The 51% attack frenzy is coming. Is this the twilight of POW coins or the end of ASIC mining machines?

PoW (Proof of Work), as a precedent pioneered by Bitcoin, is widely regarded as a solution that combines security, decentralization and fairness in the consensus mechanism. The stable operation of the Bitcoin network for nearly ten years is the best example.

The only way to defeat the PoW mechanism is a 51% attack - when an attacker controls more than 51% of the computing power of the entire network, he can generate the longest chain by himself - that is, the attacker can force a hard fork at a certain moment, causing a double-spending attack. It is not difficult to see that this type of attack requires the attacker to pay a huge price, so it only remained on paper and in theory before May this year.

Since May, multiple cryptocurrencies have been hit by 51% attacks. Let’s review the overview of each of the attacked cryptocurrencies.

BTG was attacked on May 18, 2018. The official website issued an announcement on May 24, admitting that it was attacked by 51% and explaining the situation and improvement plan.

XVG was attacked on May 22, 2018. The official did not directly announce the attack. Only the official Twitter account posted on the 22nd that multiple mining pools were DDOSed, but there was no response to the attack afterwards.

ZEN was attacked on June 3, 2018. The official website issued an announcement almost immediately, listing in detail the TXID, amount, wallet address of the attacker and other specific information of the three double-spending transactions in the attack, and stated that the official will increase the number of block confirmations to increase the difficulty of the attack. I like this attitude of actively responding to the problem!

Why has Bitcoin, which has been running stably for nearly 10 years, not encountered 51% attacks, but has been frequently attacked in less than a month? Let’s talk about it from the perspective of the attacker’s “equipment upgrade”.

Bitmain has released two new Asic mining machines with different algorithms this year, namely the X3 using the Cryptonight algorithm released at the end of March and the Z9 mini using the Equihash algorithm released at the end of April.

In the Cryptonight algorithm, the representative currency is XMR Monero. After receiving the news of the release of the X3 mining machine, the Monero community immediately launched a hard fork plan to change the algorithm to ensure the safety of its own computing power ecosystem.

In summary, the Monero community launched a hard fork before the first batch of X3 mining machines were publicly sold, which caused the amazing profits of X3 mining XMR on paper to fail in the womb. Bitmain had to fork XMC for miners to make X3 useful. At that time, another currency favored by X3 buyers was ETN - the developer of this currency was a little late - did not change the algorithm immediately, but only remembered to adjust the algorithm at the end of May to prevent the situation of concentrated computing power. However, this also encountered a very embarrassing situation, please see the picture:

Take nanopool, which accounts for a large proportion of the computing power in the entire ETN network, as an example. When the block height of the fork was reached, the X3 mining machine could not be applied to the new algorithm and had to withdraw the computing power. The computing power of the mining pool dropped precipitously from 400K to 2-3K, which directly led to the inability to produce any new blocks due to the high difficulty (the difficulty of the CryptoNight algorithm is adjusted every 720 blocks) . The ETN community had to temporarily spend money to purchase cloud computing power from cloud computing service providers and adjust the computing power to its own currency. It survived the most difficult days. As the difficulty decreased and the mining income increased, it attracted some graphics card miners to join, so that the ETN network could continue to survive.

These are the stories that the emergence of X3 mining machines has brought to CryptoNight algorithm currencies. Although ETN is in a bad situation, it has survived after changing the algorithm. It is much better than the Equihash algorithm currencies (ZEC, ZEN, BTG, etc.) that have recently suffered a series of heavy losses.

Bitmain's newly released Z9mini is designed for Equihash-based currencies. It has a computing power of 10K sol/s at a design power consumption of 300W, which is approximately equal to the computing power of 33 GTX1060 graphics cards, which consume a total power of up to 4000W . In fact, everyone knows that in order to avoid "overdoing it" when releasing the mining machine this time, Bitmain has reduced the number of chips on a single mining machine, making the parameters of the mining machine look less scary, but its appearance is still fatal to currencies that are still in the GPU mining era.

Therefore, the currencies that use the Equihash algorithm made announcements one by one in early May. ZEC, ZEN, and BTG successively stated on their official websites that they would change their algorithms.

Zen explains the specific algorithm adjustment mechanism very clearly. The following content is excerpted from Zen’s official blog:

Possibility of adjusting ZenCash hashing algorithm

The hash algorithm used by ZenCash is Equihash, which is inherited from Zcash. The optimized hash function was created by senior cryptography practitioners. The original Equihash white paper describes how it was created and how it can be optimized and adjusted. It is worth reading.

Fortunately, the Equihash algorithm is flexible enough to be adjusted without being completely replaced, so changes can be made relatively quickly without requiring a lot of software development work.

The Equihash algorithm has parameters that affect the solution, called the N and K parameters. They affect how long it takes to solve and how big the solution will be for different combinations of processing power and memory. The parameters chosen by Zcash before launch were N = 200 and K = 9. After Zcash was launched, cryptographers determined that values ​​of 144 and 5 seemed to produce smaller, more ASIC-resistant solutions.

In this case, ASIC resistance means that more memory is required, the optimal amount of memory for parameters 200,9 is about 512MB, while parameters 144 and 5 require nearly 2.5GB of memory. Integrating memory into ASIC miners is expensive, so hashing algorithms with higher memory requirements are more ASIC resistant. We can call this version of the Equihash algorithm Equihash-144-5 to distinguish it.

ZenCash community members do not know if the Bitmain Z9 Antminer can be modified to mine Equihash with adjusted N and K parameters, as the actual architecture of the device has not been disclosed. It may be a dedicated ASIC with limited memory, or it may be a reprogrammable cryptographically optimized computing device with external memory. Depending on the architecture, changing the N and K parameters of Equihash may prevent Bitmain from producing Equihash-144-5 dedicated miners for some time.

Unfortunately, perhaps some stakeholders were resentful of the fact that there was no money to mine before the new mining machines were put into operation, and 51% attacks occurred one after another before these currencies hard forked . Since Zen described the attack incident very clearly, we will analyze some of the data.

At the time of the attack the Zen network hash rate was 58MSol/s. It is possible that the attacker has a private mining operation large enough to conduct the attack and/or supplement with rental hash power.

When Zen was attacked, the total network computing power was 58MSol. It was possible for the attacker to gain more than 51% of the computing power through private pools and temporary rental computing power to launch an attack.

Before the emergence of Asic mining machines, this situation was basically impossible to happen, because the threshold for a 51% attack is a computing power of more than 29MSol/s. Based on the computing power of 300Sol/s per GTX1060 card, it would require nearly 100,000 graphics cards, or about 15,000 professional mining machines. The cost and expense of using these mining machines to launch an attack would be much higher than the final profit from a successful double spend, and no one would do something that is thankless.

With the Asic mining machine, only 2900 Z9 mini mining machines are needed to meet the requirement of 29M computing power (Z9mini is used as an example here, and attackers may use other models of Asic mining machines to launch attacks) . The site requirements and power supply requirements for deploying 2900 Z9 minis are much lower than the previous 15,000 GPU mining machines. Therefore, attacks are possible.

The following can answer our previous question: Why have 51% attacks, which have almost never happened in the history of PoW in the past 10 years, appeared frequently recently?

This is because the emergence of ASIC chips for Bitcoin and Litecoin has a relatively slow evolution process, and at the time there were only these two currencies that could be mined. The computing power of the entire network has always been very high, and the cost of launching a 51% attack has always been very high.

The situation is different now. There are already many currencies that can be mined. The majority of the network's computing power is still concentrated on a few leading currencies, while some unpopular currencies do not attract many miners. The ZEN and BTG that were attacked this time are second-tier currencies that use the Equihash algorithm. Their total network computing power is much lower than that of ZEC, but they are still listed on major platforms such as Binance and have good liquidity. They are indeed good targets for malicious attackers. That's why we say that the computing power ecology of small currencies cannot withstand the tossing of Asic mining machines. To put it in an analogy, everyone is still mining with shovels, and suddenly a group of people driving excavators come. How do you compete with them?

In the future, it is foreseeable that the communities of various currencies will pay more attention to the emergence of ASIC chips and the prevention of the influx of ASIC mining machine computing power. It is possible that after an ASIC chip manufacturer releases a chip with a specific algorithm, the currency using this algorithm will announce a change in the algorithm to abolish the ability of ASIC mining machines. This is definitely a situation that ASIC mining machine manufacturers and miners who purchase mining machines do not want to see.

Just when I was preparing to post this article, I saw the announcement of the price adjustment of Z9 mini. Indeed, the situation that the new mining machine was launched at half the price, which was hard for us to imagine before, has now appeared. However, facing the situation that "it will become scrap metal after purchase", we do not think that this form of promotion will attract many miners to buy.

PoW will definitely not disappear in the future, but the development path of Asic mining machines that cause the centralization of PoW seems very confusing.