In light of reports that the Chinese tech firm Bitmain "has already developed" an Application-Specific Integrated Circuit (ASIC) designed for the purpose of mining Ether, members of the Ethereum community have begun discussing the possibility of a hard fork to alter the network's hash algorithm, potentially rendering the device useless.
The hash algorithm is a feature of blockchains with proof-of-work (PoW) consensus mechanisms. To mine a block of a PoW blockchain, a miner must perform work, often described as solving a puzzle, to ascertain a string of characters that the algorithm outputs based on transactions in the current block.
Well-designed ASICs can generally arrive at this value much more quickly than other types of processors and, as such, their introduction can push other processors being used for mining toward obsolescence.
An influx of ASIC miners also threatens to cause miner centralization, which, among other things, can lay the groundwork for a so-called 51 percent attack, through which malicious actors can double-spend their tokens.
Following a similar revelation about an ASIC tailored to the Monero network's CryptoNight hash algorithm, a hard fork of that blockchain has been scheduled.
On March 29, Ethereum developer Piper Merriam opened Ethereum Improvement Proposal (EIP) #958, which seeks the community's input on whether the blockchain should fork and how it might "implement improved ASIC resistance." In the proposal, Merriam also alludes to speculation that Bitmain has already begun mining with its Ethereum ASICs.
A number of influential members of the Ethereum community have weighed in on the topic, both on EIP 958's Github page and on Twitter.
"Nobody was promised a stable PoW algorithm for Ethereum, and the present one was deliberately designed to be hostile to ASICs. ASICs result in more centralisation of mining power; forking them out seems like a good idea to me."
On the EIP's page, however, he added a layer of nuance to this appraisal, saying that he considers it impractical to "design a new asic-resistant algorithm until we have a better idea how the current ASICs (if they exist) function."
Furthermore, he argued, the task at hand is far from trivial and should, therefore, be approached with care:
"Each PoW algorithm has to be unique – you can't just change a key, like you would in crypto – and well vetted for issues and vulnerabilities. Then, multiple implementations are needed, and mining pools etc need to be given time to update their implementations for the new algorithm in time for the fork. Then, for each fork the community needs to be informed, and the fork needs scheduling with enough time for everyone to upgrade in a timely fashion."
A few hours after opening EIP 958, Merriam clarified his goal, writing that the community "can't make [Ethereum's hash algorithm] ASIC proof, but we should be capable of at least rendering anything currently in production either obsolete, or at least requiring more R&D."
Several of the individuals commenting on the matter noted that a proposed shift to a proof-of-stake (PoS) consensus mechanism would do away with the element of competition in mining, making (generally expensive) ASICs a poor investment. Ethereum developer Lefteris Karapetsas questioned the logic behind assigning "manpower/resources on such research if PoS is going to render this obsolete anyhow."
Ethereum developer Philip Daian also speculated that the eventual adoption of PoS would "kill longer term ambitions of any ASIC purchasers" and proposed doing "absolutely nothing." He suggested that the goal of avoiding "industrialization" could be better achieved by implementing "incentives making it not worth it," such as reducing the rewards paid out to miners.
Without explicitly stating his support for or opposition to a hard fork, core developer Vlad Zamfir said that he has "no issue with trading off miner interests for user interests. [In my opinion,] the protocol is for users, and ought to be controlled by users for their benefit." Elsewhere in the same Twitter thread, he wrote, perhaps only half-jokingly, "I would say that disenfranchising miners of their role as participants in governance is a good thing :P"
A day after introducing the EIP, Merriam summarized the responses that the proposal had received up to that point:
"This feels like a choice between 'wait until we know' or 'shot in the dark' if we work with the assumption that we will change algorithms to deal with ASIC mining.
-'wait till we know' exposes us to both the risk of changing algorithms + the risk of whatever mining centralization happens between now and then.
-'shot in the dark' exposes us to the risk of changing algorithms + the risk of maybe having to change algorithms again later."
The question, he said, then becomes one of whether the "risk of changing algorithms [is] greater than the centralization risks" inherent in doing nothing.