There's many logical issues with the argument. As should be expected. It is very rare that logic actually wins an argument, and it's usually because there's a referee ensuring the argument stays logical.
The particular issue you will likely want to have attention drawn to is the difference between "a specific word is offensive" and Alice's statements such as "A large portion of people find it offensive" and "Because this conversation has been had many times, many people find it offensive". These statements are being made in English, rather than a language with explicit logical constructs, so its expected that people will take leaps.
The first leap is from "a word is offensive" to "a large portion of people find it offensive." This statement is not justified by any further logic. If anything it's being presented as a definition for a word having the property of being offensive. In a truly logical argument, both Alice and Bob would have agreed upon a list of these definitions and axioms ahead of time. In a real argument, like this, Alice may introduce a new axiom. For this to be "logical," Bob would have to accept the statement as true, or Alice would need to justify the statement using logic and relying on statements that Bob has already accepted as true.
The second piece is "Because this conversation has been had many times, many people find it offensive." This is another implication that doesn't have a logical justification included. Bob could easily argue that 2 devil's advocates could have this conversation without finding it offensive. There's also a trend to saying less and less. We started with "the word is offensive" then changed to "a large portion of people find it offensive," and now we're down to "many people find it offensive." Define "many." 100 people may be "many people," but it's a small portion of a city of 1,000,000. Alice is assuming that these statements are justifying her more sweeping claims, but in fact they are not logically justifying them. She is relying on the human side of Bob to look at a not-quite-logical-argument and accept it as justification in the larger argument.
Bob, on the whole, is much more logical. His argument is basically taking the approach of proving Alice's argument to be non-logical at every step without adding any non-logical statements of his own. However, he did make one similar mistake. When he talked about the flat earth arguments, he made the assumption that there was a connection between flat earth arguments and offensive word arguments. While it's natural for one to argue "Alice, if you invalidate a similar argument, you invalidate your own," that statement is not logically justified unless the agreed axioms for the debate all treated offensive word arguments and flat earth arguments equally.
Because we don't have a preamble where Alice and Bob lay down the axioms and define the domain of discourse, we can only guess as to whether Bob's argument is logical or not. If I were to pick up the argument from here, I'd invoke Godwin's Law as paradoxically as I can, but pointing out that in many debates where the logical axioms are not stated, arguments about Hitler are often placed in a separate category such that one cannot logically infer anything about a Hitler argument from a non-Hitler argument and vice versa. I'd say this is typically justified because it's hard to specify logical axioms which work well in corner cases with individuals who are considered to be that evil.
The real moral of the story is "Debate using logic, not English." Just kidding! The real moral is that real debates are never 100% logical, because the devil is always in the details. Even if the debate starts as 100% logical, someone will eventually attack one of the axioms, and demand it be justified instead of simply assuming it was true. This process will continue until the debate ceases to be 100% logical, or until Baron Münchhausen pulls himself up out of the mire by his own hair.