It's the audible changes that always seem to somehow 'conveniently' get better after "break-in," in spite of the different directions/levels such an effect could go after the fact. The unit could have been at it's optimum on the initial power-up and went somewhat south from there for all we know.
Maybe, if the designer is an incompetent moron. Passive and active component manufacturers provide data sheets for a reason. If you come across such a product, run, don't walk, away from it and everything that company touches.
I will concede that we do sometimes come across such products that use digital logic, because for some reason engineers who plan to live in the world of software and logic don't only forget that they are dealing with analog electronics, they have little knowledge of analog circuitry. They picked a bad time to catch some ZZZ's in class. I have even run across some who don't know Ohm's Law. Tragic error.
That kind of person will design logic circuits based on the simulation, which is never correct; real parts don't behave that way. You can, with a lot of work or a lucky model, find simulation data that takes into account the typical behaviour, but it's not guaranteed; you may be working with a sim that assumes perfect parts; perfect parts don't actually exist in the real world.
(Data sheets generally spec parts at 25C [77F]. Strictly speaking they use degrees Kelvin, but it works out to the same thing. Capacitors are a little different; they are rated for 85 or 105C but the same thing applies, except operating temperature is expected to be lower, not higher, than the data sheet. If you just read the tables, versus studying the graphs which do represent temperature values, you will miss how the part will operate at some other temperature, temperatures typical inside an enclosure with operating circuits).
For example, the USB power is *supposed to be* 5V with logic states changing at 1V and 4V. If the power supply puts out 5.2V or 4.8V, your logic state voltages change. This *will* introduce errors. Somewhat on topic, HDMI is a one-way protocol; the receiving device has no way to tell the transmitting device it's having trouble deciphering the data stream (which is an analog stream, even though it represents digital data).
HDMI is considered in spec if there is one data error every 8 seconds (with a 1080p data stream; it goes up or down depending on the data rate). In practical terms this is inconsequential if in spec (the bad pixels won't be onscreen long enough to be noticed with human vision), but will be a problem if the data is somehow corrupted; if the characteristic impedance is badly mismatched, etc. because the error rate can increase rapidly. This is one reason why you want a decent HDMI cable (or USB cable, etc); no need for snake oil but a certain level of quality is a minimal requirement.
How well or how poorly these errors are tolerated requires a solid background in analog theory, something that seems rare amongst the "bits is bits" crowd.
Apologies for the rant, but it feels good to get it out of my system ;-)
