The Numbers Game…

It’s time to stop kidding ourselves!

By Roy Gregory

It’s mainly a man thing (at least it seems like that to me), but people buying stuff just love numbers. Whether it’s the 0-62 figure for a car or the weight of a bicycle, the world is full of guys who know every stat or measurement that applies to everything – and the value of nothing… There’s no ignoring it: there is and always has been something peculiarly seductive about numbers, not least because they promise easy answers to otherwise imponderable questions. And what generates more numbers than an audio system? The specs and measurements are legion, from the efficiency and bandwidth of loudspeakers to the rated output of power amplifiers. The problem is that the apparently ‘simple truth’ represented by numbers is often far from simple. Sometimes it’s a subtle distortion: occasionally it’s a downright lie.

That 0-60 figure that you are secretly so proud of every time someone asks you what car you drive? That was achieved on a bone dry and perfectly manicured piece of tarmac, with slick tires, a way better driver than you and no wife or kids in the car. Just how meaningful is it? Or take the situation a few years ago, where the ‘gold standard’ for serious bicycles was a frame that weighed less than a kilogram (these days it’s less than 700g!). This bizarrely abstract figure loomed so large in both reviews and the consciousness of the buying public that it became a necessity if manufacturers actually wanted to sell product. So large in fact, that manufacturers took to weighing not just their smallest frame, but weighing it unpainted and often devoid of functional necessities like cable stops, the seat clamp and bottom bracket shell!

“Not everything that counts can be counted…”

In the same way that car manufacturers and bike builders don’t actually lie about their published specs (they just don’t tell the whole truth), the audio industry adopts and promotes particular numbers, generally without being any too careful about defining the parameters of testing, or their relevance to actual performance. So let’s take the example of a loudspeaker. Most manufacturers will publish figures for efficiency, bandwidth and nominal impedance. These should (in theory) enable you to gauge how deep a loudspeaker goes in the bass, how extended it is at the top end and how easy it is to drive. Well, for starters, any quoted bandwidth that doesn’t state the measurement parameters (and it’s surprising how many don’t) is next to useless: ±3db has become something of a de facto standard, but in itself isn’t that useful. Any variation in those limits makes comparison meaningless, while what you really need to know is not just how deep or high the bandwidth goes within the measured limits, but how quickly it rolls off thereafter. Are those measurements anechoic or in-room? If they’re in-room, what are the boundary conditions? Likewise, a figure for ‘nominal’ impedance is almost as useless. When it comes to driving loudspeakers, it’s the impedance characteristic that’s critical: how much does the impedance vary and how sharply do those variations occur? And that’s before we get to the human tendency to polish results, portray them ‘”in the best possible light” or simply exaggerate. How often do you see bandwidth limits stated for efficiency? That speaker that claims a sensitivity of 93dB might only hit that figure at a couple of resonant peaks throughout its range. Its actual efficiency might well be a lot closer to 91db – which still keeps the frequency response within those ±3dB limits…