BUGGED OUT

摘要

The screens stopped working at 3.27 pm. Suddenly, flight controllers found themselves working blind, unable to access key information on the dozens of aircraft in the sky above them. Planes awaiting take-off were grounded and inbound aircraft diverted to other countries. More than 100 flights were cancelled and tens of thousands of passengers were affected. By taking the software out of the loop, and falling back on fail-safe procedures, flight controllers skillfully avoided disaster. But this glitch in the computer system of the UK's National Air Traffic Services on 12 December 2014 could have been far worse. Software bugs have plagued us since we started to code. They cost the global economy billions, and we spend billions more trying to get rid of them. But they're tenacious. It's common for software to be released with at least some bugs - and more are often discovered only once it is out in the wild. At best, these are a nuisance, causing an app to crash every now and then. At worst, they cause serious security and safety risks, or lead companies to lose fortunes. Some have even killed (see "A history of glitches", left). In just the past few years, for example, software errors have made Toyota, Land Rover and Ford recall more than a million cars between them for safety reasons. They have crashed spacecraft, let online shoppers buy thousands of dollars of goods for nothing, and even delayed the launch of the Apple watch. What to do? For a growing number of researchers, it's time to admit defeat. If we can't beat bugs, we need to learn to live with them - switching from damage prevention to damage limitation. Making computers more resilient to things going wrong could mean an end to computer crashes altogether- buggy code or not. "The idea here is immortal software," says Martin Rinard at the Massachusetts Institute of Technology.