Voltar lista
04 novembro 2019

These Researchers Are Trying to Build a Better Blockchain

Blockchain tech enables systems where no one is in charge, and keeps them secure. But it’s compute-intensive and slow, a hurdle for applications like payments.



There’s a rule in the world of blockchains so ingrained that some call it folklore. Bitcoin, the original iteration of blockchain technology, is great at two things. One is keeping data secure, with a ledger others can’t sabotage. The other is “decentralization,” or getting lots of people to work together without a central authority to call the shots. But those two nice properties come with a big tradeoff: Blockchains can’t scale.

David Tse, a professor of electrical engineering at Stanford, is skeptical of that “blockchain trilemma.” The idea, he noted at a gathering of “cryptoeconomists” last week, had never been precisely defined—he could find no mathematical proof for it. Tse and his colleagues had done the math and, he declared onstage, developed a more efficient algorithm for keeping blockchains secure. And so they formed a company called Trifecta to build out their idea, adding Tse to the growing ranks of academics-turned-blockchain entrepreneurs.

Tse’s talk followed presentations by the Turing Award–winning MIT professor Silvio Micali, founder of Algorand—which has raised more than $120 million from investors—and the MacArthur genius and UC Berkeley professor Dawn Song, founder of Oasis Labs. Each is trying to build new blockchains from the ground up, seizing upon a mismatch between excitement over the core innovation of blockchains and disappointment about their performance.

Tse is a recent blockchain convert. He began working on blockchain about 18 months ago, after a career focused mostly on wireless networks. He’d noticed parallels between the clunkiness of the early mobile web and the current state of decentralized networks. Plus, researching blockchain avoided the drawbacks of machine learning, a field increasingly consolidated in large companies that hoard talent, data, and servers. “Blockchain research, on the other hand, is much more decentralized,” he says.

Academics like Tse have added all kinds of new features to blockchains, often with the aim of improving security and privacy. Those projects include Zcash, where a team of professors brought privacy-protecting technology to Bitcoin, which is increasingly tracked by law enforcement.

But building more robust blockchains requires a more complete overhaul. Bitcoin and Ethereum, the largest blockchain networks, are kept secure by “proof-of-work” protocols, which involve a race to solve cryptographic problems. The process takes oodles of energy and is painfully slow. Bitcoin can handle seven transactions per second, while Ethereum ekes out 15. That’s not fast enough to handle basic commerce, much less complex applications. A hopeful benchmark often professed by blockchain entrepreneurs is “Visa-level performance,” or thousands of transactions per second.

The newcomers are trying a variety of approaches to get there. Bitcoin inspired some researchers to take another look at older, mostly discarded cryptographic approaches designed for decentralized systems. Others look to extend the algorithm used for Bitcoin, known as the Nakamoto consensus, named after the network’s pseudonymous creator. One strategy is to separate the security mechanisms from other functions, like data storage and computation.

“We have an in-house joke that there is no blockchain trilemma,” says Kevin Sekniqi, a researcher who left his PhD studies to start Ava Labs along with his adviser, Cornell professor Emin Gun Sirer. The company grew out of a theoretical debate in the lab about how they’d build their ideal consensus algorithms from scratch. “Not just tweak them around but really reinvent them from the ground up,” Sekniqi says. Eventually, they realized they had the ingredients for a competitive product.

The existing blockchains are making their own upgrades. Ethereum has a long-term vision called Eth 2.0, which would revamp the protocol and eventually do away with proof-of-work consensus. But it’s complicated. Ethereum is a multibillion-dollar platform—host to decentralized games, collectibles, even financial products—but it has no central decision-makers. That’s core to its philosophy and design. The upgrade process involves a complex coordination game where technology is built and shepherded through a voting process by a diffuse ecosystem of developers. The difficulty is avoiding changes that jeopardize existing products built for the platform, and that appease the voters—the miners who run each node.

There are also solutions that don’t touch the underlying blockchain itself. That includes the Lightning network for Bitcoin, which adds software to the existing protocol to enable speedier peer-to-peer payments. But such technologies have proven tricky to integrate, given the inflexibility of the underlying protocols.

In the meantime, potential competitors see an opening. “We’re starting fresh and we don’t have billions of dollars on our shoulders,” Sekniqi says. But there’s risk, he adds, in trying to turn an academic exercise into a product. “A lot of founders in crypto just build things without really thinking about use cases,” he says. “You don’t stand out if you just have performance.”

Ava is hoping to get around that problem by initially tailoring its blockchain to make certain financial products, like insurance contracts, more liquid and tradable. The system is also being designed to be modular and work with existing blockchains, including Ethereum, potentially taking advantage of their existing users and applications. Last month at Ethereum’s annual conference in Osaka, Japan, Ava announced a “friendly fork” of the protocol, intending to essentially copy Ethereum’s functionality and install Ava’s machinery underneath.

It’s too soon to tell if the competitors will catch on with users—communities like Ethereum’s, for example, are unlikely to hurriedly embrace a new platform still largely controlled by a single company. And none of the upstart networks have been fully built out, meaning their promised improvements are still mostly theoretical. In the meantime, some are trying to bootstrap their networks. Beyond the money raised to build its platform, Algorand, the MIT spin-out, is bolstered by a $200 million fund that plans to invest in companies that will build services for its network. Blockstack, another built-from-scratch blockchain with PhD-founders, has its own program for rewarding app developers.

Even by that standard, Tse’s company is just getting off the ground. Next up is a hunt for funding and getting the ideas in his demo working in practice. Tse believes this is like the early days of mobile connectivity, when the market was filled with a variety of standards. “There have been a gazillion projects on focused scalability,” Tse says. “You’ll expect the process of elimination will take a while. We believe that the best technology will win, and that’s ours.”