A Supercomputer Takes Humanities Scholars Into the 21st Century

By JOSH FISCHMAN

Article tools

Printer friendly	E-mail article	Subscribe	Order reprints
	Discuss any Chronicle article in our forums
Latest Headlines U. of Delaware Approves New Diversity Discussions for Students The new, voluntary program will replace one tossed out last fall after coming under criticism as "thought reform." In Tight Employment Market, Career Services Gain Clout Borrowers' Uncertainty Has Colleges in Quandary How It Does It: The RIAA Explains How It Catches Alleged Music Pirates State High-School Exit Tests Do Not Improve Academic Achievement, Study Finds Commentary Business Students Should Learn More About Science

The phrase "one million" in grant announcements tends to be eye-catching because it is usually linked to the word "dollars." But late Monday afternoon, when officials at the National Endowment for the Humanities used that figure in describing a new grants program, they were not talking about money.

"We are offering one million hours of high-performance computing," said Brett Bobley, director of the endowment's new Office of Digital Humanities.

That is one million hours, doled out in chunks of 100,000 to 500,000, that can be used for such research efforts as analyzing records of the U.S. Census to trace subtle economic and social shifts, or examining centuries of texts to discover the origins of a single linguistic concept.

"It's tremendously exciting," said Vernon Burton, a historian and director of the Institute for Computing in Humanities, Arts, and Social Science at the University of Illinois at Urbana-Champaign. "I think it has the potential to move forward the basic boundaries of human knowledge."

That kind of language usually comes from physicists who are describing a powerful new particle accelerator, not scholars specializing in the Civil War, as Mr. Burton does. Yet, like physicists, "humanities investigators can have huge amounts of unstructured data," said Bruce Cole, chair of the endowment. "That's where we think this can help. It's a subset of a lot of other stuff we are doing, such as the National Digital Newspaper Program, which is taking 30 million pages of microfilm, the first great draft of history, and digitizing them."

Finding patterns in the data is exactly what high-performance computing is good at, but the tools are generally unfamiliar in the world of arts and letters. "We just want to expose scholars to the possibilities," said Mr. Bobley.

The endowment, along with the Department of Energy, will be administering the grants, known as the Humanities High Performance Computing Program.

Teraflops of the Great Emancipator

All of the computing will be done at the Energy Department's Lawrence Berkeley National Laboratory, in Berkeley, Calif., in the National Energy Research Scientific Computing Center, known as Nersc. The facility has a Cray supercomputer named Franklin that can reach speeds of 101.5 teraflops per second—more than 100 trillion operations performed every tick of the clock—and four other computational speed demons. The hours that grantees get on a computer will be measured by time used by the central processing unit, or CPU.

But hardware isn't the only attraction drawing the humanities program to Berkeley. "Nersc has a large and advanced training facility," said Mr. Bobley. "We anticipate that humanities scholars will need a fair amount of training in using these machines, and Nersc has a staff to do just that."

It's not typical duty for the computing center. In 2006 the facility's big users were from the fields of fusion energy, materials science, and chemistry. They took up 54 percent of its computing time. Most of the rest was accounted for by researchers in various fields of physics, mathematics, climate science, and geoscience or engineering.

So historians and literature scholars will be as novel to the computing-center staff as supercomputers will be to most of the humanities researchers. "It will be a very interesting challenge," wrote Jonathan Carter, Nersc's user-services group leader, in an e-mail message. "While the total of the NEH time isn't a big fraction of the total time we provide, a grant of 500,000 CPU hours would put a project into the top 20 or so biggest projects."

Big projects are, of course, the whole point of the grants. Simulating entire economies or collecting everything Abraham Lincoln ever wrote to follow the evolution of the idea of "emancipation" are some of the investigations envisioned by scientists and scholars. "It is 21st-century research," said Francine D. Berman, executive director of the San Diego Supercomputing Center. "And it demands 21st-century tools."