Technical Roadblock Might Shatter Bitcoin Dreams
A study of the system that powers Bitcoin concludes that it cannot become widely used without a major redesign.
The total value of the digital currency Bitcoin is over $5 billion, reflecting how some people think it will one day become widely useful. But a new analysis of the software that powers the currency concludes that Bitcoin needs a complete redesign if it is to support more than the paltry number of transactions that take place today.
That suggests that the people, companies, and investors who are banking on the currency becoming widely used must overcome fundamental technical challenges that currently have no known solutions, not just the economic and cultural issues associated with a currency independent of any government.
The findings, from a large group of researchers mostly affiliated with Cornell University, also offer a new perspective on an acrimonious debate that has recently riven the world of Bitcoin (see “The Looming Problem That Could Kill Bitcoin”).
Factions in the community are arguing over proposals to adjust Bitcoin’s software so it can handle more than a measly seven transactions a second across the whole world. Yet no tweak to Bitcoin could allow transactions at a scale close to that of conventional payment processors such as Visa without compromising the digital currency's decentralized design, says Ari Juels, a cryptographer and professor at the Jacobs Technion-Cornell Institute at Cornell Tech, and a coauthor of the study. Visa’s system processes 2,000 transactions per second on average and can handle up to 56,000 transactions per second, the company says.
“The current debate is missing the forest for the trees,” says Juels. “We have to think in terms of a fundamental redesign if we’re going to see robust scaling in Bitcoin.” Juels worked on the new study with 11 other researchers from Cornell, the University of California, Berkeley, the University of Maryland, the Swiss Federal Institute of Technology Zurich, and the National University of Singapore. The group's analysis is being presented in a position paper at the Financial Cryptography and Data Security conference in Barbados later this month.
Bitcoin was invented by a person or persons using the name Satoshi Nakamoto, who released the software online in 2009. The currency is powered by a network of computers around the world belonging to a disparate group of companies and individuals without any central authority. The Bitcoin software’s use of cryptography allows that decentralized network to function as a reliable system to process and validate transactions.
Bitcoin’s capacity limit comes from the way transactions are recorded in batches known as “blocks.” Every 10 minutes a new block is added to the digital ledger maintained by the computers in the Bitcoin network, but those blocks have a maximum size of one megabyte, enough for only seven transactions a second at best. Gavin Andresen, who for nearly four years led work on Bitcoin’s software, said last week that the limit is already troubling Bitcoin, with transactions sometimes becoming delayed. “I think it’s a very urgent problem,” he said.
The Bitcoin community is fighting over how best to increase the size of blocks added to the currency’s ledger, with two leading proposals aiming to effectively double it in one way or another. But when Juels and colleagues measured how the network currently performs, they concluded that the strategy of tweaking Nakamoto’s design can’t go much further without compromising the decentralization claimed to be so crucial to the system. The way data moves around the Bitcoin network is too inefficient.
The researchers estimate that Bitcoin’s current design could bear at most only about 27 transactions per second, using a block size of four megabytes, without forcing a significant cut in the number of computers powering the currency, making it more centralized. There is general agreement in the Bitcoin community that the system must remain decentralized to prevent the possibility of any company or government controlling the currency.
“That’s still a fairly limited throughput,” says Emin Gün Sirer, an associate professor at Cornell who worked on the study. “If you compare it to what a big network like Visa is capable of [or] these imagined futures where computers are paying each other, it’s nowhere in the same vicinity.”
Some people believe that Bitcoin can be useful without operating at vast scale, for example functioning as a gold-like asset generally held for long periods. But some of the biggest Bitcoin companies are built on the idea that Bitcoin will come to support a very large transaction volume.
Brian Armstrong, CEO and cofounder of Coinbase, which helps people buy, sell, and use Bitcoins and has raised over $100 million, says he believes the currency will become widely used as a means of payment, particularly in the developing world.
A company called 21 Inc has raised $121 million from investors, including the networking company Cisco, and says that very small “microtransactions” paid in Bitcoin will become an economic backbone used by people and companies to pay for services and goods such as Wi-Fi, data analysis, or music.
Joi Ito, director of the MIT Media Lab, which supports three leading developers of Bitcoin’s code, says that he believes the people working on the currency will be able to find ways it can scale up. But it may not be possible to upgrade Bitcoin’s capacity fast enough to meet the expectations of some investors and the Bitcoin companies they have backed, he says. “I think there are some businesses that have promised returns based on the scaling that are not really reasonable,” says Ito. Ito is a member of MIT Technology Review’s board.
Proposals have been made to create additional decentralized systems that can move bitcoins around but run in parallel to the original network, lightening its load. However, none has been properly tested in a way that proves it can definitely help, and no one knows for sure how to build a Bitcoin-like system that could support very high transaction volumes, says Gün Sirer. Part of the problem is that predicting how Bitcoin-like systems will perform in the real world is very difficult, he says. “The performance is unknown because that’s an emergent property.” These problems are probably less severe for the Bitcoin-inspired systems being investigated by some banks, because they don’t need to operate using a large public network of computers, says Gün Sirer.
Blockstream, one company working on extra layers that might enhance the original Bitcoin network, has raised $76 million, including from insurance giant AXA (see “The Startup Meant to Reinvent What Bitcoin Can Do”). Adam Back, a cofounder of the company, says that in the next 18 months he expects to see multiple systems that can move bitcoins in parallel to the existing Bitcoin network ready for beta testing, from his company and others.
New technology is needed to scale up Bitcoin and support all the use cases imagined for the currency, but the right factors are in place to see it invented, he says. “There’s strong interest from academia, lots of new technology coming in the next 18 months, and a lot of funding coming to the industry,” says Back.
Become an MIT Technology Review Insider for in-depth analysis and unparalleled perspective.Subscribe today