I suppose that the "Y-dependent" formula is " when solving for the dependent variable Y we set it equal to the intercept, b0, plus a slope coefficient, b1, times the independent variable, X, plus an error term.", yes?
So in this formula, we are suggesting a model of the data. For example, I am saying that Y = lifetime income and X = years of education and that Y = b0 + b1* X + error. At this point I know nothing about b0 or b1 or anything about that error term. I have merely suggested a model that might help me either understand the relationship or predict lifetime income when given years of education. Probably I believe that b1 > 0 so that for each additional year of education someone gets, their lifetime earnings go up by, say, $50000. If I found that, I could sell my analysis to public television who could then have Anna Nicole Smith say that "If you want to have money, stay in schooool" (forgetting for a moment that she is dead). My next step would be to gather data, estimate b0 and b1 from the data (and I would call my estimates b0-hat, b1-hat), and see if the data suggest that my model is right.
The "population values" for b0 and b1 are unknowable. Even if you could sample all trhe people in the world, your equation is broader than that. It asks about all the people who are or could have been or will be and what would happen to them if they had more or less education. God knows b0 and b1 and your mission is to estimate a truth known only to Him (every once in a while, God sees this as hubris and throws a statistician into the lake of fire. It usually doesn't happen to CFA candidates).
The cool thing is that even though I have this very vast unobservable universe of possibilities that I am trying to make inferences about, if I do my sampling right and I make some possibly defensible assumptions about that error term, I can make reasonable good inferences that depend only on my sample size.