The Great Cable Debate

“Cables are like air — they’re only important when you don’t have any.”

I hem and haw every time I have to write something cable-related. I feel like I need to gird my loins (so to speak) every time I sally forth, lest I have my head handed back to me on a bloody platter.

The subject of audio cables, you see, appears to inspire only the very best behavior out of the otherwise cutthroat and murderous audio forums on the old Internettywebitude. I mean, can you imagine what might happen if I actually wrote about something contentious? I shudder to think.

I’ve thought about simply being a cable-Switzerland, quietly amassing my collection without offering any official public stance. It might be easier this way, even if it’s a little unsatisfying, but since I’ve already written about cables anyway, I suppose that’d be somewhat futile. Once more, into the breach! Or something. [sigh]

So, before I start plowing in, I thought that perhaps this time I’d reset the level. Again. Just so we know what’s going on here and why my discussion of these products tends to read a bit different than, say, something from Positive Feedback.

First, let me say this.

1. Audio cables matter.
2. Yes, they can make your system sound better.

Let the games begin.

Philosophy and the Zen of the Audio Cable

But even saying that begs the question: do cables matter? Which begs another question: can cables matter?

Of the two, the latter one is actually (fairly) interesting — at least in a philosophical sense — but it’s worth putting them both out there, because the forums happily conflate and confuse the two. And that’s annoying. So, just for clarity’s sake, let’s tackle the “can” question first.

The argument goes like this: wire is wire, and as long as you have some, it doesn’t matter what it is. The common variant is the zip-cord appeal — chop the cable off of a table lamp, and you can successfully use that for any of your audio system’s electrical needs. In fact, that cord will be indistinguishable from any and all other cables.

Whoops. That was another step, and yes, it’s actually different. The “can” question is, perhaps, properly qualified as a “metaphysical” kind of question. It might help to think of metaphysical questions as external questions — we’re asking about the world, or more specifically, some small piece of it, and perhaps it’s characteristics.  The other question, about “indistinguishability” isn’t metaphysical, it’s epistemological. That is, it’s about us, that is, about knowledge — what and how we know of the world, or the specifically small piece of it we’re curious about.

With me so far?

One of my grad profs, Michael Devitt, made a lot of hay about how you really need to answer the metaphysical questions before everything else — especially epistemic ones. So, say for the sake of the argument that I was paying attention that semester, and we go with that. It’s been about 20 years, so forgive the broad strokes, but in the case of cables, it’s pretty straightforward.

Yes, cables do make a difference. How do I know? Simple. Try running your system without them. Har har. No, okay, try this instead. If cables don’t matter to sound quality at all, then it shouldn’t matter what I use, whether it’s lamp-cord, magnet wire, or Nordost Odin. A connection is a connection. Right? Okay. So, take 500 feet of magnet wire, lamp-cord and Nordost Odin (remember to send that to me when you’re done). They’re going to sound different. Why? Their electrical properties. Steve Nugent over at Empirical Audio has some nice graphs that show the math behind what something as simple as cable length can do to a signal in the audio band. Remember, the question of “could we hear that” is an epistemic one, which we’re tabling for the moment.

Worse yet, it’s also not unreasonable to assume that it’s possible to design (or fuck up) a cable in such a way that it actually and actively introduces errors. In addition to attenuation, there’s also a potential for interference, too. These are some pretty obvious, first level, kind of issues — all related to the cable itself, all of which it isn’t unreasonable to assume could contribute to a negative impact on signal quality. Note that I didn’t have to have even a rudimentary discussion of reflection, propagation velocity, skin or eddy effects, or any other weird electric, signal or RF issues that may also impact signal quality. That really isn’t the point. Remember, this is a metaphysical question and the question is rather precise — “do cables matter”. The answer is “yes”.

Now that we have that out of the way, we can move to the thornier problem: how much do they matter? This is a straight-up epistemic problem, and one complicated by the ABX question. For those of you that have been sleeping, ABX is way to differentiate stimuli and uses some very clever (and arguably necessary) controls, such as double-blind testing. The Cable Doubters hold that, with proper methodological controls, humans — on average — cannot tell the difference between any two pieces of wire. So, why bother to try them out?

In general practice, I suppose a theoretical objection can prevent someone from going through the trouble of creating a methodological regime, mounting it, running subjects, analyzing data, and publishing. I mean, all that is a PITA.

But when the bar to entry is swapping an audio cable, this sort of intransigence borders on the absurd. Seriously? You can’t try out a new cable because your nifty little theory says it won’t work? This isn’t good philosophy anymore; wake up and smell the dogma.

Look, even with proper methodological controls, there are good reasons to doubt the efficacy and/or universal applicability of an ABX/double-blind approach. As I mentioned, Devitt argued (persuasively) that questions about knowing are separate from questions about existence, and we’ve already addressed the existence question — there are differences. The question of “do they matter” really ought to depend on the experiential data. If that data are inconclusive, that’s interesting. But if the math (see the Empirical reference, above) shows that at least some differences ought to be audible — and the testing design can’t reliably pick these out — then there is something wrong. Either with the math or with the design.

The clue, to me, is that there are reports that “some people” do report audible differences. It’s just a clue, but that’s not nothing. So, before we go stomping around on about detection and detectability, or what impact bias may have, just note that this is beside the point — for now. Again, we’ve already established that something is going on. The math indicates that it ought to be audible, and math is … well, lets just say that math is math, regardless of my personal feelings or beliefs on the matter. Kinda makes it a little less suspect, as a domain, at least to me anyway. So assuming that the math is correct, that leaves the methodology of the ABX/blind testing.

Generally speaking, if your model of a given phenomena doesn’t reveal an outcome that is predicted, one that is being reported anecdotally, this isn’t necessarily an indication of some design flaw or control errata or analytical outliers. I mean, it may be, but it isn’t necessarily so. It may simply be that your entire model is FUBAR — or perhaps just incomplete. Not exactly wrong. Incomplete. And that’s enough.

Here’s an analogy (I like analogies): wanna predict the flight path of a bullet? Well, Newtonian Mechanics will chart that fucker for you to a T. But Newtonian Mechanics, accurate as it is, is also … well, okay, “incomplete” might be a bit generous so let’s just call a spade a spade — it’s flat-out wrong. But being wrong doesn’t entail an inability to produce wickedly accurate predictions. As long as the domain isn’t really big or really tiny, Newtonian Mechanics is da bomb. So … perhaps that is what has happened here, in the Land of Audio. It may be that we’ve just found a corner case for ABX/blind testing. It’s a thought.

So, now that we have all that out of the way, let’s talk bias.

Bias, or the endless ability that we human creatures have to fool ourselves into doing, perceiving and/or believing absolutely bizarre shit, has been extensively studied and yes, I’m a believer in bias.

But this is not an indication of, nor does it have bearing on, metaphysical questions of existence. Bias is a purely epistemic variable. Since we’ve already done the metaphysics, let’s just leave that aside. Philosophically, what bias tells us is that humans are unreliable detectors and that we are prone to jump to conclusions that aren’t necessarily warranted. In the case of audio reviewership, bias casts doubt on conclusions of scale. Gradations and nuance (or entire events) that may not be perceivable, under bias, are believed to have been experienced. Subtle becomes obvious. Incremental becomes dramatic. Evolutionary becomes revolutionary. And vice versa.

No one can do anything about bias without some kind of help. For whatever it’s worth, this is precisely where a judicious application of ABX/blind testing can deflate some of the rhetoric.

Which brings up a related point ….

If the topic of audio cables has you, now, suddenly, worrying about subjectivity and bias of an audio writer … WTF? Were you not paying attention till now? Look, I hate to break it to you, but when someone describes the sound of a loudspeaker and uses terms other than “dispersion”, “frequency”, and “distortion”, chances are, you’re getting a subjective take. If you’ve ever read anything that has the phrases “night and day”, “not subtle”, “best I’ve heard” — you’re wandered quite a bit from the Path of Objectivity and are, actively, stumbling around that corner of the Nevernever commonly called “hyperbole”. Yeah … audio reviewing isn’t scientific in that sense — it bears far more resemblance to literary criticism. If you think of an amp as a novel, a review of that amp then would be a book review. Objecting to those claims found therein on the grounds of “improper methodology” is a bit like saying “please skip all the adverbs” or “I only like deconstructionist literature from oppressed minorities”. It’s just weird.

I think that ABX/blind testing as a method to take the hyperbole out of audio writing is a bit like using shotguns to rid your house of mosquitoes. It might work, but the result would be a mess. It also isn’t conclusive. Here’s the problem. If your random sampling of would-be detectors doesn’t reliably indicate a non-random result, the conclusion on offer isn’t that there is nothing to detect. Again, that’s already been worked out. The conclusion, more properly, is either with the logic that caused you to choose a particular methodological approach, or with the design or implementation of the method itself. I submit that it’s the latter.

Here’s my suspicion — that most of the hay being made about ABX testing has to do with an improper constraint on the sample pool in question. Mainly, my suspicion is that sensory discrimination (e.g., an audio test) is rather hard to do in a reliable way in a constrained time frame. Not impossible. Just hard — for the untrained. That is, I suspect that without some significant practice, there’s no reason to suspect that most folks would be helpfully reliable at this sort of thing. But suppose we could make people better at it?

Then the question isn’t one about whether differences can be heard by the average audiophile, but rather, whether the average audiophile can be trained to reliably detect some predicted effect. Note that such training isn’t bias, per se. Using so-called primed experts still gives you useful information — you can just build this into the methodology and correct accordingly. And note — this eliminates the confusion around the irrelevant metaphysical variable — do such such differences exist — and lets you focus on the correct epistemic one — can I expect to hear those differences?

So, who are these experts? I have no idea. But there is always this body of outliers that surround such ABX tests — people that are typically able to “beat the odds”. Assuming that they really are able to beat the odds, then they’re able do that either because they’re built for it or they’re skilled at it. If the latter, that can be transferred or learned. If it can be learned, you can create an appropriately constrained sampling pool. This might not be possible, but assuming that it is, your methodology can now successfully move past the metaphysical quagmire and actually do it’s job — to give you information about what those predicted deviations are, or more interestingly, what they might actually mean and when the conclusions would apply to me.

Another philosophical aside

Look, if all this seems to “miss the point” of ABX testing and you feel the urge to cry foul because, after all, measurements show that there is no difference and that you’ve never heard a difference between cables … then, you’re right. It does! Why? Because you’re confused.

This isn’t your fault. This isn’t even something you can necessarily help. Which is why separating out the logic of the argument is crucial. And in this particular case, you know, Science, it turns out that you have to be careful lest you take an inappropriate step at an inopportune time. Said another way, if you put the cart before the horse, you’re not taking a damn thing to market. Sorted properly, the system works. Which is why metaphysical questions need to be resolved first, before issues of methodology — at least, according to Prof. Devitt. Whether or not you agree is another matter. There are very good arguments you can make tackling Devitt’s arguments — I know because I used to beat him up with them — but his general point is sound. Worse yet, it works. So, taking metaphysics out of the question of what the specific results of a particular methodological test actually mean, really helps clarify things. Turns out, the whole problem is a bit of a red herring and relies upon a (common) misconception of how Science works.

Okay, so — just to be clear — we are here. Cables are different. Those differences may have impact in the frequency response in the band audible by the average human being. This is math, proven, and not subject to debate. This does not answer a couple of valid follow on questions, however, specifically, “can I hear the difference?” That’s where ABX can come in. If you’re a masochist. Or just bored.

A good plan

Or you could just stop reading and start talking — to an expert.

This is actually a bit harder now than it used to be, but your first stop in just about any question of relevance, impact or taste in the audio world really ought to be your neighborhood audio dealer. These guys, by and large, have been “just doing it” for years, and if anyone is familiar with the impact of any component on a system, it’s the guys that have been building, tuning, tweaking and selling them. Talk to an expert at Nordost and they’ll give you the straight scoop about Nordost products. If you’re lucky, maybe they’ll talk a little smack about their competitors (but probably not). Want to know if or when a Nordost product (or a particular Nordost product) would work best? Ask the guy who sells Nordost — and other, competing, products. Yes, you must remember that they’re sales people and their job is (and their livelihood depends on their ability) to move product. So, ask several of them. Take nothing on faith! Do your own testing.

Of course, you can just do yourself a favor, one a lot less painful than constructing and executing your own flawed ABX test, and just find a reviewer you like. Sit back, read what they write, and let them do the thinking for you. This obviously isn’t as reliable as doing it yourself, but it certainly is easy. But I have to say that I’ve read a lot of cable reviews, myself, and I have to kinda shake myself afterwards to remind myself that most of what I just read was crap. Most of the reviews you’ll read in the Mainstream Audio Media are paid for, directly, from a publisher to a contract reviewer. Yes, advertising dollars pay the publisher, but that’s irrelevant for this point — reviewers are paid to create content. A wishy-washy, maybe-I-did/maybe-I-didn’t, review is pretty bland copy. A good copy writer knows that, so the reviewer that wants to get published (and paid) for the next article makes sure that every article is interesting … even if it isn’t strictly accurate. No, reviewers don’t lie, but you have to remember this bit about getting paid to understand why the heaping piles of hyperbole are there, why the article reads as clear-cut, and why the conclusions aren’t couched, caveated, or even slightly tentative. Nature of the beast, and all that. So, remember to take what you read with a big dose of “I really need to check this out myself”.

Then do your own testing! I don’t know how many times I can repeat this before it sinks in, but neither your dealer nor your chosen reviewer is going to be stuck with your system, so why let them do your listening to your system for you?

Still, it’s a given that you can’t always do your own listening, and there are fewer and fewer “real deal” dealers these days, so … perhaps a “trusted reviewer” is helpful. Just remember that I said that no reviewer is a replacement for your own ears, okay?

Observations about cables

Okay. So, not that I should be your “trusted reviewer”, but just for the sake of argument, I thought it’d be useful to talk for a moment about my experiences with cables — in the context of preceding screed. Should give you an idea of what I look for, what I usually find, and what that might mean — for you.

First up, I find that when cables do impact the sound of the system, that impact is relatively minor. This might come as a surprise, but there it is. Yes, cables make a difference, but … by and large, they don’t make much difference.

Let me fill that in a bit.

Assuming a given cable has any contribution that’s audible, the most obvious place that this impact is audible is in frequency extension. And yes, some cables do attenuate bass — this might be something a reviewer would call muddy, slow, or soft. Or something. Same thing about the treble, though this might be called sweet, thin, or dull. Or something. Whatever — that’s the obvious stuff, and the stuff folks usually have little trouble keying on in quick A vs B testing. The “problem” is that … well, that isn’t all there is to it. And, perhaps unfortunately, it’s that extra bit that isn’t all that obvious or fails to resolve itself during a fast-switch.

So, here’s some of the less-obvious things I’ve heard cables “do” to a system. The first is change tone or timbre. This is almost always really subtle. At a gross level, the system may sound “richer”. Or “fuller”. Of course, I might have been imagining all that, but if so, it was a pleasant hallucination I was unable to replicate with some other cables, much less some randomly placed rocks, prisms, or weights, pretty though they may be. When changes are subtle, though, pulling the granularity out in a way that is meaningful is rather hard to capture — in words. So, pardon the synethesia, but I see an analogy in these Samuel Silva pics:

Which is better?

Just so we’re clear, the pic on the left is a line drawing. The pic on the right is a photograph. Many of us, based on that knowledge, would say that the pic on the right is better — “because it’s real”. But that said, just about everyone would then marvel at the reproduction on the left. It’s breathtaking, really.

Interestingly, the pic on the left is the one that has more detail. The pic on the right has more depth and better tonal gradation. Both are awesome. But — and this is important — neither are real. Both images are reproductions, and imperfect ones at that — neither are the actual, real, live, girl.

This is the kind of thing that audio reviewers are trying to capture in their reviews.

Back to cables. In this context, the impact that “good” cables can have is to help move the dial to the right (as in, to the right-side pic, above). I’m not saying that something as simple as a cable change can or does have this much impact, but this is the kind of change I’m talking about: sometimes, in my system, in my room, with my gear and using my ears, I hear a change in the overall sound quality of the system that is akin to the shift from the left picture to the right, above. A slight deepening of the soundstage. A slightly more resonant tone. A small improvement in timbre. A brassy cymbal crash becomes “brassier”, for example. And so on. Tiny things individually (perhaps), but when added together, they render the constructed audio illusion ever so slightly more convincing. To me. YMMV.

Heading out

This started being a cable review and ended up as yet another pseudo-philosophical rant. Sorry about that. It’s just that I get rather tired of having to cover and re-cover the same old ground every time I talk about cables. We audiophiles are nothing if not an overly opinionated bunch. And for whatever it’s worth, whatever applies to audiophiles generally gets amplified when applied to audio reviewers.

I have some truly wonderful cables here from Triode Wire Labs, High Fidelity Cables, Mojo Audio, Nordost, WyWires, Black Cat Cables, and Blue Jeans Cable. In some upcoming posts, I’m going to try to sort out some of what these products have to offer, but I wanted to first put forward a general discussion of the debates so we put them to bed and spend that time talking about something other than philosophy.