machine vs. man

NASA released a report this week that is a social psychologist's and complexity theorist's wet dream.

The engineers at the geekiest of federal agencies were called on to weigh in on Toyota's brakes mystery of 2009. After recalling cars suspected of having sticky accelerators and obstructing floor-mats, Toyota faced a final round of complaints claiming an electronic glitch - "sudden unintended acceleration" - which sped cars out of control and some drivers to their death. Toyota stood by their computerized driving system and blamed human error for the "runaway" cars. Consumer groups and victims (many now taking their claims to court) continue to fault Toyota's computerized cars.

NASA's engineers have now sided with Toyota's. After sifting through 280,000 lines of software code, examining 78 incidents (58 which have "black boxes" with pre-crash info), and subjecting 9 cars to extra tests - in a $1.5 million study that took 10 months - NASA concluded that any sudden acceleration happened because humans slammed on the accelerator when they meant to slam on the brakes.

Consumers are having none of it. They claim that NASA tested too few cars, especially when claims are that only 1 in 100,000 cars are going to get glitchy, testing 9 cars is bound to show negative results. And then there's the testimony of drivers like Rhonda Smith: "I looked at my feet - and I know it wasn't the floor mat - and they were firmly planted on the brake. I still stand by the truth that I told. I do firmly believe that there is a vehicle defect that they've just not found."

With two such stark - and decidedly rational and plausible - points of view, you gotta wonder what's going on here. Who's right and who's making stuff up?

Social psychologists and complexity theorists might tell you that is an answer we will never have. There are just too many things that could be going on.

• Complexity and the machine. 280,000 lines of code is a lot of code. (By comparison, the space shuttle has 420,000 lines of code.) Even though NASA's own engineers combed Toyota's code through for bugs and found none, that doesn't mean that one didn't go missing or that, each time a new car's computer is produced a mini-mutation gets picked up. (If you've ever returned a faulty Dell or iPhone, this should be obvious.)
• Complexity, the machine and the perfect storm. Even if every car came off the line with pristine code, the car's system hasn't been tested in every possible set of circumstances. Unpredictable catastrophic things occur when a new set of factors occur at once, like a "perfect storm." The code may work in all weather conditions, at all speeds, with any fuel level, when any other individual electronic feature is operating, and for all maneuvers - but may not work in severe humidity, at 73 mph, with 10% of the tank filled, while the window shield wipers are on and you're signaling left and slowing down to make a hard left turn. That might be the time the code runs into its one error.
• Complexity and the human brain. What's hardest to imagine is that Rhonda from above, and other drivers like her, could so steadfastly swear that their feet were on the brakes when, as Toyota and NASA claim, they were actually slamming the gas pedal. But brains, especially in moments of panic, have a way of getting glitchy themselves. We are all capable of seeing things - or missing things that are right in front of our faces. Cops see guns that don't exist and life guards fail to see bodies lying at the bottom of pools. It's not a vision problem; our minds have an uncanny power to see what they believe exists, not what really does.
• Complexity and human society. One of the things that helps us see things that aren't real is hearing about it in the news. I haven't seen any reports, but it wouldn't be surprising if the reports of "sudden unintended acceleration" shot up once the first new stories appeared. Just as imagined illnesses can proliferate through a school, so can imagined phenomena work their way through society. The complexity of human relations could also have something do with NASA's report; groups have a way of steering themselves to false consensus through "group think." NASA is not immune; one of the classic case studies on group think comes from NASA itself, when its engineers okayed the launch of the space shuttle in 1986.