Introduction
After a university study uncovered code designed to cheat emissions testing standards, Volkswagen Inc. (VW) has been on the defensive, admitting wrongdoing and bracing for the onslaught of regulatory fines, class actions suits, and major repairs and recalls.
The code at the heart of the controversy places the car in one of two operating modes. When the car appears to be driving under conditions simulating an emissions test, the “cheat code” is enabled, delivering proficient emissions results and better gas mileage. When driving conditions denote real-world driving, cheat mode is disabled, delivering increased power and torque, but decreasing gas mileage and outputting a level of emissions 40 times greater than the legal limit as regulated by the Environmental Protection Agency (EPA).
Discovering the Cheat Code
Researchers at West Virginia University uncovered the higher emissions during a study funded by the International Council on Clean Transportation, a nonprofit with offices in the U.S. and Europe, to test the emissions of diesel vehicles while driving. Traditionally, emissions testing occurred in a stationary location by placing the front-wheels of a car on a rolling treadmill while the rear wheels remained static. Emissions that escaped through the tail pipe were then collected and measured. The WVU researchers took the tests to the open road by creating a mobile testing rig. Sensors attached to the tailpipe captured the emissions and fed the data to testing equipment stored in the trunk and backseat of the cars. The test results captured the greater emissions and lower fuel efficiency since the cheat code was disabled on open road conditions. Upon discovering the discrepancies and conducting multiple follow up tests, WVU contacted the EPA and the California Air Resources Board, who conducted their own tests and issued a citation to VW.
Gaming the System
Just how did VW construct the code to delineate between testing conditions and open-road driving? As previously mentioned, traditional test conditions were well established and known throughout the auto industry. That transparency by testing agencies allowed the automaker to programmatically check for test-like conditions. These included:
- Were the rear wheels moving or stationary?
During traditional testing, the car’s front wheels were placed on a rolling treadmill, while the rear wheels stayed put. The cheat code gathered input from the rear wheel sensors to determine whether the wheels were moving or not. Consumer Reports successfully enabled the cheat code by disconnecting the rear wheel sensors from the wheels, and driving the car on the open road. With the sensors disabled, the car – determining the wheels were static and therefore undergoing an emissions test – enabled the cheat code normally deployed while testing occurred. In these instances, the car delivered the reduced emissions previously recorded in the test labs.
- Were other obvious test indicators present?
Besides the rear wheels registering as stationary, other known testing protocols provided additional clues as to whether the car was undergoing emissions testing. These included: turning on the hazard lights; increasing speeds up to normal road and highway speeds without turning the steering wheel; and conducting all of the actions with the driver-side door open. While none of these items on their own may definitively indicate testing conditions, using all of the information provided by the sensors and software, taken in concert these items were used to predictably determine when emissions testing was occurring.
The Fallout: Law suits, major repairs and recalls
VW admitted to knowingly deploying the code, estimating that 11 million of its diesel vehicles were equipped with the emissions cheating software. According to the National Journal (paywall), VW has been hit with over 175 class action suits across 32 states since the company admitted wrongdoing. The majority of the suits have been brought on behalf of consumers alleging they were taken advantage of when they paid premium prices for “clean diesel” cars that failed EPA tests in real world driving conditions. The company has agreed to fix the problem, though it is taking differing approaches for customers in the U.S. and Europe. Cars in the U.S. will require a “major fix” involving software updates and the addition of a new emission filtration system. This fix is expected to be put in place in 2016, with the company required to submit its final plan to U.S. regulators by November 20, 2015. Repairs for cars sold in Europe will not include the installation of the new emissions filtration system, as the pollutions standards there are not as stringent as in the U.S.
VW has set aside $7.3 billion to address the expected costs of regulatory fines, legal bills, and costs of repairs and recalls.
The cheat code, whose earliest iteration has been identified in 2009 models, highlights the challenge both regulators and consumers face in trusting providers of goods and services when so much is delivered via software that cannot easily be reviewed. Whether it’s code regulating the emissions output of a car, or algorithms determining “items you may like” within a paid service, there is an implicit trust by consumers that companies will not act fraudulently or unduly opportunistically. The VW scandal has revealed that such trust may be misplaced, and higher testing and oversight standards may be required moving forward.
