28 Sep 2015 15:16 IST

L’affaire Volkswagen: Beating the management control system

The company’s behaviour assumed that standard linkages exist among specified entities in a closed loop

Heaven forbid, but let us imagine the doctor suspects there could be some problem with the cardio-vascular performance of your heart. He suggests you undergo a ‘treadmill’ test.

You get on to the treadmill; warm up with an initial three-minute walk at roughly 3 km per hour, with your body in perfect alignment with the surface beneath. You follow that up with a further 3-minute walk at the same speed but with a 5-degree inclination to the surface.

Then follows an iterative sequence of stepping up and reducing the parameters of speed and angle of inclination at which the treadmill operates. A healthy heart must be able to withstand the stress that such running imposes for a certain duration of time for the doctor to be able to come to a conclusion about the state of your cardiac fitness.

You can find out more on all of this by searching the Internet with the phrase ‘Bruce protocol’. Now, this is where it gets interesting. A complete knowledge of what to expect under the ’Bruce protocol’ does not, by any means, guarantee that your heart will be able to withstand the strain imposed by the rigour of running under those test conditions.

If it is a creaky heart, it is only a matter of time before it starts to groan under the weight of the stress imposed by such running. In other words, any amount of prior knowledge will be of no avail in helping you pass the cardiac fitness test.

Programmed output

But if your heart is a computer and has been programmed to spout the standard output parameters of a healthy heart, it will do so the moment the treadmill starts at the 3-kmph speed. It is as though the heart-computer is telling itself, “hey, this looks like a cardiac fitness treadmill test, so get those standard numbers for heart rate, systolic blood pressure, etc., out”.

That is pretty much what happened with Volkswagen, in the emission test on cars fitted with its 2-litre diesel engine at the test facility of the US Environment Protection Agency (EPA), in California.

The moment the car was mounted on an engine test bed and started to run on a pre-programmed basis much like the treadmill that human beings get on to for measuring cardiac fitness under the ‘Bruce Protocol’, the engine’s computer inside the Volkswagen car started to spout some standard numbers that led to the EPA giving it a clean bill of health.

This worked fine for a certain number of years and would have continued thus but for a European non-governmental organisation which thought that it too, could have a say in the validation of emission norms of cars. It was joined by a professor from a university in the US (West Virginia) who thought he had a superior device to the one available inside the lab facilities at the EPA. More importantly, he also had on hand an alternative testing protocol. The rest is history. And the ramifications are still unfolding for Volkswagen.

Where it went wrong

That leads us logically to the question: Where did Volkswagen go wrong? In very simple, layman’s language, it could be said that Volkswagen thought it could cheat and get away it but was proved wrong, later. But this actually begs the question: Why did the company think it could get away with it? We therefore need a more nuanced explanation for Volkswagen’s behaviour.

The truth is that the folks at Volkswagen thought that the fitness of their cars from an ‘emission’ perspective would be evaluated only by the equipment attached to the test beds inside the EPA laboratory. In their view, their universe consisted of the company, the diesel engine fitted cars, the EPA, its testing facility and the standard protocols for emission testing.

Further, they believed that standard linkages existed among these tightly specified entities. So they modelled their behaviour on these assumptions and hoped to be judged as being in conformity with the laid down norms. In short, they thought they were subjected to a ‘closed loop’ management control system.

But the reality is that they were actually operating in a world with an unspecified number of entities beyond the EPA, such as the West Virginia University and a Europe-based NGO with an agenda of environmental protection.

Equally, the linkages extended beyond test beds and EPA-mandated protocols for emission testing. So, they were actually operating under an open-ended (in contrast to a closed-loop) system and, hence, it was inevitable that their behaviour, which would have been perfectly fine in a ‘closed-loop’ control system, would come up short under an alternative ‘open’ system with flexible linkages.

Gaming the system

In fairness to Volkswagen, what it did was completely in accord with human nature. If the stakes are high enough, all of us prone to ‘gaming’ the system that judges us and holds us to account by some exacting standards.

Volkswagen was indulging in a behaviour that is no different from that of students who take the IIT-JEE in the hope that they can crack the exam a little better than the hundreds of thousands of others with whom they are competing. It started as a test (a control system for filtering students with the right ability) for evaluating the mathematical and logical reasoning abilities of candidates aspiring for a seat in these institutions imparting technical education.

Today, it has been gamed by the students and their tutors, and the net result is that it (the control system) has been reduced to one of learning by heart the answers to a fairly large, but nevertheless finite set of questions.

The students may have succeeded in ‘gaming’ the system and emerged successful. But it doesn’t work for them in the long run. For if it did, we would not be hearing about the utter lack of employability of many of those who pass through the portals of these centres of excellence. The real world, which is another control system, is open-ended with loosely defined linkages that cannot be gamed.

Control design atrophy

For the regulatory authorities too, there is a lesson to be learnt from the Volkswagen episode. Control systems have a tendency to atrophy over time. They no longer deliver the output expected of them. There is a divergence between the expected outcome and the actual, with the added disadvantage that neither the regulators nor the general public who look to the regulators to ensure satisfactory outcome, are even aware of such divergence.

Part of the problem is that the regulator depends excessively on its own ability to control outcomes. Because it relies exclusively on itself there is a need to reduce oversight to a set of standardised processes. Perhaps there is even a case for crowd-sourcing regulation.

There was this story (apocryphal, perhaps) when Airbus first launched its fully automated (fly-by-wire) A-320 line of aircraft. The CEO had proudly announced that the new aircraft had been designed to operate with just a pilot and his dog. The pilot was required to sit with his hands held closely to his chest and the dog was meant to bite him the moment he tried to fiddle with the control knobs!

Thank God the airlines haven’t completely heeded such well-meaning advice. Just as the much-hyped driverless cars run by computers are subject to one great risk — hacking by outsiders, making drivers a safety feature in such cases, even these hi-tech aircraft could do with a certain degree of human intervention inside.

Recommended for you