Over the years, my audio system has transitioned from an analog network with a few digital peripherals, to a digital network with a few analog peripherals connected through analog-to-digital converters (turntable) or digital-to-analog converters (speakers). Not just “network” in the Ethernet sense, but a complex web of digital cables, connectors and protocols of different types: SPDIF, Toslink, I2S, USB, HDMI. And given what we have learned about the importance of quality cabling in this network, it occurred to me that it might be equally important in other digital networks.
So I did a test. I replaced all Ethernet, USB, HDMI and power cabling throughout the house with Audiophile quality cabling. (I won’t mention the brand, because this is not a a product test, it’s a concept test; suffice it to say that it was a respected quality brand, and all the cabling was from one brand since we have heard experts emphasize the importance of the synergy formed by a consistent mesh.)
For the test, I started with photography. I am a serious photographer, with quality gear (lenses, cameras, displays, software, printers, ink, paper) that would be up to the demands of showing any improvement.
I took new pictures (landscape, portraits, abstracts) and imported them fresh over the new cables. I processed them with my standard fine-art workflow, and printed them with my standard technique. I invited some friends to help me evaluate the prints.
Like with audio, photographic content has a given resolution (pixels vs. sampling rate, bit depth per sample) which does not change — “bits are bits”. But like in audio, it isn’t that simple. Looking at the prints, we found a subtle strengthening of “acutance”, which is the subjective impression of resolution, sharpness, precision, clarity. And we found improved dynamic range, subjectively beyond the level expected from the known equipment. Dynamic range is not just the level of blackness and whiteness (easily adjusted in Photoshop), but the amount of detail represented in shadow and highlight areas: does the shadow beneath a tree show the colors of moss, lichen, ferns, granite and gneiss, or is it just a dark mass? Does a summer sky show the structure and texture of cirrostratus, or is it just a white haze? And noise was reduced, especially important in dark areas taken in low light. The difference was subtle, but compared to the improvements effected by other equipment improvements and by disciplined workflow, definitely worthwhile for fine art prints.
Ok, it is perhaps not so surprising that photography shows sensitivities similar to audio, we have similar goals, even use similar terminology. But photographs are static, audio is dynamic. I should look at something dynamic, but I have no experience with quality video.
We looked at gaming. Again, I invited some friends to help with the evaluation (different friends). Subjectively, it felt like both imagery and motion were clearer, behavior more agile and responsive. One possible explanation: when rendering objects like cars, the rendering engine processes a lot of detail, not just the fine-grained geometry but also the material characteristics, to permit creation of not just shadows and highlight but specular reflection. And in a dynamic scene, the system does modeling of physics, such as the weight of the wheels, the suspension geometry and the spring rates, so a Porsche doesn’t just look different from a Ferrari but behaves differently. This is demanding of CPU and GPU power. To that end, the rendering engines do dynamic Level-of-Detail optimization: when a car is far away so it is rendered in only a few pixels, we don’t need to render so much fine detail. And we can LOD the physics as well, we don’t have to model the wheel behavior in fine detail. This is a balancing act: more detail looks better, but the system may not be able to sustain a high frame rate, making motion jerkier. And in a networked multi-player game, the rendering optimization depends on the behavior and positions of the other players’ cars. And timely rendering is important for quick-twitch player response. So given the information flowing back and forth over the network, and between the computer, the display and the control devices, it is certainly conceivable that cabling could influence both speed and accuracy, and in turn tweak the LOD optimization, driving both better rendering and more dynamic response.
But even more than in audio and photography, we have the danger of confirmation bias and the placebo effect. But here, we have an opportunity for objective testing: online games against anonymous adversaries. After trying first-person-shooters, racing and strategy games, the objective results were impressive: in every category, each of us advanced at least one level in the rankings.
Of course, there is the danger of the Hawthorne effect, where employees at a company improved productivity when lighting was raised, lowered, restored — anything that showed management attention. This effect is difficult to eliminate — did we play better because we were testing?
But I had a serendipitous opportunity for an unsolicited test, better than a double-blind test because it eliminates Hawthorne issues. My wife had not paid any attention to what we were doing, she just knew we had been futzing with my gadgets. She had been working on a research paper in cultural geopolitics, in her own office. But she came in and hesitantly asked, what did I do to the computers? She had noticed that her internet searches were going better. Not faster, bandwidth is bandwidth. But when doing the subtle, complex, shade-of-gray searches typical of social science, she felt the last two days had been more productive than usual, giving more perceptive answers. A subtle distinction, but unmistakable.
How could this be? Why would Google’s servers in a distant datacenter care what kind of cabling we use? I can’t answer that with certainty. But I think the effect may be similar to how experts answer the question, why would my final four-foot power cable matter after the utility’s miles of cabling? We shouldn’t think of this as the last piece of cabling, but as the first.
These tests illustrate how the kind of thinking common in audiophile circles can revolutionize other fields.
Claptrap, of course. But plausible-sounding, innit?