GM's ignition-switch defect. Engineers bought a fish scale during a salvage-yard investigation

In early 2002, testing by supplier Delphi showed that the ignition switch required less torque to move from Run to Accessory than GM’s specification called for.

Engineer Ray DeGiorgio told Delphi not to increase the rotational torque because he was concerned about the switch’s electrical performance. He also said the switch, then planned for the Saturn Ion, would need modifications to meet the torque specification before use in the Cobalt. His email signature included the phrase “tired of the switch from hell.”

Testing continued to show results below specification, but on May 3, 2002 the switch was approved for shipment through the production-part approval process.

The report says its investigation did not identify another GM employee who knew before production that the switch failed the torque specification.

In 2005, GM prepared a Technical Service Bulletin for dealers about inadvertent movement of the key.

Steven Oakley, who drafted the bulletin, told investigators that “stall” was language that attracted attention because it could suggest a safety issue, a recall, or NHTSA scrutiny. Other employees likewise described sensitivity around using that term in service bulletins.

After Product Investigations reviewed the draft, the engine-stall language was removed from the bulletin.

Oakley also said he had been reluctant to push hard on safety issues because he believed his predecessor had been pushed out of that job for doing so. The report also notes that discussions with engineers had alleviated Oakley’s own initial safety concern.

In 2006, GM and Delphi changed the detent plunger and spring inside the ignition switch. The longer plunger and stronger spring increased the force required to move the key out of Run.

DeGiorgio approved implementation of the change in April 2006.

Even after the internal design changed, the ignition switch kept the same part number.

That later hindered investigators who treated switches carrying the same number as the same component. DeGiorgio told the investigation that he did not remember authorizing the change or deciding not to change the part number.

After a 2006 Chevrolet Cobalt crash in Wisconsin in which the front airbags did not deploy, State Trooper Keith Young investigated the circumstances.

Young connected the possibility that the ignition switch had moved from Run to Accessory before impact with the lack of airbag deployment. An Indiana University investigation separately identified the same possibility.

The state-police report was in GM’s legal-side files by February 2007, yet the Valukas Report says GM lawyers and engineers working on airbag non-deployment cases did not learn of its existence until the 2014 investigation was underway.

On February 18, 2009, GM engineer Joseph R. Manson wrote an internal email about the long-running Cobalt issue.

The email joked that the issue had existed since humans first lumbered out of the sea and stood on two feet.

The problem-tracking record was treated as a continuation of issues recorded in 2004 and 2005. The report says it was still regarded as a customer-convenience issue rather than a safety issue.

In January 2011, GM lawyers requested a meeting with engineers about the Cobalt ignition-switch issue. The meeting finally took place in July.

Brian Stouffer, assigned to investigate after the meeting, said his manager had not told him the matter was urgent. The report found that the investigation lacked a timetable or a supervisor demanding action.

At the beginning of 2012, the report concluded, the investigation had moved not at all.

The new investigator was examining the same data as his predecessor, with some additional information that had been in GM’s files for years; the report found that little else had changed.

On March 28, 2012, GM engineers visited a salvage yard in Davison, Michigan. They intended to extract a body-control module from a crashed vehicle.

While moving cars, an engineer turned a Cobalt’s key and noticed that it moved extraordinarily easily. According to John Dolan, the engineers wanted to measure the force but had not brought measuring tools.

They dispatched an investigator to a local bait-and-tackle shop to buy a fish scale.

The engineers used it to measure the force needed to move keys out of Run in several Cobalts. Dolan said he was surprised: if the measurements were correct, they suggested a pothole impact might move a switch out of Run.

The report discusses a 2008 GM writing presentation that urged employees to “write smart” and avoid judgmental adjectives and speculation.

It suggested replacing “Problem” with “Issue, Condition, Matter”; “Safety” with “Has Potential Safety Implications”; and “Defect” with “Does not perform to design.”

It also listed sentences to avoid, including statements that something was dangerous or had almost caused an accident. The presenter described expressions such as “rolling sarcophagus” as attempted humor. A report footnote adds that a lawyer who later saw a version of the training told the presenter to remove the slide listing words never to use.

At DeGiorgio’s April 29, 2013 deposition, plaintiffs presented evidence that the switch design had changed. Internal components differed even though the switches carried the same part number.

From the deposition, GM’s outside lawyer Philip Holladay emailed GM lawyer Ronald Porter, describing plaintiffs’ evidence as a “bombshell.”

Three days later, Holladay recommended hiring Subbaiah Malladi to assist with the litigation and bring the internal investigation to closure without delay. Malladi later confirmed the design change, but told investigators that he had not been asked to determine whether low torque caused the airbag non-deployments.

The case in which the switch change was presented at deposition settled in September 2013 for $5 million.

The report identifies $5 million as the maximum the Settlement Review Committee could authorize without General Counsel approval.

Lawrence Buonomo approved the settlement. The report found that GM’s lawyers had not elevated outside counsel’s evaluation or specific issues in the case to General Counsel Michael Millikin before settlement.

The report does not establish that the amount was chosen to avoid General Counsel review.

Several GM employees told the investigation that they did not take notes at critical safety meetings because they believed GM lawyers did not want such notes created.

But investigators could not identify a lawyer who had issued such an instruction, found no lawyer who said they had given it, and found no document or email containing the instruction.

The no-notes practice nevertheless spread from employee to employee until the report described it as an internal “urban myth.”

As a result, for many committee meetings there were no clear records of who attended, what was discussed, or what was decided.

Near the end of the report, two expressions used inside GM appear in the discussion of corporate culture.

One employee described the “GM Salute” as crossing the arms and pointing outward toward others to indicate that responsibility belonged elsewhere.

Mary Barra described the “GM Nod” as a meeting where everyone agreed with a proposed plan, then left without intending to carry it out.