Richness and precision are not at odds with one another. Attempts at understanding language through precision is precisely why AI has failed miserably when it comes to actual understanding of language. I promise you that figuring out how to model all the various contexts and connotations that are associated with language is the big breakthrough that natural language processing has been waiting for.
That being said, one can have a precise discussion regarding these admittedly emotional topics if we clearly define all of the relevant contexts that are involved. But anyways, this has suddenly turned into a discussion about race unintentionally. So instead of going into great depth, I will cut to the point which will hopefully clarify my problem with your type of reasoning.
Words like racist and sexist have a certain power in our culture. If you call someone a racist, they immediately become defensive because it has become one of the worst things you can call someone. This is precisely because of the connotations involving hate, inferiority, violence, etc. Now when you attempt to prune all of that cultural context from that word (by defining it as generic discrimination based on race), and then label programs geared towards minorities "racist", you are intentionally wielding that cultural context to stigmatize something that would not otherwise receive that negative connotation. You are changing the definition without simultaneously changing the cultural connotations that give the word its power. This is what is so despicable about altering the meaning of these words: you are using unconscious associations between racism and evil things to stigmatize things you simply don't like by forcing a new association in the minds of your audience by way of the old label. The "evil" connotation is then unconsciously transferred to the thing you don't like.
A clear example of how the degenerate definition is useless, and frankly absurd, is the case of doctors that check black patients for high blood pressure, heart disease, hypertension, etc. This is your "technical" definition of racial discrimination, ie "racism". But it is absurd that this should carry a negative stigma by falling under such a label.
Without proper context, words lose all meaning.
If you want to talk about something more complicated than just "discrimination based on race", I think you need to describe that in terms of simpler things we agree on before we can decide whether it's happening and whether it's good or bad. Otherwise we don't know what we're talking about. This conversation is a perfect example. While I'm using the strict definition, you seem to think that "racism" means when the underdog is discriminated against but not the "dominant" culture, I still don't understand it. I think your definition is broken and deceptive (at least it's possible to prove something from a "degenerate" definition), and you think the same of mine. Would have been nice to know that at the start.
On the subject of pure logic, pointing out that someone's argument benefits them is strictly a logical fallacy. I'm tempted to whine that your arguments are self-serving, but tu quoque is a fallacy too. Both are usually counter-productive, except possibly as an explanation after you've explained logically why someone is wrong, which you have still not done.
This is how I try to approach any sort of philosophical or moral question, especially in writing. If you don't agree on this basic approach to finding truth, we really have nothing at all to say to each other.