The Times and General Motors: What went wrong?

In 2014 General Motors Company (GM) recalled more than 2.6 million automobiles to replace a defective ignition switch that had been implicated in more than a dozen deaths. Despite early problems with the switch, it took GM almost 11 years to initiate the recall. The recall announcement led to a firestorm of media criticism. Much of that criticism, however, was badly distorted. I concentrate on The New York Times’ coverage because it is a trusted news source, and because it devoted substantial resources to the story. Problems with The Times’ coverage were likely magnified in less reliable media outlets. I describe what we know about the switch problem and General Motors’ attempts to solve it, and try to explain the forces—mainly psychological—that warped The Times reporting. These forces—love of a story with heroes, villains, and a moral dimension; willingness to view group human activity on the model of individual human activity; overly simple theories of human motivation; a tendency to attribute beliefs and desires to organizations—are not unique to The Times or this particular story, and can thus be expected to affect media reports of other corporate decision-making. Subjects: Journalism; Journalism & Professional Media; Media Psychology; Media Studies; Newspapers; Newswriting and Reporting


Introduction
On 7 February 2014, General Motors Company (GM) announced a safety recall that eventually covered 2.6 million vehicles from model-years 2003through 2007(Valukas, 2014. The recall was to replace a defective ignition switch 1 implicated in more than a dozen deaths. 2 (The recall mainly involved Chevrolet's Cobalt model, but also included Saturn Ions, Pontiac G5s, and other GM models; I'll refer to them all as "Cobalts.") 3 Despite early awareness of problems with the switch, and of the fatalities, it took GM almost 11 years to initiate the recall. 4

PUBLIC INTEREST STATEMENT
The New York Times' coverage of General Motors' recall of defective ignition switches in over 2.6 million cars was badly distorted. The reporters ignored the only detailed independent study of the problem, instead filling the vacuum with stories that reflected their own preconceptions. These biases are not unique to The Times or the particular story, but can be expected to warp media coverage of much organizational decisionmaking, especially where complex technical issues are involved.

The Valukas report
About a month after the recall announcement, GM directed the Chicago-based law firm of Jenner & Block to prepare "an unvarnished account" of why it took so long to recall the Cobalts (Valukas, 2014, p. 5). Jenner & Block's effort was headed by Anton R. Valukas, a former federal prosecutor. On 29 May 2014 Mr. Valukas delivered a 276-page report (plus 49 pages of appendices) to GM's Board of Directors, who made the report public on 4 June.
The Valukas report is likely to be the best account we'll get of what went wrong at GM. Valukas's team had "unfettered access," 7 reviewing millions of internal documents 8 and interviewing over 230 people (some more than once), including all the important players at GM. 9 However, the Valukas team did not commission any new engineering studies, so what we know about the crashes is what GM knew, as ferreted out by Valukas and his team. As Valukas reports, the engineers at GM (eventually) came up with a plausible theory about the role the defective switch played in the crashes, a theory that explained most of the accidents they studied.
One can legitimately be suspicious of the Valukas report. It was commissioned by GM's management, was put together quickly, and ended up absolving GM's top officers. (It did, however, lead to the firing of some 15 GM employees, not all of them low-level .) But the GM engineers' theory presented in the Valukas Report has not been seriously questioned, and is thus the only theory currently on offer; anyone with a different theory will have to produce a more compelling explanation. Moreover, the Report presents a plausible narrative of organizational disfunction that fits in with other studies. In what follows, I'll accept the account presented in the Valukas report.

The low-torque switch
The ignition switches installed in the Cobalt from its introduction in model year (MY) 2003 into MY 2007 allowed the key to be turned on and off with so little force (torque) that the key could move from the on (RUN) position to the accessory (ACC) position when lightly jarred, as could happen if the key fob was brushed by the driver's knee (Valukas, 2014, pp. 59-61) or the car hit a curb or pothole. 10 When the key moved to ACC, the engine shut down and the steering and brakes, denied power assist, became heavy. More critically, after approximately 0.15 s power for the airbag sensor was lost, so that the airbags would not deploy in a crash. 11 (Shutting off power to the sensor when the key was in the ACC position was a deliberate design choice, to prevent the airbags inflating while the car was stationary. 12 ) It's not clear why the switch, manufactured by Delphi Mechatronics, had such low torque, which was outside GM's established parameters and contrary to the original switch specification. 13 Raymond DeGiorgio, the GM engineer mainly responsible for the switch, knew about the low-torque condition but approved the switch, apparently because he feared that changing the switch would compromise its electrical performance. 14 (The switch had so many electrical problems in development and early production 15 that DeGiorgio signed an instruction to Delphi not to increase the rotational torque as "Ray (tired of the switch from hell) DeGiorgio"). 16 The Valukas teams were unable, after interviewing hundreds of witnesses, to find any GM employee besides DeGiorgio who knew before 2013 that the switch failed to meet the original torque specification. 17

Moving stalls
Complaints about stalling caused by the key turning to ACC while the car was moving began to reach GM soon after the first Cobalts hit the streets, but DeGiorgio and the other GM engineers who worked on the switch (whom I'll call the "switch engineers") apparently did not know that when the key was in the ACC position the airbags would not deploy. 18 They believed, however, that a car without power remains controllable, so they considered the ignition switch problem a "customer convenience" rather than a safety issue. 19 Consequently, they felt no pressure to solve the problem quickly. 20 (GM personnel met several times in 2004 and 2005 with NHTSA officials about the general issue of engine stalling, and concluded that the Agency did not see it as a per se safety issue; there seems to have been no discussion of the Cobalt's particular stalling problem. 21 ) In response to GM's worries about the volume of customer complaints, DeGiorgio proposed replacing the switch with a higher torque version, 22 but in September 2005 an upper level GM engineering committee rejected his suggestion, apparently for cost considerations; the committee still considered the issue merely one of customer convenience. 23 Notwithstanding the committee's rejection of his proposal, in April 2006 DeGiorgio approved a modified switch with higher torque, which began to be installed during MY 2007. The stalling problem was solved. 24

The airbag problem
In 2006 a different group of GM engineers (whom I'll call the "airbag engineers") became aware of a fatal Cobalt crash early that year in which the car's airbags did not deploy and the car's "black box" (technically, the "sensing diagnostic module" or "SDM") indicated that the ignition switch was in the ACC position at the moment of impact. 25 These engineers knew that turning the switch to ACC would disable the airbags, but they thought that the airbag did not deploy because the impact had been to the car's right front corner rather than head-on. (A picture of the wreck confirms the corner impact (Jensen, 2014c).) This may be an appropriate moment to introduce a consideration that figured only peripherally in the Valukas report and was never mentioned in The Times' coverage: Airbags don't always work. A 2009 study concluded that airbags failed to deploy in 8% of fatal front-end crashes; (Braver, McCartt, Sherwood, Fraade-Blanar, & Scerbo, 2009, p. 6). Earlier studies had put the rate as high as 18%. 26 (None of the studied non-deployments involved a Cobalt.) Most of these failures could be explained by specific crash characteristics-for example, a "complete underride" with a large truck where the vehicle hood is not contacted, or an impact that is mainly to the side 27 -but the deployment failures in up to 2% of fatal crashes could not be explained.

Outsiders pinpoint the switch
In 2007 two outsiders connected a Cobalt's airbag non-deployment to the ignition switch: A Wisconsin state trooper at the site of a fatal crash reported that the ignition key was in the ACC position and surmised that the airbags had failed to deploy for that reason, (Valukas, 2014, pp. 116-118). and an Indiana University Transportation Research Center report on the crash reached the same conclusion. 28 The outsiders' reports were in GM's files and the public record, 29 but the airbag engineers only became aware of the Indiana report in 2012, at which time they discounted it because it did not square with their other information-some of it, alas, misinformation. 30 The outsiders were able to connect the dots because they had fewer dots to connect. 31 There's another reason, not discussed in the Valukas report, why the airbag engineers might have discounted information provided by outsiders: Many of these outsiders were plaintiffs or their lawyers, who had a great deal to gain if an accident's cause could be pinned on a GM design defect. This is not new territory: The MY1982-87 Audi 5000 model and various MY2000-10 Toyotas were claimed by persons involved in accidents to have had "sudden acceleration"-"unintended, unexpected, high-power accelerations from a stationary position or a very low initial speed accompanied by an apparent loss of braking effectiveness." (National Highway Traffic Safety Administration, [NHTSA], 1989). Audi sales suffered, and Toyota initiated massive recalls and a $1.2 billion settlement, as a result of these claims. The National Highway Traffic Safety Administration later concluded that most of the sudden-acceleration crashes resulted from driver error ("pedal application error" being the regulatory euphemism). 32 Outside information frequently fails to be objective.

Attention turns to the switch
The investigation into the Cobalt airbag problems did not focus on the ignition switch until mid-2009 when an airbag engineer who had been keeping a spreadsheet of the incidents noticed that SDM data indicated that the switch was in the ACC position in a number of crashes where the airbag did not deploy. (Not all Cobalts in these crashes had an SDM, and in those that did the SDM showed the switch to be in RUN about half the time (Valukas, 2014, pp. 9, 129, 135, 156, 206).) The airbag engineers, still unaware of the stalling problem the switch engineers had dealt with, began to focus on the electrical system as the possible cause of the ACC readings. They had noticed that many of the airbag non-deployments involved off-road crashes, and soon came up with a theory-"contact bounce"-that sought the cause in the internals of the switch rather than the key position. The theory was that jarring the ignition switch-as when you jump a curb or hit other offroad objects-could "open up" the switch, so that the signaling mechanism for the airbags would report the key as in the ACC position and shut off power to the airbag system. 33 The engineers conducted "abusive and teeth-chattering tests" in 2009 "in which the car was driven through steep ditches and deep potholes" without ever getting the switch to "open up." 34 The Report doesn't say, but presumably the engineers were testing Cobalts built after the ignition switch change; otherwise the bouncing would likely have caused the key to rotate to ACC. The airbag engineers were still unaware that the switch had been changed during MY 2007.

The switch is seemingly absolved
Sometime later in 2009 the airbag engineers realized that non-deployments had ceased for MY 2008 and later Cobalts, 35 which caused them to rule out the switch as a potential cause of the non-deployments. There were two reasons for this mistake: First, while the ignition switch was changed during MY 2007 to solve the low-torque problem, 36 DeGiorgio had not informed anyone else at GM, and, contrary to GM policy, had not changed the part number. 37 Worse, on several occasions DeGiorgio told airbag engineers that the switch had not been changed (except for an irrelevant change to the anti-theft system). 38 Consequently, the airbag engineers, believing the switch to be unchanged, ruled out the switch as a possible cause of the non-deployments. 39 We can only speculate as to why DeGiorgio did not inform others of the change to the switch, did not change the part number, and misinformed the airbag engineers. DeGiorgio says he does not remember many of these events. 40 Perhaps these actions were simply DeGiorgio's attempt to cover up his disregarding the engineering committee's decision not to change the switch.
When DeGiorgio ordered the switch change, he may not have been aware of the connection to the airbag system. But it seems likely that when the airbag engineers asked DeGiorgio if the switch had been changed, they would have informed him of the switch's suspected connection to the airbag non-deployments. Under these circumstances, DeGiorgio's statement that the switch had not been changed put many lives in danger.

The light dawns
In March 2012 airbag engineers examining a crashed Cobalt at a junkyard noticed that the ignition switch turned extraordinarily easily. 41 The engineers had not brought any tools with them to measure the torque, but using a fish scale they purchased from a nearby bait and tackle shop (you can't make this stuff up), they measured the torque for a number of Cobalt ignition switches in the yard. The torque on many switches was so low that they concluded that the key could turn to ACC if the car hit a pothole. 42 The next day one of the engineers, John Dolan, searched the Cobalt warranty database and discovered the numerous customer complaints about the ignition switch turning to ACC. 43 For the first time, the airbag engineers discovered what the switch engineers had known for years-that the Cobalt switch could be turned from RUN to ACC with minimal force. Dolan immediately elevated his concerns to more senior management. 44 Two months later, in May 2012, the airbag engineers revisited the junkyard and tested some 40 Cobalts. 45 (This time they brought a torque wrench.) They found that earlier Cobalts required lower torque, but this was also true of some MY 2007 and MY 2008 models. 46 (It is not explained why some 2008 models, all of which should have had the redesigned switch, had low torque.) But the airbag engineers still struggled to put the pieces together. In spring or summer 2012, DeGiorgio and his supervisor again stated that there had been no changes in the switch that would affect the torque. 47 And there remained the still unexplained fact that in about half of the non-deployments the SDM showed the key to be in the RUN position. 48 In April 2013, plaintiff's attorneys took apart pre-and post-2007 switches and showed GM's lawyers just how they had been changed: A plunger in the later switches was longer by a bit over a millimeter (about 1/25th of an inch), just enough to significantly increase the torque. 49 But it was another six months before Delphi could confirm that in 2006 DeGiorgio had approved the change to the plunger.

The double-impact theory
Finally, in November 2013, shortly before GM initiated the formal recall process, the airbag engineers came up with a theory that explained most of the airbag failures: If the sudden loss of power didn't cause the crash-that is, the car remained controllable-then something that happened between the driver's losing control and the crash must have caused the airbags not to deploy. The solution was a "double-impact" theory (my term). The engineers theorized that in an off-road crash where a vehicle with the defective switch hits a number of objects, such as when it goes over a curb before hitting a tree, the initial impact with the curb could jar the key, or make the driver's knee jar the key, causing it to turn to the ACC position, disabling the airbags before the second impact with the tree. 50 The double-impact theory explains most of the non-deployments that the engineers considered. The Valukas Report only describes 13 cases of non-deploying airbags (not all fatal), presumably because they were ones with written evaluations from GM's counsel. It may not be a representative sample, but for what it's worth, 10 of the cases involved off-road crashes with multiple impacts-the car jumped a curb or hit some small bushes before hitting a larger object, such as a tree. But in two cases there seems to have been only one impact. In the first, a car was hit from the side in an intersection, in the other the car rear-ended a tractor-trailer; these are both situations where the airbag might not deploy for other reasons. (The report does not give enough information about the one remaining case to determine if there was a second impact.) The theory also attempted to explain why the SDM often showed the switch to be in the RUN position, but the explanation is confusing and no clear answer emerges. 51 Perhaps the switch really was in RUN position and the airbags did not deploy for other reasons, or perhaps there is something important that we still don't know about the crashes.
So GM's final theory for the airbag non-deployments is this: If the defective switch goes to the ACC position in normal driving, the car will remain controllable, so the car will not go careening off the road. But if the car goes off the road under full power for any of the myriad reasons that drivers lose control, it can encounter small objects or rough terrain that cause the switch to jump to the ACC position, disabling the airbags just before the car strikes a large object.

The Times ignores the controllability issue
The Times' reporters seem never to have fully understood the double-impact theory, which, as I've said, is the only theory currently in the lists. First, the reporters did not report that a loss of power should not make a vehicle uncontrollable. Of course, the Valukas report was not available to The Times for several months after the recall announcement, but GM had claimed as early as 24 February 2014 in a notice to the NHTSA that "the car's steering and braking systems remained operational even after a loss of engine power." The Times duly reported the notice but didn't mention GM's claim about controllability (Jensen, 2014b). In covering GM President Mary Barra's testimony before Congress, The Times quoted a GM source that "When [the ignition switch turns off] the Cobalt is still controllable," . but never mentioned this claim again. Nor was the claim mentioned in The Times' coverage of the Valukas report or later. (The day The Times reported on the Valukas Report, the Times' website carried an Associated Press story which advised that "When the engine cuts off, in most cases you'll lose power steering, making it hard to turn the wheel, and power brakes will follow soon after. You can still steer and stop but it requires much more force." (Associated Press, 2014b).) The Times' coverage of particular crashes often assumed that the loss of power caused the cars to run off the road.
Were the cars controllable when power was lost? GM's engineers believed the cars were controllable, and their claim is backed by other facts: • Drivers lose power all the time-for example, when they run out of gas-without the car's becoming uncontrollable.
• As complaints about the balky switch flooded in to GM dealerships, they seemed to be about general annoyance, plus the dangers of suddenly being stopped in the middle of a busy highway; there does not seem to be a record of people complaining about loss of control.
• GM personnel (including a former GM president) had deliberately or inadvertently replicated the stalls without anyone reporting a loss of control.
• Following the recall, Consumer Reports took a GM vehicle from their test fleet to demonstrate what to do when a car stalls. They wrote that "the car will lose power steering and soon power brakes. But you can still guide and stop the car, even if it takes longer and requires much more effort." (Consumer Reports, 2014).
On the other hand, a fair number of the Cobalt crashes seem to have involved younger drivers, intoxicated drivers, or elderly women, for all of whom the heavier steering and braking might have been more problematic. Perhaps these drivers did not have the physical strength to control the car, or were more prone to panic (Associated Press, 2014a). But while loss of power makes a car harder to steer, the difficulty is the greatest at lower (and therefore less dangerous) speeds. Moreover, the additional effort may be less for a light car like the Cobalt than for a full-sized sedan. Finally, in a crisis, people are often capable of extreme strength. Of course, these last thoughts are all surmises, but no more so than The Times' surmise that the loss of power preceded the loss of control. At the least, The Times should have solicited some expert and unbiased opinion on the issue. The failure to do so, I suspect, was due to The Times reporters simply missing the issue.
The failure to appreciate that a car remains controllable after it loses power led to a more serious flaw in The Times' coverage: The assumption throughout that the faulty ignition switch caused the crashes, rather than making the crash, which would have happened in any event, much worse by preventing the airbags from deploying.
Thus, in reporting on an accident where the airbags failed to deploy and a passenger was killed, The Times stated in the first paragraph that the NHTSA asked GM Why did the [Cobalt] he was traveling in, along a rural Texas road, suddenly swerve into a tree?" Why did the air bags fail? General Motors told federal authorities that it could not provide answers.
But only a month earlier, a G.M. engineer had concluded in an internal evaluation that the [Cobalt] had most likely lost power, disabling its air bags, according to a subsequent internal investigation commissioned by G.M. .
The implication is that the GM engineer agreed that the loss of power caused the car to swerve off the road, but, as we'll see, this is not what the engineer concluded.
The same article states that "Only recently did Ms. Anderson [the driver, who survived], who pleaded guilty to criminally negligent homicide after the accident because she had a trace of Xanax in her system, learn that she was not to blame." 52 This sentence only makes sense if one assumes that the car ran off the road because it was uncontrollable, not because the driver lost control.
Another Times article (published before the release of the Valukas Report) on the same accident had also absolved Ms. Anderson: "It's torn me up," Ms. Anderson said of the death of Gene Mikale Erickson. "I've always wondered, was it really my fault?" Last week, she learned it was not. .
The earlier article went on to say that: Despite the earlier determination by the [GM] engineer, Manuel Peace, that the engine's shutting off had most likely been the reason for the crash, G.M., in its response to regulators, said there may not have been "sufficient reliable information to accurately assess the cause" of the incident.
But according to the Valukas report, Mr. Peace only concluded that the airbags did not deploy because power was lost. "Peace, however, had not determined precisely how the vehicle lost power." (Valukas, 2014, p. 125). That is, Peace had not connected the airbag non-deployment to the faulty switch, and he certainly hadn't said that a loss of power caused the crash.
Why did The Times persist in concluding that the ignition switch caused the crashes? One possibility is that it made a better story. Unless the passenger is an Albert Camus or a Jayne Mansfield, a fatal car crash isn't very newsworthy. Nor is a driver who loses control of his or her car a particularly sympathetic subject-and perhaps less sympathetic if it turns out, as it did for Ms. Anderson, her unfortunate passenger, and many other victims, that they were not wearing their seat belts. But all of that changes if the car company caused the crash. The best stories pit good against evil, the innocent driver against the guilty car company, and here The Times succumbed to the pull of that story. Of course, the defective switch might have caused Ms. Anderson to lose control of her car, but there is no direct evidence for that, and the probabilities are against it.

The Times misses the double-impact theory
The Times' reporters also do not seem to have understood that the airbag non-deployments required a second impact. In September 2014, months after the Valukas report appeared, The Times carried long stories on two crashes that didn't fit the double-impact theory: In the first, a Cobalt rear-ended a stopped tractor-trailer on an icy road (Stout, 2014a); in the second, a car coming from the opposite direction swerved across the center line and crashed head-on into a Cobalt, whose airbags did not deploy (Stout, 2014b). (It's not clear if the airbags deployed in the first case; the driver's compartment was destroyed by fire. Two eyewitnesses who pulled the driver out of the car moments before it erupted in flames said they saw no airbags. One suspects, however, that they were preoccupied with getting the driver out of the car, and in any case their testimony came from lawyers for the driver's parents.) In each of these cases there was only one impact (and there also does not seem to have been a loss of control, other than that caused by an icy road). It is of course possible that in the last moments before the crash the driver contacted the ignition key with her knee, turning it off. Or perhaps severe braking was enough to turn the key. But these are just-so stories. At the very least, The Times should have informed its readers that the accidents did not fit GM's theory. Again, I suspect that The Times' reporters didn't fully understand the double-impact theory.

The Times misses the knowledge issue
Hand-in-hand with The Times' failures to realize that a car remains controllable after power is lost and that airbag non-deployment requires a second impact-or at least that GM's engineers believed this to be so-was the failure to report that the switch engineers (as opposed to the airbag engineers) did not appreciate that the switch's turning to ACC would disable the airbags.
The Times' failure to distinguish between what the switch engineers knew and what the airbag engineers knew may have been understandable before the Valukas report appeared, but that report was quite clear that the switch engineers had failed to understand the connection between the ignition switch and the airbag system: "They did not know how their own vehicle had been designed." (Valukas, 2014, p. 95). The Valukas report makes this point in at least six different places. But The Times' coverage of the Valukas report concentrated on the report's assignment of responsibility to GM, and not on why the GM engineers failed to find the problem. Indeed, The Times coverage of the report compounded the problem by referring to "a safety problem that was detected even before the first of the cars came on the market." . As the report made clear, the problem that was known about early on (though not "before the first of the cars came on the market") was the stalling, which was not considered a safety issue; the connection to the airbag nondeployments only emerged years later.
Three months after the Valukas report appeared, The Times was still writing that "General Motors has acknowledged that some of its engineers knew about the problem for more than a decade before the company disclosed it publicly," (Stout, 2014b). which suggests that there were GM engineers who knew both that the switch could turn to ACC with light force and that this would prevent the airbags from deploying.
Another story written well after the Valukas report appeared similarly jumbles up what the switch engineers knew and what the airbag engineers knew: "When the switch began failing because the ignition key could be inadvertently bumped, shutting off the engine and disabling airbags, Mr.
DeGiorgio tried to have it replaced with a newer part" (emphasis added) (Abrams, 2014). The phrasing suggests that DeGiorgio knew about the airbag connection, contrary to the Valukas report.
The failure to understand the difference between what the switch engineers and the airbag engineers knew is partly the result of the way we, including the media, often talk about corporations: We imagine the corporation as a kind of person that can know things and make decisions. But of course, only individuals have these abilities. Where the individuals work for a corporation, it's natural to ascribe their knowledge and decisions to the corporation; it's usually a harmless shorthand. But occasionally it gets us into trouble, as when we assume that General Motors could somehow combine the bits of knowledge of its various employees; some GM employees knew about the stalling problem while others knew about the airbag non-deployments, but it took years before any GM employee put those facts together. The Valukas report is largely the story of why it took so long for anyone to make the connection.

Cases in litigation
Another problem with The Times' coverage is a failure or unwillingness-remarkable for business reporters-to understand a basic asymmetry in civil litigation: A plaintiff seldom has anything to lose by a clumsy statement to reporters, whereas such a misstep can be fatal for a defendant. Plaintiffs and their lawyers are therefore happy to give their version of events to the press, while defendants are usually instructed by their lawyers not to talk. The result is that the press usually gives more favorable coverage to plaintiffs than defendants. The media cannot change the basic situation, but in reporting on the switch litigation, The Times not only made no attempt to correct the built-in bias, but actively aided it.
For example, a long article about the switch problem, titled "In General Motors Recalls, Inaction and Trail of Fatal Crashes," gives an accident victim's mother the last word: "It makes me very angry -they didn't do their jobs," she said. "I am surprised that it wasn't investigated and also surprised that G.M. would allow faulty cars out on the road. They knew there was a problem, and they didn't do anything about it." (Jensen, 2014c).
There is no comment from GM on the case, which was in litigation; the most relevant quote from GM is a comment about the accidents generally: "These crashes occurred off-road and at high speeds, where the probability of serious or fatal injuries was high regardless of air bag deployment," G.M. said in a statement on Feb. 13. "In addition, failure to wear seatbelts and alcohol use were factors in some of these cases." 53 The story did not mention that in the particular crash at issue, the car was being driven illegally by a young girl with a learner's permit unaccompanied by a licensed driver (Valukas, 2014, p. 116), that the car was going fast enough that it became airborne at one point, (Jensen, 2014a) and that neither of the two passengers who were killed nor the driver, who was seriously injured, was wearing a seatbelt (Valukas, 2014, p. 114).

An example
Most of the problems with The Times' coverage can be seen in an article describing a settlement between the compensation fund and the parents of Brooke Melton, who died in the 2010 crash of her Cobalt (Vlasic, 2015a). The parents had earlier settled with GM for $5 million, but sued again after the ignition switch problems became known. (Settlement agreements usually provide that settlement amounts are to be kept secret, so it's not clear where the $5 million figure came from.) The Times' one-sentence description of the accident shows that the reporter, who covered the story from the earliest days, still did not understand the Valukas report, or chose to ignore it: Ms. Melton was killed after she lost control of her 2005 Chevrolet Cobalt on a Georgia highway when the ignition key slipped into the accessory position, shutting down engine power and disabling the car's airbags.
A picture caption is to the same effect: "Brooke Melton was killed in 2010 after a faulty switch cut her Chevrolet Cobalt's engines, leading to a crash in which the car's airbags did not deploy." The implication is that Ms. Melton lost control because ("when," "after") the ignition key went to the ACC position. The double-impact theory is simply ignored.
While The Times does not give any details of the accident, a CNN Money report described the accident as follows: [Melton] lost control of the car. It spun out, hydroplaned, hit an oncoming vehicle and rolled off the road, dropping 15 feet into a creek. (Wallace, Harlow, & Hobor, 2014).
It's not clear that the crash can be fitted into the double impact theory-presumably most of the damage was done by the first collision with the other vehicle. And hydroplaning would explain the loss of control.
Then there are the statements by the family's lawyers (the parents declined to comment): A lawyer for the victim's parents … said an aim of the second suit was to force G.M. to admit it had concealed evidence of the deadly ignition problem.
"One of the important issues for the Meltons was accountability," said … the family's lawyer. "They are grieving parents who simply wanted the truth and for no one else to suffer a similar loss." Another lawyer for the parents added that: … the initial suit was "directly responsible for alerting the government and the public to a massive cover-up by General Motors." The Times was certainly within its rights to publish statements from the only sources they had, but it would have been appropriate for anyone who read the Valukas report to question the accusation of "a massive cover-up," or to point out that the eventual settlement with GM's victim-compensation fund (of which more below) had nothing to do with getting at the truth or shielding others from a similar loss, and everything to do with money.
The Times' failure to say anything about the "massive cover-up" allegation may stem from their swallowing the allegation whole. Later in the story the reporter states that "G.M. had secretly upgraded its switches without changing the part's identification number-further frustrating efforts to pinpoint the cause of dozens of accidents that caused deaths and injuries." Technically, this is correct: Ray DeGiorgio approved an upgraded switch without telling anyone and without changing the part number, and in law Mr. DeGiorgio's actions can be taken to be the act of the corporation. But that doesn't make the sentence less misleading. For those unfamiliar with the actual story, the sentence suggests that the switch was changed in secret to prevent anyone from finding out its role in the airbag non-deployments. Only the most knowledgeable readers would know that the change was the act of a single employee who may not have realized that the switch problems had a safety aspect, or that the people misled by the failure to change the part number were other GM engineers who were trying to figure out why Cobalt airbags were failing to deploy.

How many deaths?
In the wake of the public outcry, GM set up a fund-the GM Ignition Compensation Claims Resolution Facility-to compensate victims. 54 The fund was overseen by Kenneth Feinberg, the lawyer who ran the compensation fund BP set up in the wake of the 2010 Deepwater Horizon oil spill. GM initially set aside $400 million for the ignition switch fund, but there was no cap and the eventual payout was approximately $525 million (GM Ignition Compensation Claims Resolution Facility Final Report, 2015. All the statistics in this section relating to the Facility come from the Final Report).
The fund ceased accepting compensation claims on 31 January 2015, by which time 4,343 claims had been received, of which 473 were for fatalities and 280 for serious injuries. Mr. Feinberg eventually found 124 of the fatality and 18 of the serious injury claims eligible for an award, rejecting the rest. 55 The media tends to accept that these "eligible" deaths and injuries were caused by the faulty switch; for example, a typical story, from the Associated Press, states that "The death toll from crashes caused by faulty ignition switches in 2.6 million older General Motors small cars has reached 100" (emphasis added) (Associated Press, 2015). The Times report of the same milestone is more cautious, referring only to "the 100th death claim tied to faulty ignition switches" (emphasis added) (Vlasic, 2015). But even this may overstate the connection in many cases.
There is an inevitable ambiguity about causation: When a distraught and somewhat inebriated teenager drives away from a party without bothering to attach her seatbelt, loses control of her car at high speed, and crashes into a tree (the Amber Marie Rose case) (Jensen, 2014c), it's a bit simplistic to say that the airbag's failure to deploy caused her death. There were numerous causes; there always are. 56 Even when airbags deploy, drivers and passengers may not survive. All cars sold in American since 1998 have been required to have front seat airbags, 57 and these airbags usually deploy, yet in 2013 over 22,000 drivers and passengers died on the nation's roads, 58 the vast majority protected (inadequately) by airbags.
A more tractable question is "How many people died in Cobalt crashes where the airbags failed to deploy because of the defective switch?" But this, too, is a difficult question, and so Feinberg chose to base settlement on a simpler standard: Did the airbags fail to deploy? This means that settlement offers were made even when it was not obvious that the ignition switch was the cause of the nondeployment, or that deployment would have made a difference.
Take the case of Trenton Buzzard, left paralyzed from the waist down by a crash a week before his first birthday that killed his grandmother and 13-year-old aunt (Stout, 2014b);. It's difficult to make the accident fit the double-impact theory: The car collided with an oncoming vehicle. It's unclear why the airbags didn't deploy, but in any case young Trenton was in the back (in a car seat), where there was no airbag to protect him. Nonetheless, Mr. Feinberg offered a settlement. Settlements were also offered in the single-impact crash described above where a Cobalt rear-ended a tractortrailer (Stout, 2014a), and for at least one other back-seat passenger (Jensen, 2014a;Stout & Ivory, 2015).
There are no doubt worse uses for GM's money than compensating the victims of automobile crashes, but we (and The Times) should understand that Mr. Feinberg's decision to compensate a crash victim didn't necessarily mean that the ignition switch was at fault or that the airbag's deployment would have lessened the damage. As the Final Report stated, [T]he Facility's goal was to create a compensation program that was efficient, speedy, cost effective, and consistent in resolving the claims of individuals who applied for compensation. For this reason, the Facility did not conduct rigorous scientific or technical determinations or engineering analyses as to whether an ignition-switch defect manifested itself in a particular accident or whether a particular death or injury was "caused" by an ignition-switch defect.

Why?
The Times reporting of the GM ignition switch story was unique in that much of the reporting followed the release of the Valukas report, which remains the only detailed independent 59 description of what happened. In most cases of technical disaster-air crashes, space shuttle explosions, nuclear meltdowns-detailed technical explanations only appear after the story has vanished from the headlines. In the GM case, however, much of The Times' reporting, including several long articles about particular accidents, appeared well after the release of the Valukas report (e.g. Stout, 2014aStout, , 2014bVlasic, 2015b).
The Times did not ignore the Valukas report entirely: The report was covered the day after it was released (Vlasic, 2014), and Times reporters must have gone through it carefully in compounding a chronology of the events at GM (Ivory, 2014). But the GM engineers' theory of the causes of the accidents did not figure in these stories.
It's not that understanding the engineers' theory required any particular expertise; I've explained it above in pretty much the non-technical way it was explained in the Valukas report. One suspects that The Times' reporters missed it simply because it did not fit into their preconceptions about what must have happened. Thus, The Times' handling of the GM story allows us to examine those preconceptions without our having to make allowances for the reporters simply lacking full information.
Why was The Times' description of the ignition switch problem so distorted, and so relentlessly critical of General Motors? One possible reason is a general human failing: We tend to see events as the product of individual human actions. It's the standard way primitive peoples explain storms, earthquakes, crop failures-there's a god who directs these things for His human-like ends. If you want to avoid these natural calamities, you'd better stay on his good side by offering prayers, praise, food, and virgins.
We're a lot more sophisticated than that now, but we can still slip into thinking of organizational behavior on the model of individual behavior: A corporation, it is supposed, does things for human reasons, which must be the reasons of the top management. It's as if the corporation was an organism, with a mind (top management) controlling the actions of the corporate limbs (the employees) much as our minds control our muscles. As a result, when things go wrong, we often assume that the events happened because the top management-the corporate mind-willed them. Thus, when the Valukas report failed to fault any of GM's top executives, U.S. Senator Richard Blumenthal stated that "It seems like the best report money can buy," presumably because "It absolves upper management, denies deliberate wrongdoing and dismisses corporate culpability." (Vlasic, 2014). The Senator no doubt saw these conclusions-for which he presented no empirical evidence-as following from the essential nature of the corporation-as-person. 60 This vision of the corporation often goes hand-in-hand with a simplified picture of human motivation. We accept that corporate executives want to maximize the corporation's profits (fair enough), but then assume that they will do anything, within or without the law, to achieve that goal. In Hollywood, corporate executives frequently plot the murder of rivals or whistle-blowers to protect their profits. But how often does this happen in real life? I personally can't think of a single example. In real life, motivations are constrained by law and by a sense of what actions can and cannot be justified, to others and to oneself. Moreover, for a corporate executive to perform any of these illegal or immoral actions usually requires that he enlist the aid of many underlings, each of whom is likely to carry their own sense of legal and moral limitations. This is not to say that businesses do not defend harmful products: The urge of corporate executives to find the evidence for harm lacking (like the urge of corporate critics to find such evidence overwhelming) is a sad psychological fact. But that's a lot different from knowingly selling a dangerous product.
Of course, tobacco companies pedaled cigarettes long after they must have believed that the product was harmful. But in their defense, lots of Americans wanted to continue smoking whatever the risk, and the tobacco companies had no alternative policy. But it's hard to see why, if GM executives knew that the defective switch was killing people, they would delay acting when it was obvious that the costs of delay would be much greater than the costs of a recall.
For anyone familiar with large-organization decision processes when faced with technical problems, GM's response to the problem was to be expected (leaving aside DeGiorgio's actions, which remain mysterious). For a classic example, we have the sociologist Dianne Vaughan's book on the decision process leading to the 1986 loss of the space shuttle Challenger (Vaughan, 1996). Vaughan presents the narrative twice, first as a journalist (or a lawyer) might, as a series of stupid and/or venal decisions that any sensible person would have seen likely to lead to disaster. But her second account, told in greater detail, shows how the launch decision reflected years of experience with a process that everyone acknowledged was fraught with risk. Let me assure you: Nothing you've ever done in your life is likely to have been as carefully thought through as NASA's decision to launch Challenger. Then, at the end, Vaughan shows how the experience that the NASA engineers relied on, had it been organized in a different way, might have told them not to launch.
For those not up to Vaughan's 592-page account, the lessons are summarized in the "Blowup" chapter of Malcolm Gladwell's What the Dog Saw: And Other Adventures. The chapter's tagline says it all: "Who can be blamed for a disaster like the Challenger explosion? No one, and we'd better get used to it." (Gladwell, 2009, p. 280).
The Times' business reporters are no doubt familiar with organizational disfunction; The Times has had some of its own well-publicized incidents in recent years. So how could they have bungled things so badly? I don't know any of them personally, but it's hard to believe that, as business writers, they are generally antagonistic to the business world. I can only surmise that they were carried away by their conviction that the exposing skullduggery in high places is the highest kind of reporting, and by the pressure to tell a compelling story, like you see on TV. For most of us, the most compelling stories are those with heroes and villains, and if a suitable human villain of sufficient power and evil can't be found-Ray DeGiorgio just doesn't meet the requirements of this scenario-then an imaginary all-powerful villain will do, like Dracula, Voldemort, or the American business corporation. In that light, the reporters' failure to understand the Valukas report becomes a helpful oversight, and their focus on the corporate villain a casting decision.