Lies, Damn’ Lies…

The trouble Volkswagen have got themselves into may be symptomatic of a wider malaise, and we may find that their main failing is breaking the 11th Commandment.

Most people, quite naturally, tend to believe the information provided by their gadgets. Between my training as a physicist, my fascination with numbers and my professional leanings, I’m definitely inclined to the view expressed in the famous quote "never believe anything you read in a newspaper except the date, and that only after you have checked it in a calendar". I’m always trying to cross-check the instrumentation of everyday equipment, to understand which are accurate, and which not. This goes especially for all those read-outs in a car, with ideal opportunities on long journeys.

A car’s speedo, for example, can be cross-checked against a GPS with a speed readout. The latter tend to lag slightly behind the actual value, but can be very accurate once you are travelling at a constant speed, such as on the motorway with the cruise control engaged. I reckon a GPS is good to within about 0.5 mph under those conditions. Alternatively, there’s always the old "Sherlock Holmes" method, which I used to use before the GPS days: travel at a constant speed and time yourself past 17.5 of those little blue posts. That’s one mile, and as the great detective says in Silver Blaze, "the calculation is a simple one".

Over the years I’ve seen a steady improvement in the accuracy of speedometers. In my early motoring years it wasn’t unusual to find the speed being exaggerated by as much as 5mp at motorway speeds, but my latest car, the Mercedes E-Class, seems to be accurate to about 1mph at speeds as fast as I can safely check on British motorways.

For some reason, that’s not true of fuel efficiency. The most accurate way to measure that is the old one: fill up to the brim (or at least the pump cutout) and zero the trip counter. When the tank is nearly empty fill up again, and divide miles by gallons, or litres depending on your persuasion. That measurement is probably accurate to about +3%, maybe better, or less than 1mpg in the 30-40mpg range.

Now on my VW Eos, I found that the average fuel economy readout from the trip meter consistently agreed with my own calculation to within about 1mpg. Good enough that I stopped checking manually. Not true of the Mercedes. The error varies, but it’s always considerably optimistic, sometimes by as much as 3 or 4 mpg on a real figure in the range 32-35mg. That’s an error in excess of 10%. In absolute terms it’s still very impressive for a big heavy car which can do 0-60 in around 6s, but not as good as you are led to believe…

If you think about it, the reasons are obvious. In older cars, accurate speed measurement was a challenge. Both regulations and psychology inclined to flatter a car’s performance: the regulation states that any error must be to show a speed above actual, and that was also desirable in sales terms when cars were slower.  Nowadays there’s no benefit to exaggerating the real speed, and a distinct benefit to providing an accurate value if possible so the driver can maximise use of the speed limit.

The opposite is unfortunately true of fuel economy. My own VW experience suggests that it’s perfectly possible to provide a fairly accurate report (although it’s always possible that I may just have been lucky), and I struggle to understand any technical reason why the Mercedes is so inaccurate. I’ve checked the obvious sources of error, such as an inaccurate odometer, and can’t find anything. However when you consider the psychology, the reason is apparent – we all want to think that we’re driving efficient cars, and my Mercedes tells a very good story, if only I wasn’t a cussed so and so who checks things!

While an inaccurate fuel economy read-out is nothing new, and probably hasn’t broken any laws the way the VW diagnostic software did, it does appear that the general issue may be broader than we think.