Demystifying political gridlock with supercomputers

Political infighting has halted the US federal government 12 times in the last 34 years. With the help of Texas Advanced Computing Center (TACC) supercomputing, University of Texas-Austin researcher Eric Svensen hopes to determine why this political strategy emerged. 

Federal functions cease when the government stops. Employees and soldiers are not paid; national parks and monuments are closed; government offices are shuttered — everything just stops. But before the Reagan era, US government shutdowns didn't happen.

"Even during the Nixon administration, where there was a Republican president and Democratic Congress and a major war, government passed a large number of landmark laws," said Svensen, a researcher and lecturer in the Department of Government.

What changed? Political realities: Mandatory spending for programs like Defense, Social Security, and Medicaid cannot be cut without a great deal of political pain, which means discretionary funding is dwindling.

With a completely Republican-controlled Congress and a Democratic president, the tenacity of legislative gridlock is now unprecedented. To find out why, Svensen compares biennial legislator ideology scores with annual spending. Ideology scores range from -1 (extremely liberal) to +1 (extremely conservative), and are derived from roll call votes.

"I want to understand how party polarization in Congress has changed over time due to both increased and divergent spending priorities,” says Svensen. “By recovering estimates of legislator preferences on all roll call voting, I create polarization scores to describe the change.”

His study looks at US voting since the second World War. Seventy years of voting history quickly adds up to more than a typical computer can handle, however.  Fortunately, TACC allocated time on their Stampede supercomputer to help Svensen make short work of those numbers. Humming along at nearly 10 petaFLOPS, Stampede is a formidable force.

“I’m dealing with more than 10 million individual data points, which I then have to aggregate into 40 or 50 data points for an annual polarization measure," Svensen says. "It's a data-intensive process, and why TACC has been essential." 

Stampede’s analysis is showing Svensen how ideology scores change over time and grounding his chief argument that both parties are arguing over how to prioritize spending.

"Nearly everything government touches revolves around spending," says Svensen. "As polarization moves up, changes in federal spending move the same way."

--Amanda Aubuchon