My thought wasn't that he wouldn't have anything true to say. It was that if he's still defending good and evil as obviously existing, in that context, he's far enough behind me on the issue that I can safely assume that he doesn't have anything major to teach me, and that what he says is untrustworthy enough (because there's an obvious flaw in his thought process) that I'd have to spend an inordinate amount of time checking his logic before using even the parts that appear good - time that would be better spent elsewhere.
That's not a good heuristic. There are a lot of people - Eliezer would name Robert Aumann, I think - who are incredibly bright, highly knowledgeable, and capable of conveying that knowledge who are wrong about the answers to what some of us would consider easy questions.
Now, I know Berserk Buttons (warning: TV Tropes) as well as anyone, and I've dismissed some works of fiction which others have considered quite good (e.g. Alfred Bester's The Demolished Man, TV sitcom The Modern Family) because they pushed those buttons, but when it comes to factual information, even stupid people can teach you.
(Granted, you may be right about the worthlessness of this particular speech to you - I haven't watched it. But the heuristic is poor.)
The heuristic isn't widely applicable, but I disagree about it being poor altogether. As I pointed out above, it's not just that he defended good vs. evil. It's that he did it in the context of a presentation on a subtopic of how we conceptualize the world. He may have things to teach me in other areas, obviously.
That's why I compared it to someone bringing God into a discussion on ethics specifically. (Or, say, evolution.) That person may be brilliant at physics, but on the topic at hand, not so much.
It also occurs to me that this heuristic may be unusual...
Tyler Cowen argues in a TED talk (~15 min) that stories pervade our mental lives. He thinks they are a major source of cognitive biases and, on the margin, we should be more suspicious of them - especially simple stories. Here's an interesting quote about the meta-level: