In This Story
Let’s unpack the portmanteau word “cryptocurrency.” The combination of two unlikely bedfellows—cryptography, a subfield of computer science, and currency, a topic in economics—is at the heart of the transformative potential of its underlying blockchain technology. But the uniqueness of the pairing can make it very difficult for research professionals in either field to predict, let alone positively influence, blockchain’s future development.
Jiasun Li, an assistant professor of finance at Mason, is among an elite group of academics who are bridging the divide by merging relevant concepts from computer science with game theory—a subfield of economics that studies the interactions of decisions made by interdependent economic actors.
“Learning from the interaction between both economics and computer science is going to be very fruitful,” Li says. “People working in crypto already know this. The problem is that there are very few people well-versed in both fields; economists don’t understand what computer scientists are talking about, and vice versa. Part of my value-add as a researcher is that I can interpret between them.”
For example, in a 2021 Review of Financial Studies article (co-authored by Lin William Cong of Cornell University and Zhiguo He of University of Chicago), Li looked into whether the phenomenal growth of Bitcoin mining pools would lead to dangerous centralization over time—a possibility concerning to many practitioners. If too much activity were concentrated within one or a few pools, it might undermine the purpose of a decentralized system.
As Li’s game-theoretical models demonstrate, however, the very force driving miners to form pools—i.e. the desire to stabilize mining income via risk sharing with peer miners—acts as a drag on centralization. Optimal risk-sharing strategies for an individual miner would include distributing computing powers across multiple pools, rather than putting all their eggs in one basket. A large pool will optimally charge a higher percentage fee, so it tends to grow more slowly as miners choose to reallocate their computing power to different pools. Therefore, Bitcoin mining has a certain degree of built-in resistance to monopoly.
Li's working paper with co-author William Mann (of Emory University) has significant policy implications for initial coin offerings (ICOs) and digital tokens. Tokenization is widely considered to be the capitalization engine of many blockchain projects, yet regulators are divided on how to treat tokens. Are they securities like any other, and thus subject to existing laws such as the Securities Act of 1933?
Li’s model, incorporating a concept from game theory called forward induction, shows that purchasing a platform-specific token can signal future intention to adopt the platform, in addition to potentially supplying capital in the here and now. For example, purchasers of the FIL token can be reasoned to be future users of Filecoin’s “decentralized storage” services, which in turn motivates potential storage providers to contribute. The token serves as both a medium of exchange and a coordination tool for building—and binding together—both sides of the platform. Therefore, the researchers suggest that tokens that do both deserve their own regulatory category.
The above dynamic
presumes a token with little or no worth outside the originating platform. As the off-platform value of the token rises, its coordinating power drops because the temptation to cash in their chips becomes stronger for token holders. Therefore, the authors also recommend that entrepreneurs who seek to build successful platforms curtail token holders’ ability to engage in speculation, e.g. by offering separate security and utility tokens rather than issuing only one type of token.
Li’s most recent paper employing game theory (co-authored by Hanna Halaburda of New York University and Zhiguo He) examines the economic incentives involved when nodes within a blockchain form consensus. How well this consensus formation process works is a critical issue for blockchain systems. For example, in the application of cryptocurrencies, transaction histories must remain the same across blockchain nodes in order for the system to remain reliable. Furthermore, blockchains must have a way of reaching consensus even if some nodes do not work properly (for example, in cases of some nodes malfunctioning or being hacked).
Over decades, computer scientists have developed “Byzantine fault tolerance” (BFT) protocols to work around compromised, or “Byzantine” nodes and obtain consensus. These protocols are now widely used by almost all major tech companies for their in-house services. The permissionless nature of many blockchains, however, distinguishes them from those existing in-house applications. Since they don’t fall under one organizational umbrella, nodes in a blockchain may develop an adversarial relationship due to misaligned incentives. Hence, protocol designers must carefully specify the payoffs and penalties each node derives from successful or failed consensus. Enriching prior BFT research, Li’s game-theoretical model in this paper shows that there are many equilibria within such systems (an equilibrium of a game specifies a set of strategies by all nodes so that each does what is optimal for itself).
For example, there is always an equilibrium in which nodes fail to commit to anything at all. The model provides a framework to further analyze alternative versions of BFT protocols with explicit account of economic incentives.
Game theory is a valuable tool because it enables Li to model the “action space” of a decentralized system like in a blockchain and thus start to make sense of how it works. From there, academics can theorize about the answers to critical questions about this emerging technology.
Sources: Lin William Cong, Zhiguo He, Jiasun Li (2021). “Decentralized mining in centralized pools,” Review of Financial Studies
Jiasun Li, William Mann (2020). “Digital tokens and platform building,” working paper.
Hanna Halaburda, Zhiguo He, Jiasun Li (2021). “An economic model of consensus on distributed ledgers,” working paper.