by operators.

As long as no dishonest entity controls more than half of BTC’s computing power, the network is safe . During the 51% attack, a node that controls more than half of the computing power will use the computing power to change the block transaction to perform double-spends operations.

The minimum number of mining pools required for a 51% attack is called the Nakamoto coefficient. At this time, the top 4 mining pools on the network need to collude with each other to reach 51% of the computing power. In the history of BTC, this number has generally risen, indicating that the degree of decentralization is steadily increasing.


Source: Coin Metrics Network Data Pro

The Nakamoto coefficient is not a perfect metric, it will overestimate the degree of centralization of BTC . Because individual mining unions face a large amount of upfront expenditures such as mining machines and other capital, this also makes them lose the enthusiasm to attack the network, that is, these miners may still choose to leave the maliciously operated mining pool in the end.

However, the mining pool will choose the blocks that miners in the organization will dig and limit their departure to a certain extent to achieve relative control. Moreover, an attacker may use a technology similar to feather-forking to use less than half of the network’s computing power to change historical transactions. Therefore, it is very useful to have a metric such as Nakamoto Coefficient to quantify the degree of centralization among miners.

Stratum V2 is an implementation of Betterhash (but with modifications to the original protocol). It recommends that individual miners choose the blocks they want to mine instead of the pool operators. This potential improvement to the operation of the mining pool is to put more power in the hands of individual miners and further decentralize the network.

In addition to the distribution of computing power between different entities, the distribution of computing power of different types of mining machines has a great impact on the security of the network.

Strengthen the mining machine (Hardening Hardware)

BTC miners will try to find a nonce value (or arbitrary value) to add a new block to the existing blockchain, the nonce value will make the hash value of the block header lower than the target value, and these hash values The speed of calculation and verification is called computing power, and the nonce that meets this condition is called golden nonce. Theoretically, the golden nonces are evenly distributed in the space of potential nonces and valid blocks. The threshold of the hash value of the block header must be met, which is set by the difficulty parameter of the network. This parameter is based on the block in the chain. The accepted speed is adjusted regularly.

Although mining was initially done by the CPU, the GPU was parallelized at an early stage, and mining became more efficient. Today, mining is almost always done using mining equipment that contains specialized chips, which are called ASICs. Compared with other mining machines, the speed of these devices is significantly faster, the parallelization capability is stronger, and the energy is saved.

The existence of old mining miners has changed the current security model of BTC , because although the subsequent operating costs are large, the initial investment required is small. Even in addition to the cost of mining machines, there are still various obstacles (actual or other logistical costs) to open a mine composed of old mining machines, but the existence of these mining machines also greatly reduces the fixed cost of entering the market.

Due to the anonymity of the mining industry, it is often difficult to distinguish what type of mining machine is used to ensure network security. However, the latest technology allows us to numerically estimate the computing power provided by certain types of mining machines.

We can calculate the type of mining machine used for mining on the network from the nonce distribution of BTC. By combining these data with the price information of mining machines on the secondary market, we can also quantify the degree of risk posed by these cheap and slightly outdated mining machines.

Since the golden nonces are evenly distributed in the nonce space of all potential blocks, we previously expected that the golden nonces will be evenly distributed over time, but in fact their nonce distribution is not like this.


Source: Coin Metrics Network Data Pro

Near the left side of the figure above, nonce is concentrated in the lower range. This is the result of the sampling technique used by miners in the CPU mining era, which is to increase the number of iterations from zero to test the value.

In 2015, the nonce distribution of BTC for the first time showed stripes, and this pattern has only recently begun to fade. The stripe shape was wider at first, then suddenly narrowed, and then gradually subsided. In total, we can see four distinct stripes, and the narrow band and wide band of each stripe can be confirmed in the figure.

The appearance of these stripes stems from the sampling technology of the gold nonce by the Bitmain Antminer S7 and S9 miners. These miners were once the dominant miners on the network, and the dominant position of S9 was only recently replaced by Antminer S17.

Broadband and narrowband are caused by the sampling technology used by the S7 and S9 series mining machines, respectively. We can use the information obtained and the principle behind the technology to numerically estimate the computing power provided by S7 and S9 in the network proportion.


Source: Coin Metrics Network Data Pro

We can see from the figure above that the percentage of computing power provided by S7s and related miners reached its peak in May 2016 (about 61%). Today, S7s accounts for a small percentage of the computing power of the entire network. The proportion of blocks mined by S9s and related miners peaked in May 2018 (about 78%). And now about 32% of the blocks are generated by S9s.

This estimation method is based on the assumption that both S7s and S9s are not within the exclusion zone for nonce value determination, and that all mining machines except for the exclusion zone have the same algorithm for determining the nonce value. These assumptions are not true in the era of CPU mining, and it seems to have been there since the beginning of GPU mining. The exclusion zone is determined manually, and the estimated value outside the range of 0 ~ 100% is corrected and standardized.

The estimated value will be affected by a certain amount of noise (we can see this on the left side of the figure). The small bulge of the estimated proportion of S9 computing power in 2015 may be caused by noise, or it may be a signal of other things, such as the test of an experimental mining machine.

The proportion of computing power provided by each type of mining machine further confirms the current view that old mine opportunities threaten the security of BTC.


Source: Coin Metrics Network Data Pro

The most notable in the above figure is that the computing power to maintain network security has increased exponentially

The estimated computing power provided by S9s reached its peak in August 2019, when the computing power was 52EH / s. In February 2020, the computing power of S9s reached the bottom, about 21EH / s.

It now appears that a considerable number of previously offline S9s have been reopened, which is most likely the result of the recent appreciation of BTC prices. Now the computing power of this miner is 37EH / s.

Due to the rapid changes in the market environment, the computing power distribution of various miners at this time will not continue in accordance with the status quo. However, our analysis illustrates the feasibility of using old mining machines for mining under favorable conditions and the ease of deploying such cheap mining machines.

S9 mining machines are now sold at a very low price in the secondary market. The second-hand prices of these miners range from US $ 20 to US $ 80, while their original price is about US $ 3,000. Considering today’s economic environment and cheap electricity brought about by the rainy season in China, miners found that the operation of these devices is likely to make them profitable.

Although the computing power generated by S9, which was offline (now put into use) before, is far from enough to launch a 51% attack, the changes that their presence brings to network security dynamics are huge. Other platforms that use the BTC mining algorithm may feel this effect more strongly, including BCH and BSV. The current security computing power of these two platforms is about 2.5EH / s and 1.8EH / s, respectively.

BTC opens a new era

Under the influence of the expectation of halving and the increasing interest in investing in BTC with institutions, the price of BTC has risen sharply. Since the halving, BTC market volatility has eased, but prices still continue to rise. The halving also drove the increase in transaction fees and caused a decline in computing power.

The sentiment related to halving will continue to affect the market, and the halving event itself will continue to test whether BTC can successfully transition to it, so that the income of miners depends mainly on the stage of transaction fees. The long-term impact of this incident remains to be seen, but its impact on the mining machine economy and the entire market is already obvious.

Network data observation

Summary of various indicators


Source: Coin Metrics Network Data Pro

The transaction fees of BTC and ETH continue to climb, but most other indicators are relatively flat compared to last week.

The median transaction fee for BTC reached US $ 2.88 on May 14, the highest level since June 2019. At the same time, the median transaction fee of Ethereum reached US $ 0.25 on May 14, the highest level since August 2018. The median transaction fee tends to surge when the block capacity is crowded. The reason for this surge will be discussed below.

Network highlights

The computing power of BTC has fallen to 81.66 TH / s after halving, which is about 40% lower than the high point before halving. As this week ’s weekly special article pointed out, as less efficient miners withdraw from the network, a decline in computing power can be expected. It may recover after a period of shuffling, during which time efficient miners replace inefficient miners. However, it is unclear how long this turnaround period will last.


Source: Coin Metrics Network Data Pro

The decline in computing power has caused the average time interval of BTC blocks to rise to the highest level since the end of 2018 (excluding the block interval time around March 12, 2020. Caused by the drop and subsequent drop in hash rate).

As the speed of block generation becomes slower, the competition in block space is becoming more and more fierce, which leads to an increase in transaction fees. Paying a higher transaction fee results in a higher probability that miners will include the transaction in the block. Therefore, in the period when the block space is at a premium, the median transaction fee tends to surge.


Source: Coin Metrics Network Data Pro

The decrease in the total number of blocks also causes the size of each block to increase. The average block size of BTC reached a record high of 1.32MB on May 17.


Since the halving of BTC, the median transaction fee of Ethereum has also shown signs of growth. However, the median transaction fees of Bitcoin Cash (BCH), Bitcoin SV (BSV), Ripple (XRP) and Litecoin (LTC) did not increase significantly.


source: Coin Metrics Network Data Pro

Market data observation

BTC’s performance has surpassed other Crypto currency assets for three consecutive weeks, and has formed a trend that cannot be ignored. Although it cannot be ruled out that this trend is just the result of random price walks, a seemingly reasonable explanation is that the stored value attribute of BTC is becoming more and more important in today’s market environment. And as we observe that institutional investors are increasingly including BTC in their investment portfolios, BTC has naturally become the first choice for institutional investors looking for Crypto currency investment, and eventually led to Crypto currency assets as an independent asset class. use.


Source: Coin Metrics Network Data Pro

An underlying call option for inflation

Now there is a new argument: the global central bank and the government have adopted unparalleled monetary and fiscal policies in a market environment where we need the existence of BTC. To study this phenomenon, we show the year-on-year change in the Fed ’s balance sheet, highlighting the speed and magnitude of the Fed ’s response to COVID-19. The Fed’s policy response has exceeded the balance sheet expansion in the first three quantitative easing plans after the financial crisis.


Similarly, whether it is based on survey-based inflation expectations, such as the University of Michigan consumer survey, or market-based inflation expectations, such as 5-year and 5-year forward inflation derived from TIPS It is expected that people will continue to pay attention.

In the context of rising inflation, how can we reconcile such concerns: on the one hand, BTC is an anti-inflation asset during the period of rising inflation; on the other hand, inflation does not exist in the short term, and the long-term inflation expectations are very low? According to the Fed ’s Consumer Expectations Survey, the median inflation expectations have not changed significantly due to COVID-19, but the uncertainty of respondents ’inflation expectations has risen unprecedentedly.


To explain this phenomenon, we can think of BTC as a call option with inflation as the target, and study its Greek indicator: the sensitivity of option prices to inflation. Even if the median expectations for future inflation levels remain unchanged, the potential volatility of future inflation is increasing.

Standard option price theory shows that an increase in the underlying implied volatility should lead to an increase in the price of a call option. Therefore, we can attribute the recent rise in BTC prices to “expected inflation volatility.”

CMBI observation

Both the CMBI and Bletchley indexes performed very well this week, closing at 5% to 15% higher than the previous week’s closing price. The biggest news this week was that after BTC halved, the CMBI BTC index performed the strongest, rising 14.4%. The CMBI Ethereum Index also performed strongly this week, closing 11.9% higher. Despite these two strong performances, the best performing market capitalization weighted index is the small-cap Bletchley 40 index, which closed up 12.1% this week.


Source: Coin Metrics CMBI

Coin Metrics Author

Sub-Ming translation

Sub-ming editor

The content is for reference only and not at your own risk as investment advice

All rights reserved without permission

Leave a Reply
Related Posts

Reasons why Bitcoin cannot break through $ 10,000

Since Bitcoin fell below $ 20,000 in 2017, the five-digit figure has been a huge resistance level for Bitcoin. Breaking through this price point is temporary, and the price usually falls without holding for long. Experts believe that the integration of Bitcoin…
Read More