UPDATED | March 26, 2018:
Following the recent news that the cryptocurrency mining hardware firm Bitmain plans to sell a Monero-specific ASIC called Antminer X3, Monero core developer Riccardo Spagni published plans for a hard fork that would undermine those ASICs' efficacy. The update to the Monero network is scheduled for April 6, more than a full month before the first of the mining rigs are expected to ship out.
When the Antminer was unveiled, Spagni took to social media to proclaim that the hardware would "NOT work on Monero." Before the Bitmain ASIC sold out, customers had the option to preorder it for $11,999, according to the company's website.
ORIGINAL | February 16, 2018:
A recent post credited to "dEBRYUNE [sic], dnaleor and the Monero project" relates that a group of individuals, ostensibly those referred to collectively in the communiqué as "the Monero team," have taken a position on the possibility of Application-Specific Integrated Circuit (ASIC) miners making their way into the Monero ecosystem.
Due to concerns over centralization, the team is against the development of ASICs for Monero mining and proposes "slightly modifying" the blockchain's PoW hash algorithm during scheduled semi-annual forks, as well as executing "an emergency hard fork to curb any potential threat from ASICs [whenever one is] needed."
Before going any further, let's break down what this means.
In order to mine blocks on a PoW blockchain, miners use computing power to solve a puzzle, known as a hash, which is generated by that blockchain's hash algorithm. The hashes that these algorithms produce are based on the transactions contained in the current block. Different blockchains use different hash algorithms, and while a single algorithm will always generate the same hash if it's fed the same data, different algorithms will produce different outputs based on that same data. Particular mining equipment can be better or worse-suited to solving a particular hash algorithm's puzzles.
In the words of the team:
"An ASIC … is custom mining hardware that is specialized for a particular use. For example, Bitcoin ASICs are specialized to mine cryptocurrencies utilizing a SHA256 hash algorithm, [which is used on the Bitcoin network]. Due to the fact that they are designed for a specific use, they will typically have a large performance advantage over hardware that is intended for general-purpose use like CPUs, GPUs, and even FPGAs [which are circuits that can be configured for specialized purposes after they are manufactured]."
In other words, an ASIC that is tailor-made for one hash algorithm will not perform nearly as well in solving the hashes of a different algorithm, but one designed around Monero's hash algorithm "will obtain a significant majority of the network hashrate and introduce centralization." This happened to the Bitcoin network, on which ASICs are now "the only viable way to mine."
The post's authors worry that only a few firms will be able to build such ASICs in a cost-effective way, and with this centralization will come the threat of a "single point of failure. For instance, a government could require these ASIC manufacturers to add a 'kill-switch' which allows them to shut down a miner remotely or otherwise control it."
They concede that cryptocurrency mining is generally "prone to the rich-get-richer effect," because the more capital miners have at their disposal, the more mining equipment they're able to purchase and operate. They argue that while this tendency "ultimately leads to centralization," Monero's hash algorithm, known as CryptoNight, aims to "close the gap between CPU and GPU, FPGA, and ASIC mining by using a proof-of-work system that is memory bound over a moderate amount of memory." This feature slows centralization.
Continually modifying the hash algorithm is a way of kicking the can down the road. It may not stave off centralization forever, but it will throw wrench after wrench into efforts to develop CryptoNight-specific ASICs.
One possible obstacle to the smooth functioning of this scheme: users failing to mine the desired branch of the hard fork. If some users move their operations over to the newly created branch while others stay on the old one, the division of the mining community poses its own threat of centralization.
Riccardo Spagni, who has described himself as the "lead maintainer for Monero," told ETHNews that, as is the case "with every hard fork, as developers we present the software to the userbase, and then we hope that they'll run it. We have no control over what miners and economic nodes will do."
The first routine fork containing a change to CryptoNight will be executed in March, and the Monero team will "continue to research alternative proof-of-work functions that may provide better ASIC resistance than CryptoNight."