For six years, Volkswagen cars programmatically cheated emissions tests to appear “clean” under scrutiny, and polluting 40 times as much otherwise.
This deceptive code was built into vehicle software to skirt regulations algorithmically, and naturally, the scandal has left the company in hot water.
As cars and other objects get smarter, the downsides of intelligent objects are becoming more apparent. One being that manufacturers can write in software cheats, as VW did, for their own interests, and not without consequence.
In terms of corporate cheating, cheating software is unique in that it happens in early design stages, far removed from actual testing, and is typically quite subtle in nature.
Not unlike VW, smartphone manufacturer Samsung was caught programming phones to detect when they’re being tested and artificially enhance performance temporarily.
The consequences may not seem huge, but when things like this slide, products are substandard — in VW’s case, at detriment to the environment, government subsidies, and the consent of buyers.
The potential for others to do the same is frightening, especially when it comes to electronic voting ballots, city infrastructure, medical devices, and other important objects.
Can it be stopped?
There are various methods that could help curb the probability of software cheats.
- “In the wild” testing: If devices are tested out of the lab or at random, it will be more difficult to for them to cheat programmatically
- Open software: If software can be made open to researchers instead of using copyright claims to block it, it could benefit from oversight and regulation
- Regulatory oversight: As cheaters get smarter and harder to detect, only frequent and extensive research by government and private code testing may catch software cheats
Overall, the incentive for companies to cheat and their ability to get away with it may get more difficult to combat. But in a world increasingly dependent on algorithms and software, it’s important that the devices we rely on are exactly what they purport to be, instead of malfeasance in disguise.
Executive management should also bear more responsibility by becoming familiar with what exactly is being built into their software by engineers, as they are often blind to the details of coding that could compromise their products, services, and entire business.