Will More Auto Safety Investigators Save More Lives?

Last Friday, the U. S. National Highway Traffic Safety Administration (NHTSA) announced that it was planning to improve the way it keeps tabs on automakers and the safety of their products.  NHTSA administrator Mark Rosekind admitted that the changes were largely inspired by his agency's failure to catch the GM ignition switch problem early enough.  The defective switches on older-model small cars such as the Saturn Ion and the Chevy Cobalt could accidentally cut power to both the engine and the vehicle's airbags, and have been identified as the cause of over 100 deaths and 200 injuries that took place before a massive recall of some 2 million vehicles last year to fix the problem. 

Rosekind acknowledged that although his staff had access to some of the data on the switch problem, they didn't understand that the switch could disable the airbags, and so in two reports issued by the agency, there are calls for improved technological expertise and more investigators in the agency's Office of Defects Investigation (ODI), which currently has about 60 full-time employees.  The report calls for increasing that number to 150, and maybe more.  Rosekind says that NHTSA has already undergone a "culture change" which relies less on the automakers to do self-policing and more on the agency to hold automakers accountable for producing needed data, and also for the agency to do more on-the-ground accident investigation itself.

The question here, as in any change of resources to address a problem with engineering implications, is:  will it do any good?  And will the good that might result be worth the resources expended?

As bureaucracies go, the NHTSA's ODI is pretty small compared to the total number of non-military Federal goverment employees—about 2.7 million people in 2014.  A few million dollars will give the NHTSA all they're asking for and more.  If you asked any of the relatives and friends of those 100 people who died because of defective GM ignition switches about this proposal, they would say the expenditure is worth it if it will save even one life in the future.  And here we get to an issue that tends to come up a lot in discussions of engineering ethics:  the monetary worth of a human life.

I'm not going to waste time playing with dollar/life quotients, because doing that means you have slipped into the never-never land of utilitarianism.  The philosophical approach to happiness called utilitarianism is conveniently (if not entirely accurately) summarized by the phrase "the greatest good for the greatest number."  It has a sneaky kind of appeal to engineering types, because it holds out the (false) promise of reducing complex morally-freighted issues to a straightforward process of mathematical optimization. 

I reject utilitarianism for a number of reasons.  While it has limited usefulness in extreme cases—it tells you, for example, that placing the value of a human life at zero is unwise—the thing usually falls apart well before you can actually sit down with a calculator and do a mathematical operation that will tell you which of several alternatives in an ethical problem is the right one.  It falls apart for me because the very act of putting a dollar value on a human life makes one no different in principle from the slave traders who did exactly that for profit.  It is simply a thing not to be done.

Well, if we can't make the situation into an optimization problem in mathematics, how should we decide if giving the NHTSA more bucks to hire more inspectors is a good idea?  To begin with, a little background might be helpful.

The history of U. S. auto safety has been one of gradual but cumulatively remarkable improvement, starting with the adoption of safety glass in the 1920s by Ford Motor Company, and progressing to seat belts, air bags, and many other safety-related technologies, only some of which were federally mandated.  The result of these improvements has been a pretty steady decline in deaths due to automobile accidents from a high of nearly 3 per 10,000 people per year in the 1930s to about 1 per 10,000 today.  So to begin with, things are getting steadily better, and if self-driving cars realize their early promise, the operator errors and drunk-driving accidents that cause most crashes today may largely go away too.  In the perspective of these numbers, while the 100 or so people who died in the GM switch accidents died needlessly, they were only 0.3% of the 30,000 or so people who died in auto accidents in each of the last few years. 

Ironically, it looks like the public is less upset when a single bad guy can be identified as the cause of a fatal crash, than when the immediate cause is something mechanical like a weak ignition switch spring.  This has nothing to do with mathematical optimizing of resources, and everything to do with how people perceive risk and danger.  You get in your car and tool out on the highway.  Maybe there's a drunk driver out there, but you think you can do something about that.  You can see him weaving around and steer out of his way, for instance.  But some un-thought-of hidden mechanical defect like a defective ignition switch—there's nothing you can do about that.  It's a lot scarier, and the news media know that, and the automakers know that.  Which is why they have been highly motivated to avoid causes of safety recalls, and deeply regret the ones that slip through their own bureaucracies (as the GM ignition switch problem did) and cause only a few verifiable fatalities.

So what is the bottom line here?  I think the bottom line is, there is no bottom line.  This is not an accounting problem.  Maybe three times as many investigators at the NHTSA will make the automakers three times as vigilant to catch the next major safety flaw before it gets built into millions of cars.  While that would be nice, I doubt it will happen, and I don't know of any objective way to tell whether it's happened after it happens, if in fact it does happen.  But the public has read in headlines that the feds are doing something about the problem, and that perception itself, rather than any meaningful actions that may happen afterwards, may have been the main point in this exercise.

Sources:  The Associated Press article on the NHTSA reports was carried by a number of outlets, including the Los Angeles Times under the headline "U. S. auto safety agency admits flaws, starts reform after GM case" at http://www.latimes.com/business/la-fi-hy-nhtsa-gm-20150605-story.html.  I referred to the Wall Street Journal for the statistic on the current (2014) number of Federal employees at http://blogs.wsj.com/economics/2014/11/07/the-federal-government-now-employs-the-fewest-people-since-1966/.  I also referred to the press release of June 5 at the NHTSA website, http://www.nhtsa.gov/About+NHTSA/Press+Releases/2015/nhtsa-forming-new-safety-teams, and the Wikipedia articles "List of motor vehicle deaths in U. S. by year" and "Windshield."  I most recently blogged on the GM ignition switch issue on April 7, 2014 at http://engineeringethicsblog.blogspot.com/2014/04/the-gm-ignition-switch-recall-too.html.

The GM Ignition Switch Recall: Too Little Too Late?

How much stuff do you carry on your keyring?  Besides keys, I mean.  Some minimalists like my wife carry car keys separately from other keys, with nothing attached except maybe a small plastic tag to make it easier to find in her purse.  Other people, many of whom are younger, may carry a whole bundle of stuff on their keyrings:  those little barcode cards that give you discounts at retailers, miniature plastic poodles, handcrafted bits of knitted yarn, and I don't know what all.  But it probably never occurred to you to think that a heavy keyring could be hazardous to your health. 

Brooke Melton probably wasn't thinking of her keyring one rainy March night in 2010.  She was driving her 2005 GM Cobalt when the ignition switch suddenly moved from "run" to "accessory."  This had the unfortunate effects of killing the engine, disabling the power steering, and turning off the airbags.  The sudden loss of power caused Melton to cross into oncoming traffic.  The Cobalt crashed into another car at 58 mph and wound up in a creek, killing Melton and starting a chain of events that revealed the true cause of more than a dozen similar crashes going back more than half a decade.

As long ago as 2001, engineers at General Motors knew that a certain model of ignition switch assembly that was later used on a number of models had a problem.  The mechanical design of an ignition switch is a compromise, as are so many things in engineering.  Most mechanical ignition switches use a device called a detent, which divides the continuous rotation of the switch that would occur without the detent into a small number of discrete positions, typically four:  "off", "accessory", "run" and "start."  If the detent provides too much resistance, the switch will be hard to turn and might eventually wear so much that it would fail to work at all.  But if the torque (twisting motion) required to move the switch is too small, you take the risk that unbalanced forces resulting from heavy stuff on a keyring, for instance, may spontaneously make the switch turn from one position to the other.  This is apparently what happened to Brooke Melton and the 12 or more other drivers who died in ignition-failure accidents in GM cars having the suspect assembly.

At this remove, it is obvious what GM should have done.  The guilty part, No. 10392423, should have been redesigned with a more forceful switch detent plunger—a 57-cent piece that consists of a rounded plastic cylinder backed by a coil spring.  It is the force exerted by this plunger that sets the amount of torque needed to turn the ignition key from one position to the next.  Changing the spring fixes the problem by increasing the torque needed to turn the key from "run" to "accessory."  Then, the ignition assembly part number, or some documentation somewhere, should have been changed to reflect the fact that the new part was substantially different.  And GM should have recalled however many cars they had sold with the defective ignition switch and replaced them free of charge. 

If this had been done early, before too many cars had been sold with the defective ignition, it would have cost something, but the earlier such things are dealt with properly the less expensive they are.  But at the time, a few other things were happening at GM that provided distractions, namely, bankruptcy.  So matters drifted along, and at some point, Delphi (the company that makes the switch in Mexico for GM) changed the plunger to fix the problem.  There is contradictory information as to whether Ray DiGiorgio, a GM engineer, approved a design change in April of 2006 making this fix.  He has testified that he did not, but a Congressional committee claims it has documentation showing that he did.  Whatever was done in 2006, it had no effect on the thousands, if not millions, of cars already on the road at that time with defective switches.

After Brooke Melton's death, her parents decided to sue GM.  Their lawyer, Lance Cooper, hired a consulting materials engineer named Mark Hood to look into why the ignition turned off—an event that was documented by the car's black box.  After plowing through numerous Cobalts of various vintages in junkyards, he discovered that the critical plunger had been silently altered around 2006 or 2007.  Switches made before then took less torque to turn off than the newer switches.  Armed with these facts, Cooper took depositions from GE engineers and reached a settlement with the firm.  But the publicity surrounding the lawsuit attracted enough attention that others with similar crash incidents on their hands began looking into the matter.  And just last week, GM CEO Mary Barra testified before Congress about the whole thing.

To her credit, Barra took action to issue massive recalls, affecting some six million cars, on this and other problems within weeks after learning about them when she took the helm of GM in January of this year.  But these recalls are too late for Melton and at least a dozen others who died in ignition-related crashes of GM cars.  Although the investigations are continuing, it appears that at least one GM engineer may have lied under oath about the matter. 

This story has heroes and villains, although most engineering ethics cases are not black and white, including this one.  Consulting engineer Hood and GM CEO Barra appear to have done the right things with what they learned.  Investigations may prove that the GM engineers involved with the faulty ignition switch may have made the best decisions they could have, based on the information they had available.  No automaker can afford to do as much prototype testing as they would like.  It took making and selling thousands of cars to reveal that a few people with exceptionally heavy keyrings could end up getting killed by a switch that took just a little less torque than usual. 

But the truly blameworthy actions happened after GM began receiving reports of such ignition-caused crashes.  One fatal accident due to a defect that can occur under certain conditions should be looked into, and if necessary, a recall—not just a service advisory, which GM issued about the matter in 2005—should be issued. 

This situation shows that corporations, like people,  have good times and some not so good times.  GM's financial troubles possibly dissuaded decision-makers from issuing the massive recall that would have been needed to fix the ignition defect early, before more defective cars were sold.  But the result has been an even larger and more costly recall later.  Let's hope GM can fix all of the defective ignitions soon and move on, a sadder but wiser organization.

Sources:  I referred to the Wikipedia article "2014 General Motors recall," as well as the following online news articles.  CNN reported on the problem at http://money.cnn.com/2014/04/02/news/companies/gm-recall-part/.  Engineer Mark Hood's detective work is described at http://www.bendbulletin.com/home/1949311-151/a-florida-engineer-cracked-gms-ignition-flaw#.  Also, a Reuters article at http://www.reuters.com/article/2014/03/31/us-gm-recall-congress-idUSBREA2T0HO20140331 correctly describes the critical component as a "detent" plunger (it has been elsewhere described incorrectly as an "indent" plunger).  And National Public Radio published a helpful timeline of the issue at

http://www.npr.org/2014/03/31/297158876/timeline-a-history-of-gms-ignition-switch-defect.