Leasing computing power to implement 51% attack, a problem in the liquid mining market that Satoshi Nakamoto could not foresee

This article is from Coindesk, original author: Anthony Xie

Translator | Moni

Editor | Lu Xiaoming

In order to ensure the decentralization of the network and avoid the emergence of "mining tyrants", cryptocurrencies using proof of work (PoW) generally do not allow a single mining company to control most of the computing power.

But as the global computing power pool continues to grow and become more liquid, cryptocurrencies may need to resist other potential threats, because attackers may rent global computing power for a specific mining algorithm. Once they do so for profit, they can concentrate a large amount of computing power in a short period of time and then carry out a 51% attack.

In the past, 51% attacks occurred in the cryptocurrency mining market, mainly due to the following reasons:

1. Miners based on specific algorithms: Many miners are optimized for a certain mining algorithm. If you switch to another mining algorithm, such as switching from SHA256 to X11, these miners will not work at all;

2. Illiquid mining market: Most global computing resources are illiquid and cannot be rented. Therefore, a large amount of upfront investment is required to build a large amount of computing power. It is obviously not a cost-effective "deal" to invest a large amount of money in the early stage in order to implement a 51% attack;

3. Opportunity cost: Cryptocurrency usually provides greater rewards to outstanding network participants to maximize the benefits of the entire network. In other words, whoever contributes more can get more incentives and greater returns. Any attack requires the risk of failure, including loss of mining rewards, reputation loss, and network damage. For a mining company that is preparing for long-term development, it certainly does not want the network to be attacked, which will affect market confidence and cause a drop in cryptocurrency prices to destroy future profit potential.

However, times are changing and the mining market is becoming more liquid.

Why can the liquidity mining market continue to develop?

The computer storage market has historically had illiquidity issues, but has now become a highly liquid online “commodity”. The same thing is happening in the hashrate market. There are two main factors driving the hashrate market to become more liquid:

1. The long-term rise in cryptocurrency prices will stimulate miners to increase their investment in computing power until the benefits they eventually receive are equal to the costs. In other words, if cryptocurrency prices continue to rise, global computing power will also increase.

2. Whether it is the buyer who rents the computing power or the buyer who rents the computing power, they can profit from the computing power rental transaction, so the percentage of rented computing power in the market has been growing. The separation of the buyer and seller markets has led to the emergence of more specialized mining machines, which also provides the market with higher mining operation efficiency. This is why mining machine hardware manufacturers are willing to sell mining machines directly instead of mining directly themselves. If the computing power renter focuses all his time on finding investment opportunities with the highest mining return rate, then they are likely to get the most value per unit of computing power. At the same time, the "lenders" who rent computing power can also reduce their business risks because they can cover more cryptocurrencies by renting computing power, that is, they don't "put all their eggs in one basket." Like "lenders" in other industries in the world, they only focus on the rental relationship and the utilization and maintenance of the leased assets.

Renting hashrate to conduct a 51% attack is now possible

The Crypto51 website calculates the cost of renting enough computing power to match the computing power of a given blockchain network for one hour. NiceHash is a world-renowned cryptocurrency mining computing power market where computing power can be bought and sold according to demand, but for some cryptocurrencies with larger market capitalizations, NiceHash still does not have enough computing power to support them, so these numbers can sometimes theoretically exceed 100%.

The hashrate data comes from Mine the Coin, the cryptocurrency price data comes from CoinMarketCapabilities, and the hashrate rental price data comes from NiceHash. There are two points that need special attention:

1. The attack costs cited by Crypto51 do not include the bonuses miners receive in the form of block rewards, so in many cases the true attack costs may be further reduced;

2. Crypto51 quoted the spot price of available computing power leasing on NiceHash, but in reality, due to the influence of supply and demand, the more demand there is for computing power leasing, the more expensive it will be.

Which cryptocurrencies are vulnerable to attacks by miners who rent computing power?

As shown in the figure above, the cryptocurrency ETP ranks 91st by market value on the CoinMarketCap website. If you want to attack this cryptocurrency, you can rent more than 21 times the computing power of the entire network, and the cost of the attack is only US$162 per hour (ETP's price data comes from Bitfinex, which provides ETP/BTC and ETP/USD trading pairs).

The amount of rentable computing power has tripled, and the top 40 cryptocurrencies by market value are vulnerable to 51% attacks

At this stage, a 51% attack on some cryptocurrencies seems out of reach because the total amount of computing power that can be rented on NiceHash is not enough to match the total computing power of the entire network. But let's imagine that if NiceHash triples their computing power rental scale, then cryptocurrencies such as Ethereum Classic (ETC, currently ranked 18th in market value on CoinMarketCap) and Bytecoin (BCN, currently ranked 40th in market value on CoinMarketCap) will be vulnerable to attacks.

With a fivefold increase in rentable computing power, the top 20 cryptocurrencies by market cap are vulnerable to 51% attacks

If NiceHash were to increase its rentable computing power fivefold, cryptocurrencies like DASH (currently ranked 15th by market cap on CoinMarketCap) and Bitcoin Gold (currently ranked 28th by market cap on CoinMarketCap) would be in danger.

So if these cryptocurrencies really suffer a 51% attack, how do the attackers make money from the attack?

While it is impossible for an attacker to create transactions for a wallet that does not have the private key, if you control a majority of the computing power of a particular cryptocurrency network, it means you can perform a "double spend" attack by reversing certain transactions on the ledger.

Double Spend Attack Mechanism

When miners find a new block, they should broadcast the block to all other miners in the network so that they can verify the block and add a new block to the blockchain. However, "corrupt" miners who control the computing power of the entire network can secretly create their own blockchain.

To perform a "double spend", the attacker spends their cryptocurrency on the real blockchain, but omits those transactions on their own secretly created blockchain. Once this miner has the ability to create the longest chain faster than all other miners on the network, they can then broadcast their own secretly created blockchain to the rest of the network.

Since the blockchain protocol follows the longest chain principle, the newly broadcasted longest chain will become the "real" blockchain, and the attacker's previous spending transaction history will be deleted. Please note that even if miners have the ability to control 51% of the network's computing power, it does not mean that they can always broadcast the longest chain, because other miners may also generate the longest chain. Therefore, in order to quickly create the longest chain, attackers usually want to control a larger proportion of the network's computing power, generally 80%.

Where will the hacked cryptocurrencies be spent? Exchanges are likely to be targeted

To profit from double spending, you need to find a way to spend cryptocurrencies that have actually been used but have not yet been recorded in the blockchain. If you can't spend these cryptocurrencies in the first place, the attack becomes meaningless.

The most likely place for an attacker to transfer cryptocurrencies is through exchanges, as they tend to be the largest buyers of various cryptocurrencies. In this case, the “routine” of a 51% attack would look something like this:

1. Choose a cryptocurrency network target that looks profitable.

2. Hoarding a large amount of cryptocurrency on the cryptocurrency network;

3. Rent NiceHash computing power and secretly develop a stealth chain;

4. Trade the cryptocurrency on an exchange and convert it into another cryptocurrency (such as Bitcoin);

5. Withdraw Bitcoin to another wallet;

6. Broadcast the invisible chain to the cryptocurrency network;

7. Obtain the target cryptocurrency;

8. Then switch to another cryptocurrency exchange and continue the above steps.

How will cryptocurrency exchanges respond?

As we can imagine, cryptocurrency exchanges don’t like being cheated. If this behavior by the attacker increases the operating costs of cryptocurrency exchanges, then they may respond with a series of measures to combat the 51% attacker, such as extending withdrawal and deposit periods, improving account verification security, etc.

Extending the withdrawal time will make the attacker's attack costly, because the time they "rule" the computing power of the entire network will also be extended accordingly, which means they need to invest more money to rent computing power. But at the same time, this method will also have a negative impact on legitimate traders and exchange users, and then arouse their anger. After all, no one wants to wait more time to withdraw their cryptocurrency.

Of course, cryptocurrency exchanges may have other ways to avoid some 51% attacks, such as carefully screening when listing cryptocurrencies and abandoning some cryptocurrencies that are vulnerable to attacks. However, this method also has problems, because the trading volume and revenue of some listed altcoins will gradually decrease, which means that these cryptocurrencies that are about to be delisted are also vulnerable to attack threats.

Generally speaking, cryptocurrency exchanges will basically use a combination of the above two methods. The easier it is for an exchange to escape a double-spending attack, the higher the cost for an attacker to implement a "double-spending" attack. In the long run, these two forces will gradually reach a balance in the same market.

How will cryptocurrencies respond?

For altcoins, there may be a solution to the threat of 51% attacks in the following ways:

1. Use a more "obscure" mining algorithm that is not used by many miners. However, this method can only be regarded as a short-term solution at best, because the fewer miners use your mining algorithm, the more difficult it will be to increase the computing power. If you want your network to grow and develop, the mining algorithm cannot be "obscure".

2. Put your project on a more secure and larger blockchain network, such as ERC-20.

3. Promote the use of more flexible new formula algorithms, such as Proof of Stake (PoS), which can also avoid 51% attacks. Of course, the Proof of Stake algorithm itself is not perfect and there are some challenges.

Satoshi Nakamoto did not expect the emergence of a liquid computing power market

How much will the computing power rental market grow in the future? There seems to be no clear answer to this question yet. 100 times? 200 times? Such a growth is not impossible. So how many cryptocurrencies will be affected?

Obviously, if your cryptocurrency has a relatively high market value and a low attack cost, it is vulnerable to a 51% attack. Does this mean that the market needs to eliminate some unsafe cryptocurrencies? Conversely, does this also mean that cryptocurrencies with large mining networks may continue to be overvalued in the future under the influence of 51% attacks?

There is a comment on the Hacker News website that may be a good ending for this article:

“Attacks via rented hashrate could be a very interesting example, and a market for rented hashrate has emerged that could directly disrupt a system. Satoshi foresaw that people might try to buy a large number of mining machines to carry out a 51% attack, so he went to great lengths to design a protocol designed to ensure that centralized mining is unlikely. However, Satoshi probably did not foresee the emergence of a liquid hashrate market, where renting hashrate can provide the hashrate for a 51% attack in a short period of time, which is 1,000 times easier than buying mining machines to carry out a 51% attack.”