Crashes happen. Researchers at Stanford University and the University of California at Berkeley are designing systems that recover rapidly when they do — the recovery-oriented computing (ROC) approach.

ROC design principles are based on speedy recovery, better tools to pinpoint fault sources, an “undo” function, test errors to permit evaluation of system behavior and assist in operator training, and selective rebooting to minimize loss of data.

“If the industry continues traveling blindly down the current path, computers in 2023 may end up another 10,000 times faster yet no more dependable than today’s machines. But with dependability-enhancing software tools and benchmarks to inspire us, computing may one day become as reliable as users expect it to be.”