Imagine an egg sitting on a crossroad. It’s fragile and vulnerable. Everyone can see it, and anyone could break it yet it remains whole, even years after you first encounter it.
This is the image Yaojin Sun, an analytics doctoral student at the Haslam College of Business, describes when explaining why he first became fascinated with blockchain in 2015. Blockchain, the technology underpinning cryptocurrencies such as Bitcoin, creates security via transparency and replication. It allows a network of computers to maintain and update a single ledger simultaneously. Altering records in a blockchain means accessing every instance of the ledger at the same time and changing all transactions before and after the record in question.
Sun wanted to understand precisely how blockchain connects records so strongly and soon devoted his time to researching some foundational, technical questions about it.
“I met with many investors,” he says. “I tried for around a year, but no one seemed to have answers.”
After joining the Ph.D. program at the University of Tennessee, Knoxville, Sun met a research partner who helped him to solve several of those basic questions using graph theory representation. Harrison Hicks, another analytics doctoral student, initially came to the program to study machine learning.
“Yaojin started talking to me about blockchain, and I decided that if I wanted to work on something really cutting edge, I should make this the focus of my research,” Hicks says. “I hoped that together we could bring some practical business applications to the theoretical research.”
At the recent ETH Denver Conference in Colorado, Hicks and Sun won a challenge by start-up Polymath to develop a better way to meet Know Your Customer regulations. They created an algorithm that determines a users’ trustworthiness based on their previous transactions in the blockchain.
Such an algorithm can help protect Bitcoin against money laundering and create a means of assessing reputation for anonymous users. “We provided a toolkit to access more information from the blockchain, modeling the ledger as a massive network of interactions,” Hicks says. “The information we extract could be used to derive something like a credit score, but many other interesting questions could be posed as well.”
Hicks and Sun met several companies pursuing similar research that may offer possible collaborations. Attending the conference and being exposed to other experts in the field changed the direction of their work.
“We really had this idea that we had to do everything from scratch and start with really low-level code,” Hicks says. “At the conference we realized that everyone else has started using the higher-level protocols, and a lot of non-trivial work could be done there.”
Since blockchain technology is so new, Hicks and Sun say little has been published on it in academic texts. That makes their research interesting, but also presents a unique situation for their doctoral progress.
“Our advisors took a bit of a risk with us,” says Sun, who works with Hamparsum Bozdogan, the McKenzie Professor in Business and an expert in statistics and data mining. Hicks’ advisor is Wei Zheng, whose research focus is in machine learning and design of experiments.
“For them to take a chance on us and give us moral support in this area is huge,” Hicks says. “It might not be their field of study, but they ask excellent questions and help us make connections.”