Does NASA study end all Toyota concerns?

A 10-month study by the space agency's engineering experts appears to close the chapter on Toyota's unintended-acceleration complaints.

But some outside experts remain a little skeptical - and they came in for sharp criticism yesterday by Transportation Secretary Ray LaHood, quoted here in a Detroit News item by David Shepardson after LaHood met with reporters during a White House round-table discussion: 

[LaHood] singled out Clarence Ditlow, head of the Center for Auto Safety, who noted the government tested just nine vehicles. He noted that shortly after the reports from the National Highway Traffic Safety Administration and NASA was released yesterday, Ditlow has been quoted criticizing them.

"[Ditlow] says they were looking for a needle in a haystack. He didn't even read the report," LaHood said. "This guy makes his money trashing DOT and trashing our safety program."

In an interview today, Ditlow said he pointed out that NASA said complaints of unintended acceleration are 1 out of 100,000 vehicles — and they only examined 9 vehicles. "The chances of finding something were slim," Ditlow said. "NASA did not give Toyota a clean bill of health and if they had more money and more time, they might have found something."

Ditlow said he respected NHTSA's safety program. "People who are on the same side, which is auto safety, can disagree about what the data shows. This time we simply disagree about what the data shows."

LaHood said the National Highway Traffic Safety Administration and NASA exhaustively reviewed the issue for 10 months in determining that electronics were not to blame for sudden acceleration incidents in Toyota vehicles.

"Give me a break," LaHood said.

NASA engineers reviewed 280,000 lines of software code in Toyotas, and bombarded Toyota vehicles with electromagnetic radiation at a Chrysler facility in Auburn Hills as part of their 10-month investigation.

"We did a very thorough report. NASA did a very thorough report, and for people to take cheap pot shots is ridiculous," LaHood said.

Computer-risk experts were among those who voiced concerns early on, after Toyota blamed the complaints on two mechanical problems: sticky gas pedals, and loose floor mats that could trap them.

It seems very likely that those factors, along with human error, caused all or virtually all of these incidents. Transportation officials plan to study whether pedal placement or other engineering choices involving "human factors" can be tweaked to minimize the chance that a driver will press on the gas when he or she is trying to hit the brake.

But one programmer reminded me why reviewing 280,000 lines of code wasn't necessarily so impressive, aside from the fact that a modern car's systems may rely on 20 million lines of computer code.  Basically, he said, failures happen for reasons that even the most careful debugging can't identify - except in hindsight. He wrote:

"I've seen bugs that only appear after 48 hours of continuous execution on a 4 CPU computer. The error required that two parts of the program execute sections of code that interact in the same small window of time."

In other words, they're close to random failures. But that doesn't mean the code wasn't to blame.