The Defeat Device

When cars get self-conscious: a brief look at the Volkswagen scandal.

The Defeat Device

When cars get self-conscious: a brief look at the Volkswagen scandal.

Diesel is cheap — cheaper than petrol. That’s why car companies have been investing heavily in developing diesel engines over the last decade.

Diesel is also more polluting, releasing nitrogen oxides and several times more soot and particulate matter.

Countries like the U.S. and the U.K. — developed, and so widely polluted ones — have passed laws and regulations to limit further pollution. To be able to appear on the market, a car must pass several stringent pollution tests that prove a low concentration of toxic gases. These tests are conducted, of course, in laboratories.

Psychologists use a concept known as “reliability”.

A psychologist must consider, when conducting a study, whether the context and environment would make the results unreliable when applied to a real-life scenario. A participant in a study about empathy for others, for example, would likely be more likely to help a stranger if she were aware she was being watched — we’re all afraid of being judged, after all.

This would render the results inaccurate, as that may not be her true reaction if faced with the same scenario, say, on the way home from work.

You could say that such a study is unreliable.

In the world of automobiles, the testing phase is widely known as being almost-but-not-quite rigged. Car doors are taped up, different tyres are used, and several other variables are altered to allow for the best possible performance of the car. Given this, it comes as no surprise that cars’ emissions are usually lower in laboratory conditions than on the road.

These results are never truly reliable. Some increase in pollution levels is always expected — and allowed.

In 2013, a group of engineers from the Centre for Alternative Fuels, Engines and Emissions (CAFEE) of West Virginia University tested three diesel-engine powered cars using a very innovative method: actually driving them around the road. They did this in various kinds of places and situations — like high elevation, peak hour city traffic, and short-burst driving that simulated errand-running.

These were not standard testing conditions.

And, they revealed something. Two of the three cars they tested were polluting between five and thirty-five times more on the road than in a lab!

Both these cars had been made by Volkswagen.

Volkswagen had been one of the first to jump onto the “clean energy” bandwagon. By this time, they were considered a leader in the field. They announced their emissions-friendly Clean Diesel cars back in 2008, and went on to release a series of cars — each cleaner, greener, and more fuel-efficient than the last.

Clean Diesel won several environment awards, and received multiple tax-breaks to encourage the good work they were doing.

Turns out, it was nowhere near as good as everyone had assumed.

On 18th September 2015, the United States Environment Protection Agency ordered Volkswagen to recall 482,000 cars. They’d found evidence that the tests had been rigged, using what came to be called a “defeat device”.

Just like the subject of a study on empathy might look out for when she was being watched and then be suitably nicer, the cars had been implanted with software that could deduce when it was under testing conditions, and become suitably more efficient just for that time.

Once out of the lab, the program would relax, letting emissions reach up to forty times the prescribed limit.

It was a disaster.

The brass at Volkswagen — the Directors, the VPs, the senior managerial executives — claimed that the defeat device had been the work of a “few rogue engineers”. But that doesn’t quite make sense. The higher ups at Volkswagen publicly pride themselves on their knowledge of automobiles and engineering. A car company from the top down — not just expensive suits supervising greasy engineers.

If Volkswagen had just made the breakthrough of the decade by revolutionizing diesel emissions without compromising on costs, wouldn’t they have been curious? Wouldn’t they have asked how?

There is an explanation as to why not.

Say it’s going to be winter soon, and you need a new sweater. And perhaps you’re putting off buying the sweater because you’re lazy, or you don’t feel like making the drive to the mall, or you hate shopping.

“It’s okay,” you say. “Today wasn’t much cooler than it was yesterday.” It is cooler, to be sure — but the difference is so tiny, it doesn’t really make a difference.

And this continues. Every day is just a little chillier than the last, but you don’t realise this until you wake one morning to snow and end up shivering on the way to work.

The same could have happened to the Volkswagen engineers.

Diesel emissions are hard to cut down, so they tweak the code that regulates it. And continue to tweak the code through every review and rewrite and nobody realizes just what has been created until the CAFEE test results are published in March 2014.

This is an entirely plausible explanation. Modern high-end cars require millions of lines of code, a volume impossible for any one person to fully know or even grasp. It really could have just been oversight.

Or not.

Morally, stealing a book from a shop would be considered theft. There is no question of this. In contrast, would the same be felt about downloading a pdf version without paying? Would it feel as wrong?

The intangible nature of software tends to blur many of the virtuous lines we draw when it comes to material things. It makes cheating seem less like cheating, more like bending the rules.

Makes things like the defeat device just one more in a long list of artificial courtesies, extended to the car-companies to make their rides easier.

Have something to say? At Snipette, we encourage questions, comments, corrections and clarifications — even if they are something that can be easily Googled! Or you can simply click on the ‘👏 clap’ button, to tell us how much you liked reading this.

This article uses “The Defeat Device”, originally published in Sirius #204, dated 11–24 October 2015, as a starting point. Other references used can be found here.