Others have already explained how 'random' numbers are produced by deterministic machines and why for all practical intents and purposes these can very well be called random.
Concerning the talk of 'true random': To this date*, there is no accepted definition of what that even means. Not surprisingly, philosophers continue to debate this. Mathematicians have tried hard to define the concept ca. 100 years ago and have simply given up. Instead they settled on a purely formal definition that does not even aim to cover the natural language use of the term.
Most of what you read about 'true' or 'pure' random is half informed pseudo knowledge consisting of titbits of quantum mechanics, philosophy and maths. Nobody who knows what they're talking about will just uncritically use the term and pretend that it is has precise meaning.
*more specifically: until 10 years ago, when I left uni and stopped following the debate.
This will generally hold true if you equip the items you find the moment you find them regardless of how they compare to your existing items. Lets suppose over a period of 6 hours, you found swords A.B,C,D,E & F. If you equip each sword then it’s likely you will see a 50% chance of usability when going from one to the next.
However, if you play like me and only equip a better items, then this changes everything. Lets suppose again that you found 6 swords A.B,C,D,E & F.
Sword B is an almost perfect ( let’s say 95% ) roll of a Thunderfury, the chances of swords C, D, E and F being an improvement on it are extremely low.
This is natural selection at play.
If Blizzard were to make sure “we had 50% chance to get a usable item”, they would have to make sure every upgrade is controlled ( say +1 int/dex point ) from one item to the next otherwise weapon stats would go through the roof.