Blockchain’s Weakest Links

But not all systems being called blockchains work the way Nakamoto described, and not all are decentralized. When some companies talk about integrating blockchain into their operations, they’re describing something with a central authority—a database that does not involve mining or maintaining trust between anonymous parties, says Chicago Booth’s Eric Budish. On this type of blockchain, transactions are recorded and a database is maintained by a central authority, such as a computer or person at the company. A copy of the verified chain sits on every computer on the network, making the system distributed but not decentralized.

Some researchers call these “private” blockchains, and Budish questions whether they should be called blockchains at all, rather than simply better databases. “A lot of the excitement about blockchain is [actually] excitement about better data-management processes,” he says.

The business applications of this technology could still improve how companies interact and run their operations, and represent better record keeping. But centralized databases may not be as safe, as each has a single point at which its system could potentially fail. One of the cornerstone precepts of blockchains is that they operate at the highest levels of security, but that’s not true of private blockchains, says Imperial College PhD candidate Engin Iyidogan, one of dozens of researchers who have been studying blockchains from every angle.

Moreover, if private blockchains fail, it could tarnish all blockchains by diminishing people’s faith in them—although it could also spur innovation. “If people realize that private blockchains are not disruptive or efficiently implementable, the hype over blockchain technology vanishes and we focus more on applications of true decentralized systems,” Iyidogan says.

Private blockchains aside, much of—if not most—economics research taking place looks at Nakamoto’s proof-of-work system, which many economists treat as the default blockchain design. One celebrated advantage of the proof-of-work blockchain: it can make markets more efficient by helping transactions to settle more quickly, eliminating the time between when parties agree and when the deal closes, and by extension the costs. “Most people agree the potential for blockchain is very high as a mechanism for streamlining the transaction costs of settlement,” Cornell’s Maureen O’Hara says.

If deals were to close more quickly, the money tied up in a pending transaction could be put to better use. In the leveraged-loan market, for instance, where companies borrow cash at steep rates, it reportedly takes 27 days to settle a transaction, compared to just two or three days for a stock trade, and agents spend a quarter of their overhead expenses on answering investors’ calls to confirm the details of loans. Distributed-ledger settlement would obviate these problems, experts say.

But it won’t be easy for the Nakamoto blockchain to evolve into a mostly seamless market like the major stock exchanges. Some research suggests the transaction fees involved will be an impediment that prevents blockchain from growing.

Transaction fees are an increasingly important component of these blockchain systems, and they are closely related to transaction times. Today, if you want to make a transaction, you offer miners a fee to include your transaction in a block. The fee you offer depends on how quickly you need that transaction processed; if you want your transaction prioritized, you pay a higher fee. In addition to the miners’ fee, you may pay a fee to a bitcoin exchange to process the transaction and, if wanted, to transfer bitcoin back into dollars or another noncryptocurrency.

However, only 21 million bitcoins can be mined under Nakamoto’s design, and about 17 million are already in existence, according to blockchain.com. When the final bitcoin is created, the market will switch to a purely transaction-fee-based system. At this point, these fees will be the only way to attract miners to a block. The higher the fees offered, the more mining power will be directed to the blockchain.

Transaction fees help the market reach equilibrium by allowing miners to earn more for processing some blocks faster—while helping more time-crucial transactions to be processed more quickly than others. But O’Hara, Cornell’s David Easley, and Cornell PhD candidate Soumya Basu find that transaction fees are an impediment to bitcoin holders who want to use the cryptocurrency as a medium of exchange to pay for real-world things, rather than to hold as a speculative investment. “As users battle to get transactions posted on the blockchain, transaction fees are rising to levels that discourage bitcoin usage, highlighting an important structural issue confronting the blockchain,” O’Hara, Easley, and Basu write.

According to the researchers’ economic model, transaction fees make mining profitable over time—but these fees won’t speed up transaction times enough to counteract the number of users who get fed up with waiting and leave the system. “The fees directly induce some users to drop out, while increasing wait times cause other fee-paying users to depart as well,” they write. The researchers identify the problem, but they don’t propose any remedies for it.

Moreover, according to another group, once all bitcoins are mined, transaction fees may not raise enough money to support the system’s infrastructure. In a comparison of bitcoin’s payment mechanisms and those of a traditional company, Columbia’s Gur Huberman and Ciamac C. Moallemi and Chicago Booth’s Jacob Leshno find that the transaction delays and high fees that plague blockchain-based settlement are an inherent part of the proof-of-work system.

Nakamoto’s system has some congestion baked in. It makes people wait for transactions to be compiled into blocks, and then wait again for the grouped transactions to be verified. And it will eventually make some people wait longer than others, if they’re unwilling to pay higher fees for faster service. “Congestion is not merely an engineering necessity, but also a device to motivate users to pay transaction fees,” they write.

In a theory paper, Ma, Gans, and Tourky argue that the Nakamoto system allows anybody to mine—granting free entry to whomever has the software to compete. If the system instead were to limit the number of miners, this would do more to reduce the amount of computing power, and electricity, that miners expend. Granted, that would require Nakamoto, if still alive, to make changes to the system.

Another trio of researchers—Chicago Booth’s Lin William Cong and Zhiguo He and George Mason’s Jiasun Li—say that Nakamoto’s system creates a wasteful energy arms race. Miners, to stay competitive, invest in computing capabilities by spending heavily on networks, as well as electricity and cooling. The investment may improve a miner’s chances of winning a computational competition. However, Nakamoto’s system is a zero-sum game, so if investment benefits one miner, it directly hurts the chances of other miners.

But regardless of who wins a competition, or how many miners compete, or how much energy they use, one block will be added to the chain every 10 minutes, on average. The effort that goes in may grow ever larger, but the outcome remains the same. So the intense competition produces no benefit for the system or end users. “The arms race nature of this technology is what’s underlying the electricity usage,” says He. He and his coresearchers point out that risk-sharing considerations lead miners to work together in the form of large mining “pools.” The rise of mining pools is a financial innovation that improves miners’ risk sharing; but it enables miners in a pool to devote greater computation power, aggravating the arms race that consumes a tremendous amount of energy. (For more, see “Are blockchain mining pools problematic?”) Excessive competition, say the researchers, is an inherent part and problem of Nakamoto’s design.

But again, not all blockchains resemble Nakamoto’s original vision. Some competing cryptocurrencies are based on proof-of-stake, which validates transactions differently. The proof-of-stake system doesn’t involve races to solve mathematical puzzles, and thus there is no reward for doing so. Instead, the amount of cryptocurrency a miner holds serves as a validation of trustworthiness and functions something like a bond: a miner has to hold a certain amount of cryptocurrency before being allowed to verify and add blocks to the chain. Miners with more cryptocurrency have more mining power, and they make money simply through transaction fees. This system is still a public blockchain, not a private one, because it has a distributed ledger and no central authority. But because it doesn’t require many computers to run the same equations as they race to solve a puzzle, proof-of-stake uses less electricity.

Ethereum is reportedly switching to a proof-of-stake system, in part to mitigate the environmental costs of proof-of-work’s massive electricity bills.

McGill’s Fahad Saleh argues that such blockchains using proof-of-stake can be economically viable because this system can achieve consensus: miners agree that a block is valid and allow the blockchain to continue. It’s important, however, that proof-of-stake systems incorporate rules that require users to have a sufficient stake of cryptocurrency to participate in validation, he writes. Otherwise, users could get in the way of consensus to drive up their own rewards.

“My conclusions emphasize the need for developers to heed economic guidance when designing consensus protocols,” Saleh writes. If participants own a lot of the cryptocurrency being used to run the blockchain, they have an incentive to not tank the currency’s value.

Sabotage is another looming issue, according to Booth’s Budish. He took a theoretical look at the economics and security of Nakamoto’s blockchain, focusing on the large relative cost required to make blockchain secure. In a series of equations, Budish lays out his concerns.

He started by asking how much computational power miners are buying for tournaments. The answer, he says, is that it depends on how much miners are compensated. The more miners can make, the more they will mine.

But for miners to stay honest, they have to know that they will make more money by mining than by sabotaging the system, Budish notes. The amount miners receive over time as they help to run and maintain the blockchain is known as a “flow,” while the one-time haul from an attack is known as a “stock.” “The recurring ‘flow’ payments to miners for running the blockchain must be large relative to the one-off ‘stock’ benefits of attacking it,” Budish writes.