Researchers at UC Berkeley’s School of Information used game theory to demonstrate that governments tend to abuse existing surveillance systems to stay in power in a new study published last month.
The model displays how different levels of surveillance affect the actions of a government, as well as citizen welfare and popularity of the government. By imparting each agent in the model with a specific set of goals and incentives, researchers determined that once a surveillance system exists, a government will always have incentives to abuse it to achieve its objective of staying in power.
The model is based on theory and mathematics, and it is not modeled on any specific government, allowing it to predict the decisions of different types of governments.
Governments, which in the model have the object of remaining in power, will invariably try to conduct excessive surveillance. Either too much or too little surveillance ultimately decreases citizen welfare in a popular government, although some surveillance can protect citizens from undesired change. In an unpopular government, however, any amount of surveillance is harmful to citizens, according to the model.
Although abuse of power is a different concept from surveillance, the two are directly related in that surveillance and surveillance technology make abuses of power more effective, said visiting assistant professor Paul Laskowski, a co-author of the study.
“If you want to stop people from expressing political dissent, it helps to have technology online to censor key words,” he said. “You have to choose how corrupt to be, and surveillance technology tips that balance because … it relaxes one constraint that holds back the level of abuse.”
The model is limited in that it cannot currently be run through simulations of real governments, said postdoctoral student Thomas Maillart, a co-author of the study. Policymakers should use tools developed by academics on specific problems, said Stephen Maurer, a lecturer at the Goldman School of Public Policy and director of the Information Technology and Homeland Security Project.
From a public policy point of view, the model is too general to provide specific insight about agencies like the National Security Agency or the Internal Revenue Service, he said.
“The paper still makes a valuable contribution by reminding us that the optimal level of surveillance is almost certainly not zero,” Maurer said in an email. “This suggests that political rhetoric like ‘let’s eliminate the NSA’ is not very helpful.”
The effects of completely secret surveillance remain undetermined, but Maillart said such a system in the real world is close to impossible.
“There’s always going to be leaks,” he said. “At least one must be loyal to the people and be a whistleblower.”
Contact Madeleine Pauker at [email protected]