Posts

Brown CS faculty member Maurice Herlihy is widely known for his work in practical and theoretical aspects of concurrent and distributed systems, but readers may be less familiar with one of his earliest research efforts. As a teenager, he traveled to Florence, Italy, where he helped his father, David Herlihy, on a project where he created punch cards based on data in the Florentine castato, or land registration system. 

David, a historian and early demographer of Medieval European populations, pioneered the use of computers for data-driven historical analysis and is often acknowledged as the first historian to apply the power of the UNIVAC computer to historical census, property, and tax records, providing social and economic insights to our understanding of the past. Herlihy utilized computers for the bulk of his career to compile and analyze large quantities of information, often referred to as “big data” today.

“The background of this project, which was done in my high school years, is that the city of Florence did a very detailed census of Tuscany in 1427 that was kept in this room in the Uffizi Gallery, a famous museum that used to be the State Archives,” Maurice says. 

“My father realized that this amount of data could be tabulated on a computer, and at the time, the idea that you would use a computer for history was considered weird because you couldn’t get funding for it,” he adds. “Because the National Science Foundation said, this is history, science, and the humanities, so why use a computer?”

David’s process involved obtaining microfilm copies of the data and then transcribing it by hand for the key punch machine to transform into cards. Maurice’s job was to read the microfilm and enter names and data from other fields, which he could do because of his knowledge of Italian and paleography.

Maurice notes that the work was challenging because people in the 1400s wrote very differently than we do today, and he had to take many interesting discrepancies into consideration. For example, the tail of the letter “g” ascended over the top of the word instead of descending below it. Abbreviations like “GVI” for the name “Giovanni” were common, but consistency was rare and still required him to know about 10 different abbreviations for “Giovanni”. 

“I lived with my family in Italy and knew Italian and knew about Tuscany, so this knowledge proved useful,” Maurice says. “I would also sometimes give advice on historic record and once discussed how there were more efficient sorting algorithms than Bubblesort, which was developed during the project.”

Maurice says the research project received funding from the Sorbonne in France when he and his father visited Paris because they were working with a collaborator there, and a book on the project was published in the 1980s.

“Now, the idea that you do data science is so normal that we don’t think about it, especially…the idea that computers could be applied to the humanities,” Maurice says. “In those days, a computer would literally fill a room; you had to drop off the cards in the evening and come back in the morning to check, so it was a very different world.”

Maurice explains that computers were a completely new innovation and econometric modeling had just begun, so this was a “radical vision” on his father’s part. It’s difficult to convey, he says, how odd their work seemed to the average person of the time, who only viewed computers as calculation devices.

When asked about his favorite moments working with his father during the project, Maurice mentions that he enjoyed deciphering the handwritten census entries because the paleography was difficult, and the process of determining what certain words truly were, like the word for “carpenter” or other professions, proved to be quite interesting.

One of his most intriguing findings during the project, he says, was his conclusion that many people didn’t bother to report female babies. This was due to discrepancies that he noticed between the census numbers and infant reports.

“This is an example of a kind of question that would have been impossible to answer normally at the time, but with very simple tabulation, you can ask questions like 'how old were men when they got married’,” Maurice says. “In order to get married, you had to have a dowry, so men got married in their thirties to teenagers, which had a side effect that the men would die and leave young widows behind running their businesses, turning into a trend.”

“There were some unique stories that I didn’t actually see, but I remembered. For example, there was one person who had been kidnapped by pirates,” Maurice says. “Probably for real. Some of these things would have little narratives explaining some kind of anomaly, and the guy was kidnapped by pirates and then finally released later, and he made his way back.”

“One thing I learned about is that it’s possible to make connections between areas that superficially seem very different. For example, the idea that you could use the power of computation to learn about history,” Maurice adds. “This opened me up to the idea that it’s possible to find connections between one thing that is really good at something and your problem, and that there might be a way to bring these two together when defying conventional wisdom.”

Maurice says that several of the first-year students in computer science that he advises also have multiple concentrations, and his favorite combination is when students combine computer science and classics.

“I think the ease with which you can study multiple disciplines is a huge reminder that this is the last time in your life where this is going to be the case, which is something that many students here take the advantage of,” Maurice says. “I always try to encourage the students that I advise to take French literature courses, Eastern philosophy courses, and just any other things that are not STEM courses.”

For more information, click the link that follows to contact Brown CS Communications Manager Jesse C. Polhemus.