An isomorphism is a way to relate two different sets such that each element is paired with exactly one in the other set such that it does not matter if you do operations before or after.
In my calculator the set of reals is isomorph to the set of complex numbers with imaginary part zero:
ℝ ⬄ { c | c ∈ ℂ and im(c) = 0 }
So I can coerce to complex numbers without impunity because of that isomporphism! And I know it is an isomorphism because if adding two reals then coercing to complex is the same as coercing first then adding.TIL: In a way mathematics has types like programming languages.
I hope I am not too far off here. I didn't look up anything, this all went into my head this morning.
The closest thing I'm aware of to "pretty much equal" is "unique up to unique isomorphism", so they may not be equal, but they're isomorphic, and there's no flexibility to make any choices of which isomorphism to use. But "isomorphism" also implies you have a particular context/structure in mind that you want to preserve, e.g. a set isomorphism (a bijection) may not be a linear isomorphism (~an invertible matrix). In practice, you may be working in multiple contexts at once, so you invent Functors which map one type of morphism to another, and now you care about Categories.
In the pre-rigorous stage, you don't know it's an issue. In the rigorous stage, you know it's an issue. In the post-rigorous stage, you know it's not an issue (i.e. you know you know how to write down all the details if you needed to, and you know it will all work how you might hope it would, and so you don't need to).