Monday, April 12, 2010

Toyota Revisited: Unintended Acceleration of Judgment?

On April 5 the U. S. Secretary of Transportation announced the intention of the federal government to assess a $16 million fine against Toyota, the maximum allowable penalty in such a case. The government alleges that the car maker failed to notify the National Highway Transportation Safety Administration within five days of identifying the defect that causes unintended acceleration, as U. S. law requires. Toyota has two weeks in which to either accept or contest the fine.

About a month ago, I addressed this issue with a blog that criticized Toyota for its foot-dragging and circle-the-wagons mentality concerning the unintended acceleration incidents that have come to light in the past year or so. What I failed to do then is what many news sources are still failing to do now, which is to put this situation in a historical context, as columnist Walter Olson did in a recent issue of National Review.

Olson points out that at least part of the sudden appearance of wildly accelerating Toyotas everywhere is due to mass psychology, on the part of the public in general and the media and lawyers in particular. Once the issue hits the newsstands, drivers and their lawyers are primed to look for it, and so often what you look for you will find. Back in 1993, a similar flap blew up about suspected unintended acceleration in Audis. Despite extensive coverage, investigations, and government inquiries, no cause for the documented cases of Audi acceleration was ever found other than operator error: in other words, people were hitting the gas pedal by mistake.

This kind of thing happens all the time, often with older or less experienced drivers. Olson cites a Los Angeles Times study of twenty years' worth of Toyota unintended acceleration cases, and of those in which the driver's age could be determined, the median age was 60. Since the typical auto accident victim is in his late twenties, there is clearly an age-related factor involved.

None of this is to say that there may not be a genuine problem, or array of problems, that are causing the deaths and near-accidents that we have heard so much about recently in Toyotas. The carmaker itself has identified and gone to a lot of trouble with recalls to fix a carpet-pedal interference issue that evidently accounts for at least some of the acceleration incidents. But so far the firm has steadfastly denied that there is a problem with the electronics, in particular the software that runs the car. Two separate scientific panels have been convened by the U. S. government to look into that issue, and although they will take a while (one is set to run for over a year), we have hopes of getting to the bottom of that particular question.

The challenge of such an investigation is that software glitches, especially if there is a random hardware effect involved such as electromagnetic interference, can be extremely hard to detect or fix. In principle, computer software follows an exact mathematical deterministic path: if you set up the same initial conditions every time, you will get the same result. But when software is embedded in the roaring, buzzing, mechanically dynamic environment under the hood of a car, the goal of "same initial conditions" can be elusive. You can always do lab tests, but to the extent they are more tightly controlled, they are just as much more removed from the reality in which the accidents supposedly occurred. If such a problem is found, it will likely be the at the end of a long and intricate detective effort on the part of the investigators. And Toyota will need to cooperate fully in revealing its proprietary software and hardware, unless it wants to actively hinder the work of the safety investigators. So far, the firm's track record in this area has not been stellar, but under the present conditions of intense political and legal pressure, a vindication by an outside panel would be worth whatever bean-spilling of secrets that Toyota has been so reluctant to do up to now.

An interesting system-wide question that no one so far has brought up would be a comparison of the way automakers vet their control software with the methods used by aircraft manufacturers. Back when "fly-by-wire" software was first being adopted, there was a lot of concern on the part of pilots that when they pushed the stick, they would no longer be directly moving flaps or ailerons or whatever—instead, they would be simply sending instructions to a computer, which would interpret the pilot's action along with everything else going on and issue the orders it thought best to the control surfaces. This idea seemed to rub a lot of pilots the wrong way, and it took a while for fly-by-wire systems to be accepted. There were one or two problems with it along the way, but then as now, the bulk of airplane accidents involve at least some personal responsibility on the part of the pilot.

It's ironic, but automobile drivers as a group probably carry a lot less clout with the automakers than airline pilots do with aircraft companies. I sense that the automakers regard their customers as, if not quite sheep, then certainly uneducated in the technicalities of automotive engineering. The transition from the day when the connection between the accelerator and the carburetor (as it was way back then) was just a pull wire, to the present situation where a computer sits between your foot and the engine—this transition came silently and unheralded, possibly out of fear that people would not be comfortable with it. Such silence may have made the present situation worse as folks discover the fact for the first time, and find that they don't like it—especially those who believe their problems with unintended acceleration are the fault of the automobile, and not due to inappropriate actions of their own.

As Walter Olson pointed out, it took years for the Audi flap to die down, and we can expect the Toyota situation to linger at least that long—longer, if the investigations into software problems are as complex as I expect they will be, and if Toyota is less than fully cooperative in them. The only sure thing is that lawyers will be making money off this situation for a long time to come.

Sources: A report on the $16 million fine appeared in the Washington Post online edition at http://www.washingtonpost.com/wp-dyn/content/article/2010/04/05/AR2010040503200.html. Walter Olson's article in the Mar. 15, 2010 issue of National Review is online at http://article.nationalreview.com/427879/exorcising-toyotas-demons/walter-olson.

No comments:

Post a Comment