A terrific rumpus broke out in the world of Bitcoin last week, when veteran developers Gavin Andresen and Mike Hearn published Bitcoin XT, a competing version of Bitcoin Core, the open-source program that generates new bitcoins, verifies all transactions, and records them on the massive distributed ledger known as the blockchain. Andresen is Bitcoin’s most senior developer, and has been the project’s de-facto leader since 2011; Hearn is a pioneer in the development of Bitcoin wallets. According to an explanation Hearn posted on Medium, they released XT because “the decision-making process in Bitcoin Core has broken.” The real crisis, as this suggests, is less about Bitcoin’s code than about the power structure that governs it.
The dispute centers on a fast-approaching computational bottleneck. The blockchain is composed, as you might guess, of “blocks,” each consisting of a list of recent bitcoin transactions, time-stamped and verified by one of the network’s “miners”—computers that secure a block’s inclusion on the chain by solving a computational puzzle called a proof of work. The first computer to win this computational race “mines” a reward of twenty-five bitcoins, and then competition recommences for the next block.
According to a Bitcoin Foundation post by Andresen, Satoshi Nakamoto, the creator of Bitcoin, imposed a maximum block size of one megabyte in 2010, in response to some denial-of-service attacks.* This cap effectively restricts the number of bitcoin transactions to about seven per second, since it takes a certain number of characters to record a given transaction, and only so many fit in a megabyte. (Visa, by contrast, routinely handles thousands of transactions per second, and claims a capacity of more than twenty-four thousand transactions per second.) Hearn anticipates that Bitcoin transaction volumes will blow past the seven-per-second limit by 2017, at the latest, creating backlogs, settlement delays, and perhaps even mass outages. The primary goal of Bitcoin XT, then, is to increase the size limit of each block to eight megabytes.
Prior to the release, a block-size debate had been blazing for months on Bitcoin forums. Opponents of an increase argued, for instance, that mining larger blocks would require more computing power, thereby discouraging small operators in favor of the massive mining farms that have gradually concentrated the network into fewer and fewer hands. It was widely feared, as well, that a serious disagreement among the core developers might further destabilize public faith in Bitcoin. Those in favor of an increase contended that “forking”—creating a competing version of a program in response to diverging aims—is the very essence of open-source software development.
Such decisions stand to affect millions of participants in the bitcoin economy, but ultimately lie with a very few people. Theoretically, any user of an open-source program is free to create, adopt, or reject any alterations he pleases; in practice, software shared on networks or between users requires painstaking standardization. This means relying on decision-makers with “commit access,” who have the right to amend a software project directly, on their own. This status represents a high level of developer control, and of user trust.
Andresen told me that, to his recollection, when Nakamoto withdrew from the project (and from public view), in 2011, only he, Nakamoto, and possibly one other person had commit access to Bitcoin’s software. Andresen eventually granted this level of access to four additional developers, for a total of five “core devs.” In April, 2014, Andresen decided to devote more of his time to other projects, and named one of the core devs, Wladimir van der Laan, to succeed him as lead developer. Even today, only van der Laan and Andresen can grant commit access to other developers of Bitcoin Core.
“Right now,” Andresen added, “just Mike and I have commit access to Bitcoin XT.”
The publication of XT threw the Bitcoin forums into turmoil, with partisans duking it out in what has amounted to an old-fashioned flame war. Some of the public debate was constructive, though. Roger Ver, a Bitcoin investor and evangelist famously nicknamed “Bitcoin Jesus,” tweeted, “If you run a full node, and support bigger blocks so more people around the world can use bitcoin, install XT today.” Nick Szabo, a respected cryptography expert whom some believe to be the real Satoshi Nakamoto (whose identity and location remain mysterious), publicly stated his opposition to XT, tweeting, “A rapid block-size increase is a huge security risk: a reckless act to be performing on a $4 billion system.”
One of the most vocal critics of XT has been Gregory Maxwell, one of Bitcoin’s five core developers. Aside from the many other enhancements he has made to Bitcoin Core, Maxwell is a developer of “sidechains,” a clever proposal for scaling Bitcoin by processing transactions partly off the main blockchain. But he and fellow core dev Pieter Wuille have faced criticism, during the recent debate, over an apparent conflict of interest: both are co-founders of Blockstream, a startup venture, capitalized to the tune of twenty-one million dollars in a recent seed round, that is experimenting with a number of scaling options, including sidechains. Should any of Blockstream’s proprietary innovations become essential to Bitcoin, the company could make a fortune. On Reddit, a user with the handle yumein observed as much, suggesting, in imperfect English, that preventing the move to a larger block size might one day force Bitcoin users to adopt Blockstream. “Nothing against Blockstream,” he wrote, but, “might be a big reason why this gentlemens stand against bigger blocksize, isn’t it?”
Adam Back, another co-founder of Blockstream and a pioneering cryptographer in his own right, maintains that his company only wants what is best for the system as a whole. “We really want and need Bitcoin to function well, in terms of its security and scale and safety,” he told me. I asked Back, who strongly opposes XT, why he thought that Andresen and Hearn had released their software independently of the other core devs. “I’m a little confused about why that is,” he replied, adding that he had spoken with Andresen and met with Hearn for several hours in the weeks before XT was published. “I think the tech community was expecting and hoping that Gavin would work within this process . . . and see which [proposal] is accepted.”
“Why did that not happen?” I asked.
“You would have to ask Gavin that,” he said, laughing a little ruefully.
Problems of governance have beset Bitcoin from its inception. Notably, there have been mass resignations from the Bitcoin Foundation, the cryptocurrency’s standards body. Furthermore, some of its most prominent members have been in trouble with the law: Charlie Shrem is serving a two-year sentence on money-laundering charges, and Mark Karpelès, C.E.O. of the bankrupt MtGox bitcoin exchange, is being held in a Tokyo jail on suspicion of fraud. In the absence of institutions capable of implementing clear standards, it’s plain that Andresen and Hearn decided to take matters into their own hands. XT is above all a path toward establishing new leadership. I asked Andresen whether, if XT were to achieve full acceptance, he would then include all the earlier Bitcoin core devs in the new XT team. He replied that “[XT] will have a different set of developers. Part of the reason for forking is to have a clear decision-making process for the software development.”
Just before Nakamoto released Bitcoin, in January, 2009, he authored a post at the P2P Foundation. “With e-currency based on cryptographic proof, without the need to trust a third party middleman, money can be secure and transactions effortless,” he wrote. Six years in, the total value of all extant bitcoins is roughly three and a half billion dollars. You can now pay for your Dish subscription and buy gifts from 1-800-FLOWERS or plane tickets from Expedia using bitcoin. Nearly a billion dollars in venture capital has been invested in Bitcoin-related businesses. But Nakamoto’s predictions regarding trust, security, and effortlessness have failed, repeatedly and dramatically.
It will take several months, at minimum, to learn whether or not Bitcoin XT will prevail, because Andresen and Hearn designed the code to require a kind of vote before the block-size increase takes place. For the moment, XT is very nearly identical to Bitcoin Core, and is writing to the same blockchain. But Bitcoin XT identifies each newly mined block with a version number that acts as a sort of polling device; those miners whose blocks are signed by XT are thereby signalling their approval of increasing the block size. After January 11, 2016, if (and only if) seventy-five per cent of the thousand most recent blocks are signed by XT, a waiting period of two weeks will kick in, during which miners will have time to adopt XT in advance of a true “hard fork.” Absent a rapprochement between the two camps, the ledgers will then begin to diverge, with larger blocks appearing on the XT blockchain but not on the Core one. This could give rise to problems, such as the same bitcoins being spent twice, once on each ledger.
Andresen seems fairly certain that sufficient consensus will be reached, at least to preclude diverging ledgers. He wrote in an e-mail:
Within days of its publication, the first blocks signed by Bitcoin XT were recorded to the blockchain. “I think there’s a very good chance XT will eventually reach 75% of miner support,” Andresen wrote, adding that he expects this to take six months or more. The number of new XT nodes was being recorded at xtnodes.com, which soon found itself targeted by denial-of-service attacks and flooded with false statistics via “pseudo-nodes”; it is still hopelessly messed up.
On Monday, a letter signed by the chief executives of eight leading Bitcoin companies, among them Stephen Pair of BitPay, Peter Smith of Blockchain, and Jeremy Allaire of Circle, threw some significant weight behind XT, pledging philosophical and technical support for a block-size increase (though not explicitly for XT, versus, say, a redesign of Core). These eight firms represent enormous technical expertise and experience, and are vested with a massive amount of venture capital—well over a hundred million dollars, in all. Unless the dissenting core devs manage to build a better mousetrap very quickly, it’s quite possible that Bitcoin XT will achieve the target adoption level and emerge as the dominant system. But one might be forgiven for supposing, if the wild ride of the last six years is any guide, that there are more surprises in store between now and January.
Andresen seems to think so, too. I asked him about Bitcoin’s checkered past—the rise of mining pools, the implosion of MtGox, the Silk Road debacle, and the huge number of irrecoverable bitcoin thefts from individuals and businesses—have all these disasters altered his thinking about the future of Bitcoin, and about the blockchain?
“No,” he replied. “I think if you go back and look at even my earliest interviews from 2011, I was warning people to expect ‘chaos and drama.’ New technologies are always full of chaos and drama, because the startup companies that first experiment with them often fail. I still expect lots of chaos and drama over the next few years.”
*This sentence has been revised to clarify when Bitcoin added the one-megabyte block-size limit.
