A lot of these sorts of inconsistencies in cable performance crop up--we hear about them all the time and it can be hard to figure out just what's going on. We do have some ideas, though, which probably explain a large part of it all.
First, there is the mysterious and weird realm of the DDC (Digital Display Channel). Excessive capacitance can be a real problem here because this is not an impedance-matched circuit and the impedance is typically run "low into high." Not a problem at short lengths, because this is low-speed data, but it can become a problem at longer lengths. It is very hard, unfortunately, to limit DDC capacitance very effectively--and, at some length, the impedance mismatch itself will start to make a mess of it.
Second, intermediate devices (switches, boosters, splitters, etc.) sometimes are active when they appear to be passive, and sometimes are linked to the 5v DC channel. In some cases even when there's an external power supply hooked up, the continuity from the 5v pin to the power supply for the device may be connected, and this is potentially problematic because the 5v is intended as an "on" indicator, not as a source of significant current. When the cable length is long, and when something is loaded on this line, there can be a voltage drop at the display end, with somewhat unpredictable consequences.
Third, it has always been true that devices vary in their ability to generate strong and clean signal or to properly interpret weak and dirty signal. The differences here can be profound; we have found cases where a given source to one display through a given cable will go one distance, and changing out the display cuts the working distance to half.
Fourth, the jack/plug interface is problematic. Some jack/plug combinations are fairly tolerant of a bit of side stress or up/down stress on the interface, while others are not. Users often interpret the funkiness of the jack/plug interface as meaning that the cable is intermittent, because when they wiggle the end they can get it to work--but we find that, while cable intermittency at the connector is always a possibility, we are usually unable to reproduce reported intermittency when the cable is tested.
Fifth, it is certainly clear that cables can introduce problems which can screw the signal up. Not only is impedance control difficult, but crosstalk or skew can mess things up very badly. There is a tremendous amount of pressure on the market to keep costs down and we have seen some nasty cable assemblies.
Sixth, it is likely that sometimes a cable swap fixes an issue simply because it's a swap. Unplugging and replugging reinitializes (or should) the HDMI interface, and so if what is really needed is a sort of HDMI "reboot" a cable swap will do that, leading to the perception that the swapped-out cable was broken and the new one is good, when both may be good.
This list is not exhaustive. There are all manner of potential issues with device interaction, especially when there is switching, boosting, EQing, et cetera in the mix; and whenever you think you've seen or thought of everything, something else happens.
Kurt
Blue Jeans Cable