Ola, Iām grateful for that detailed explanation!
Also in my experience are the details of risks, severity and hazard, way to seldom discussed. (Iā more used to probability and impact).
I had with āambiguousā more the concept of āmultiple concepts under the same phrase or different phrases for the same conceptā in mind. Therefore at first I did not got your direction.
Iām sadly not surprised to see that important topics are not sufficiently discussed in our industry. 
Bug.
We get a bug raised via helpdesk. Someone looks at the steps to reproduce and says the customer was using the software wrong, āuser errorā not bug. Someone else says the steps to reproduce are possible to carry out, so it is a bug. Someone else says it is an issue, but the way of resolving it involves new functionality so itās an improvement. Or that the behaviour only goes against an implicit requirement, so it needs an improvement making that requirement explicit, but someone else says that it should be obvious so itās a bug still. And then the developers get involved and start debating whether the software was ever developed to specifically address the behaviour or just happened to cover it as a consequence of other workā¦
Thereās a lot going on behind the scenes when people talk about bugs, and I think there would be benefit in classifying different types, eg regressions, incorrectly carried out requirements, behaviour that is not what a reasonable user would want/expect, necessary improvements to current features to avoid problemsā¦ I need to think about it more, but in the meantime the different ideas people have when they talk about bugs is wasting a lot of time and causing a lot of confusion, at my company at least.
I often play devils advocate when it comes to the word bug. For me it can go by so many different names, and thatās why as testers, language is a super important tool that we need to learn how to wield like a gigantic orc smashing battle axe, and at other times like a dagger to the heart of a specific communication error. My favorite is to use āundocumented featureā when covering any non-regression bug during triages.