A Fight Over Specialized Chips Threatens an Ethereum Split

In July 2016, Ethereum endured an early test of faith. The people behind the barely year-old blockchain had taken Bitcoin’s idea of decentralized money and run with it, building a digital landscape where users, based on a mutual trust in code, could interact and create applications. Then hackers emptied $50 million from one of those applications, the DAO. Facing a crisis, a core group of developers swiftly altered Ethereum’s code to return the lost funds.

The heist raised a key question for Ethereum and for dozens of other platforms that purport to be “decentralized:” Who decides when the code needs changing? Even at the time of the DAO hack, when Ethereum was a small community, the move sparked dissent. Some people kept using the old code, splitting Ethereum in two. And since then, even as Ethereum has grown into a multibillion-dollar operation with a vast constellation of businesses, developers, and users in its orbit, its process for making hard decisions remains ill-defined, a warren of Twitter polls, developer votes, and in-person conversations. Now a controversial new proposal—involving not just code, but politics—is testing what that means for Ethereum’s future.

The effort, led by developer Kristy-Leigh Minehan, centers on the kinds of machines people use to mine cryptocurrency. Like Bitcoin, Ethereum is kept secure by a system called proof-of-work, which involves computers racing to solve complex math problems. The solver reaps the rewards, hence the term mining. A decade ago, when Bitcoin was just getting started, a miner could turn a profit using chips found on their home desktop—first with standard computer processors, and later with the GPUs that power graphics cards. But the rising value of Bitcoin set off a race for more powerful hardware. Along came another type of chip, ASICs—which, like a finer, sharper chisel, are specially designed to handle the computations used to mine a particular coin.

In 2014, as he developed Ethereum, Vitalik Buterin watched the rise of these specialized chips with trepidation. In the Ethereum white paper, he observed that Bitcoin had become dominated by large and well-financed mining pools using ASICs and described ways in which Ethereum would curb their advantage. But that changed last summer, when Chinese chip maker Bitmain announced it could produce an ASIC optimized for Ethereum’s unique math. In September, another Chinese company, Linzhi, said it would follow suit.

Last spring, even before Bitmain’s new chips were available, Minehan and two other developers, under the pseudonym IfDefElse, began working on a new mining algorithm to slow them down. They called it ProgPoW. To Minehan, ASICs present a threat to Ethereum’s decentralized ethos. They tend to be more expensive than GPUs and require more power than a standard home outlet. And because they’re made by a short list of private companies, primarily based in China, access can be more limited; critics also point to cozy relationships between manufacturers and big mining farms. The home Ethereum miner, Minehan argues, would become obsolete.

Not everyone agrees. The most vocal opposition has come from Chinese ASIC makers. Bitmain cofounder Jihan Wu recently turned the tables, arguing that ProgPoW was a play at centralizing Ethereum under two public American GPU makers, AMD and Nvidia. A spokesperson for Linzhi, in an email to WIRED, called it a move to “counter Chinese supremacy in crypto.” But many critiques have centered on Minehan herself, focusing on her experience working with GPU makers and alleging that she had been paid to develop the algorithm. Minehan denies the collusion. (Nvidia CEO Jensen Huang, facing a “crypto hangover” as GPU sales slumped along with the crypto prices, recently said revenue from the industry would be “negligible” moving forward.)


The WIRED Guide to Bitcoin

Some developers are also skeptical of ProgPoW, for technical reasons. Long-term, the type of mining chip doesn’t matter much to centralization, says Phil Daian, a researcher at Cornell Tech who studies Ethereum. Mining with GPUs and ASICs will both “end up pretty oligopolistic due to electricity costs and economies of scale,” he says. Daian adds that ASICs can improve security because, unlike GPUs, they mine only one coin. If you’ve invested millions into Ethereum-specific ASICs, chances are you won’t conspire to collapse the network.

Last fall, Minehan and her colleagues submitted the code proposal to Ethereum’s core developers, who debated the merits on biweekly calls. The proposal appeared to pass technical muster without objections, as well as broader votes by users (based on the coins they hold), miners (based on their computing power), and a rash of Twitter polls. But the lingering questions and controversy have left ProgPoW’s inclusion in Ethereum’s next update unclear. After months of debate and uncertainty, Minehan believes some GPU miners, spooked by the ASIC threat, might well strike out on their own with the code from Minehan and her colleagues, causing another Ethereum split.

Those issues have led some in the Ethereum community to declare a “governance crisis,” which is especially acute as Ethereum considers bigger, and potentially more controversial, changes to its code than ProgPoW—the centerpiece of which would, ironically, involve abandoning mining. Buterin has described the ProgPoW debate as a “distraction” as developers work toward pivoting Ethereum to “proof-of-stake,” where security would depend on how many coins a person holds, not mining power.

“[ProgPoW] is a stress test about how do you measure consensus from core developers, from users, from miners, from exchanges.” Minehan says. “How do you measure that without one single authority?”

“It’s one of the big unsolved problems of Ethereum,” says Hudson Jameson, a developer who works on community relations at the Ethereum Foundation, a nonprofit charged with helping the network grow. “It’s been hard separating the signal and the noise, finding who the real community is and trying to find out what they want.” Along with a few other developers, Jameson recently convened the Ethereum Cat Herders, a team devoted to guiding projects through the messy governance process. To assuage some of the concerns, the Cat Herders offered to arrange an external audit of the algorithm; but no one has stepped up to fund it.

For now, that’s left ProgPoW in limbo. And that poses a problem for Ethereum’s present, too, Minehan argues. Businesses looking to experiment with blockchain, including Microsoft and JP Morgan, have often looked to build their projects on Ethereum first, she notes. But newer blockchains are releasing designs that are faster and more adaptive; and “permissioned” alternatives from companies like IBM, that offer some of the benefits of blockchain but without the core component of decentralization, are making inroads with corporations. “When someone thinks about using a blockchain for their own internal process, do they want to build on top of a network that has this much infighting?” Minehan says. “They’re not sure Ethereum is fit for enterprise.”

More Great WIRED Stories

Posted by Web Monkey