Why do digital cables sound different?

It seems to me that this is more of an issue with data loss than it is with timing errors, although I have read in other threads RE transport quality that re-clocking can only go so far, and that the more conservative the initial translation and transport of the digital signal, the better the outcome, even with the most successful implementation of re-clocking on the DAC end of things.

Knownothing, Imagine that you play clean 1kHz sinewave but digital stream jiters in time back and forth at 60Hz. It will result on analog side in expected 1kHz signal and two other signals (sidebands) at 940Hz and 1060Hz. Their amplitude will depend on the amplitude of time jitter. These sidebands will be at very low level, in order od -50dB but still audible since they have no harmonic relation (like overtones do) with original 1kHz tone. It is form of modulation (similar to FSK). While music has many frequencies time vibration (jitter) of the signal appears at many frequencies, together causing whole bunch of new frequencies - a hash. This has has amplitude proportional to sound level and is inaudible without signal. It will make music less clean, imaging less precise etc.

Mapman, very short coax should be free from reflections. Rule of thumb says that to avoid transmission line effectes (reflections) use cable so short that propagation one way thru it is less than 1/8 of transition time. Assuming average transition at 25ns it will be <3.125ns equivalent to about 2ft of the cable (propagation = 5ns/m). Since it should include internal connections in transport and the DAC I would not risk putting more than 1ft. Longer cable might cause reflections and to avoid first reflection affecting first transition cable has to be longer than about half (threshold point) of this transition 25ns/2=12.5ns. At 5ns/m it will be 2.5m and since it is both ways 2.5m/2=1.25
It shows that 1.5m or even 2m cable might be better than 1m.
I said might because nobody knows exactly what transition time is so it is pretty much trial and error. If your fancy transport outputs 5ns transition it will give you great immunity from electrical noise affecting transition time, but will require well matched characteristic impedance to avoid reflections. Toslink is immune to ambient electrical noise and does not produce transmission line effects but driver transitions are slow making it more sensitive to system noise (on both sides).

My system images and does detail very well, with detail as good as most any of the best headphone setups I hear or my phono source, essentially as best I can tell. IE no detectable issues ommissions, flaws, etc. So I am sure jitter exists and could be made better (nothing is perfect), but impact in my case to-date is just not an issue.

Of course, although I am always listening to various things to know better, you never know for sure what you may be missing until you hear it, but I do listen to a lot of music in various ways enough to be able to make a good determination whether things are going well or not.

Note that the architecture used for digital streaming can be a variable in determining end results, but I find if done well a certain way, then the wire between the source and DAC seems to make little difference, as long as of good quality, and perhaps not too long as Kijanki suggests.

The architecture I like is wireless network connection from server to streaming device (for physical and electrical noise isolation from potential noisy commercial computer gear), and SPDIF/TOSLINK from streamer to DAC. Most network streaming devices buffer data which help to avoid timing issues at the source.

Then choose your DAC of preference for the sound you like. If it does reclocking very well, then that is an extra insurance policy in case of any doubt. The trend seems to be for more and more DACs to do this these days.

THen use a good quality digital wire that is not too long and you should have all teh bases covered.

I second Mapman's recommendation of network streaming. Computer speed, amount of RAM, playback program, file format etc. make no difference since timing is recreated on the other side of the bridge (saves a lot of money). Sound still can be affected by ambient electrical noise but in this respect slow computer might be better than the fast one.

Asynchronous USB should be good as well since, as name suggests, it does not use noisy/jittery timing from computer by buffering the data and requesting different amount of samples each frame to keep buffer between underflow and overflow.