How common is it for someone to buy one of the retail-grade cables they're discussing in this article, and find that it doesn't work reliably?
I recall making a very crude crossover cable by cutting and soldering wires together and still got 100mbit (as high as I could test, this was early 2000s I think).
I want to say this was Bob Metcalfe, but not 100% sure he was involved.
Though, mind you, right tool for the job—I'm surprised that everybody is talking about "getting perfect cables" for maximum interference-tolerance in business-critical applications, when what I'd expect the professionals here to be thinking in terms of is taking measurements of the local RF environment they're going to do cable runs in, and then using the cheapest thing that's within tolerance.
Most people buying retail cables aren't going to be using that for that distance, and are more likely to run at 1Gbps than 10Gbps; where cat5 (not e) meets the spec at full distance. I'm a bit surprised that the cables didn't all meet cat5 (e) spec, but even that is not really required to have a solid ethernet connection at 1Gbps if you're not in a tricky environment -- I ran a 1Gbps drop over mostly cat 5e with a ~ 20 foot section of cat 3 for many years with no apparent issues (repurposed a buried 4-pair pots drop to a separate structure for ethernet).
Most likely any reliability issue with this cables is going to be if the connectors were poorly crimped and the connections loosen over time. That said, ethernet negotiation is a slower protocol, that's easy to decode, so it's very possible for both link partners to negotiate to something the wiring can't sustain. I've seen some drivers end up negotiating down to a lower speed in this case, but it's not part of any formal ethernet spec AFAIK.
Oh man that takes me back. I did the same thing though it was for 100Mpbs at the time. I distinctly remember staring at the twisted pairs prior to starting and wondering, “Is the actually going to work?”
If manufacturers want to sell Cat 5E (or lower grade) cables they are free to do so, but the market won't accept the higher price they want to fetch for those cables.
It also means the manufacturers are either incompetent (doing no testing) or are knowingly testing the cables but not bothering to test them to the specification they are claiming on the package. Either way that's a crime.
That certainly requires a creative reading of what I wrote. No, I'm saying that the test specs may be rigged to favor high-end cable manufacturers, and we may all be paying too much for cables as a result.
That is to say they are testing the _very definition_ of Cat-5E, Cat-6, or Cat-6A. That's the thing about cabling standards: either you meet the requirements of the standard or you don't. I presume the committee that designed 100Base-T, 1000Base-T, and 10GBase-T didn't pick numbers out of nowhere for no reason.
Your comment seemed to be using circular reasoning... if the standard is too stringent than it's just a free-for-all, or the manufacturer's made-up non-standard, but there is only one way to be Cat-6A: follow the specs laid out in the standard. Otherwise why even have standards?
Change the label, double the price. The temptation is very strong.
Similar amusement from the Capacitor Plague: https://en.wikipedia.org/wiki/Capacitor_plague
Absolutely nothing in the real world needs twisted-pair cables to exhibit 54 dB of NEXT. If it does, that's an engineering failure in itself, IMO.