The Case Against ASIC Resistance
There has been a lot of discussion recently around ASIC mining, and how to prevent it. The communities of Monero, Sia, Ethereum, and others, have all had discussions about the coming “threat” of ASICs. However, I believe that the benefits of ASICs significantly outweigh their costs. In this post I will first briefly describe the history of mining and the types of hardware used in the process. I will then go into two types of potential mining attacks, and how their effectiveness depends on the hardware being used. Lastly, I will sew up my argument for the benefit of using ASICs to mine as opposed to GPUs.
History of Bitcoin Mining
On January 3rd, 2009, Satoshi Nakamoto became the first miner on the Bitcoin network. And for the first year of its existence, Bitcoin was mined using CPUs. CPUs are general purpose processors found in every computer and smartphone. However, towards the end of 2009 there were talks about the ability to utilize GPUs, which can perform mining calculations hundreds of times faster than a CPU. GPUs are larger processors that are commonly found in gaming computers.
There was a huge profit incentive for people to get their GPUs to start mining, especially if everyone else was still mining with CPUs. By 2013 GPUs were being used almost exclusively to mine, as CPUs were no longer profitable. However, in mid-2013 the first ASICs were introduced, which would quickly replace GPUs.
ASIC stands for Application-Specific Integrated Circuit, which means that the chip is specifically designed to perform one task. In the case of Bitcoin this task is performing the SHA-256 hashing algorithm as fast and as efficiently as possible. For this reason, ASICs can mine hundreds to thousands of times faster than GPUs. By the end of 2013 it was no longer economical to mine Bitcoin with GPUs.
Where GPUs can be used for other things such as gaming, mining other cryptocurrencies, or video rendering, ASICs can only do one task. Therefore, a Bitcoin ASIC miner is useless for anything except mining Bitcoin. This naturally makes their market significantly smaller, giving them a much bigger price tag than a GPU. A top of the line ASIC could run you $2000, whereas GPU’s fall in the range of $200–$800.
The Problem with ASICs
ASICs however, have generally not received a very warm welcome from the crypto community. The main argument against them has to do with their tendency to increase centralization.
Since GPUs are used for a variety of tasks, they are used by a relatively large number of people all over the world. This wide distribution of GPU ownership allows anyone in the world with a GPU to begin mining if they desire to, giving the network more decentralization of people who own the mining hardware.
With an ASIC, most people are not willing to drop thousands of dollars on a computer that can just be used to mine Bitcoin. Additionally, the ASIC you buy will eventually be significantly less effective within a few months of purchasing it as better ones come to market. This makes the resell value of ASICs virtually non-existent. ASICs high price and lack of alternative uses naturally makes them better suited for corporate mining as opposed to hobbyists. For this reason, the Bitcoin network becomes controlled by fewer and fewer actors.
This distinction between the corporate controlled ASICs vs. the rogue GPU mining crypto anarchist is a narrative that I believe to be at the heart of the ASIC-resistant mindset. However there this argument does have some merit, there are risks of having hashrate centralized in its control. As I will go into in the next section, if an entity is able to gain a majority of the hashpower on a PoW blockchain, they are essentially able to attack the network. One would think this is less likely to occur when your mining base is GPU hobbyists. But due to economies of scale for hardware and cheap electricity, GPU mining can be fairly centralized among mining companies too.
The idea of ASIC resistant cryptocurrencies has probably led to more Bitcoin forks than any other reason. Litecoin, Bitcoin Gold, and Zcash are all forks of the Bitcoin code base that, though not their primary alteration, changed the hashing algorithm in an attempt to be more resistant to the development of ASICs. And recently, Monero forked in order to stop the development and use of shadow ASICs on their network.
I will now take a counterpoint to talk about how ASICs actually provide Bitcoin more security than a cryptocurrency that relies solely on GPUs for mining — despite the potential increase of centralization.
When it comes to blockchain attacks, most are game theoretical. This means that we can expect rational actors to behave a certain way within the network. However, we also know that humans can be anything but rational, and potential motives can exist outside of what I will describe.
The most common blockchain attack is called the 51% Attack. A 51% Attack occurs when one miner (or colluding group of miners) are able to sustain more than half of the networks hash rate for a given length of time. While this attack would not allow the actor to change account values or old transactions, it would allow them to fix current blocks, leading to double spending or censorship of transactions. For this reason, a 51% Attack is only a real threat if it is maintained over a length of time.
In his paper titles Hostile Blockchain Takeovers (https://fc18.ifca.ai/bitcoin/papers/bitcoin18-final17.pdf), Joseph Bonneau outlines the four methods in which a nefarious actor could gain enough hashpower in an attempt to takeover the network. I will discuss 2 of these attacks: the rental and bribery attack.
A common theoretical attack vector on a Proof of Work blockchain is a rental attack. This is when an attacker temporarily gains access to non active mining hardware in order to act nefariously. However, nearly all available ASICs are already being used to mine the chain already, which is an important difference. This makes this attack significantly more feasible with GPU mineable blockchains.
For example, if you wished to perform a rental attack on the Bitcoin blockchain, it would be nearly impossible since nearly all SHA-256 ASICs are either already being used to mine bitcoin or are too old to make a significant difference. Alternatively, on an ASIC resistant blockchain like Monero or Zcash, a nefarious actor could rent GPU power from a variety of other sources: another GPU miner from a different chain, cloud computing power, etc. This could result in a potential 51% takeover of the network.
The primary takeaway from this is the idea that it is nearly impossible for a rental attack to occur on a ASIC mined blockchain because there are not any ASICs to rent.
Another possible attack vector on a PoW blockchain is a bribery attack. A bribery attack is when a nefarious actor bribes hashpower that already exists on the network, and gets it to work in their favor. At first, one would think that this would be an issue for both ASIC and ASIC resistant blockchains because a rational actor should accept any bribe that exceeds the amount that they are making by mining the chain.
However the decision of the rational actor becomes significantly harder to quantify than what I mentioned above. This has to do with the idea that if the bribe resulted in a widely public attack, it could significantly damage the value of the network (the price). And if your company has spent millions on ASICs that can only mine bitcoin, the amount that you would need to get in bribes would need to be significantly more than just your daily revenues. As an ASIC miner you care much more about the price of your coin because you have nowhere else to go.
The point here is that Bitcoin miners are less likely to be bribed because of their skin in the game, that is they want to protect their investment (quota met for Taleb reference in crypto post). In a blockchain mined with GPUs the miners have almost no downside risk if they improve their economic output through a bribe, because they are able to direct their hashpower to a large amount of different uses. If the Monero price tanks due to a large scale attack, Monero miners could simply point their GPUS to Zcash or Ethereum.
So while it’s true that ASICs can lead to more corporate controlled centralization, this is overcome by the skin in the game that is provided by having only one chain to mine. By having a dedicated mining base as opposed to miners who can switch at will provides significantly more security. While there are other reasons that I did not go over, this is primarily why I believe blockchain communities should embrace ASIC development on their networks, rather than resist it.
Disclaimer: The views expressed in this article are solely the author and do not represent the opinions of the author on whether to to buy, sell or hold shares of a particular cryptocurrency, cryptographic asset, stock or other investment vehicle. Individuals should understand the risks of trading and investing and consider consulting with a professional. Investors should conduct their own research independent of this article before purchasing any assets. Past performance is no guarantee of future price appreciation.