Using a suicide junction can be an alarming subjective experience. However, given the huge financial investment made as the result of road planners' decisions and the possibly horrendously expensive liability suits that would result if such crossings were shown to be dangerous, there must be objective evidence to prove that they're as safe as the more usual cloverleaf.
A similar disparity has long existed in audio, where some components with superb objectively assessed performance produce subjectively disturbing experiences. In the THD Wars of the 1970s, amplifiers offered successively lower levels of harmonic distortion without their designers once trying to confirm whether the "improved" performance correlated with improved sound. In fact, the reverse appeared to be true: by reducing steady-state distortion to vanishingly low values with extremely high levels of negative feedback, further problems were introduced that led to bad sound.
When I decided, back in 1989, to start a program of measurements to accompany our reviews, I believed that if someone could hear something, it would be possible to find a measurement—or, more likely, a collection of measurements—that would begin to explain what was heard. However, my experience in the years since then has convinced me that making the essential causal connection between measurement and observation is far from simple.
The problem is that, as Audio Research's William Z. Johnson said in a Stereophile interview back in August 1994, "hi-fi is one of the few industries where products are measured one way and used in another way....Every measurement that we're aware of falls into the realm of what we call repetitive, or static, measurement. In the real world, the simplest musical signal has component signals one ten-thousandth the size of some of the other signals present, and at many, many frequencies at once. It simply defies the abilities of static circuitry measurement."
Yet static measurements are still all we have available. I remember hearing the late Richard Heyser in London in 1986 present the idea of the hi-fi experience as being multidimensional. The reproduction of music involves five or more separate parameters—as many as you care to identify, in fact—changing instantaneously at any one time (time, of course, also being a parameter). But if you want to measure that performance, you can effectively plot only one parameter against another—or, if you have 3-D graphics available, one parameter against two other parameters. Any single measurement can therefore give you only a tiny part of the whole picture.
The result is questions: Should we publish a particular graph? Does it throw light on the component's sound? Does it suggest the opposite? In which case is something overlooked to blame?
Sometimes the answers offered by designers suggest that they've been asking the right questions. With the big Mission speaker in this issue, or our 2002 Product of the Year, Halcro's dm58 monoblock amplifier (p.61), it can be assumed that at least some of the reason they sound so good is because of the way they measure. Yet the Sutherland DAC and the PS Audio amplifier—also featured in this issue and both designed by talented electronics engineers—have measured aberrations that psychoacoustic theory would imply are audible at least some of the time. Yet both were well-liked by the reviewers. Are their designers asking the right questions but my measurements providing the wrong answers?
More work on the suicide junction is called for.